1
00:00:00,090 --> 00:00:01,780
The following
content is provided

2
00:00:01,780 --> 00:00:04,019
under a Creative
Commons license.

3
00:00:04,019 --> 00:00:06,870
Your support will help MIT
OpenCourseWare continue

4
00:00:06,870 --> 00:00:10,730
to offer high quality
educational resources for free.

5
00:00:10,730 --> 00:00:13,330
To make a donation or
view additional materials

6
00:00:13,330 --> 00:00:17,215
from hundreds of MIT courses,
visit MIT OpenCourseWare

7
00:00:17,215 --> 00:00:17,840
at ocw.mit.edu.

8
00:00:26,225 --> 00:00:27,600
PROFESSOR: Welcome
to Foundations

9
00:00:27,600 --> 00:00:30,330
of Computational
and Systems Biology.

10
00:00:30,330 --> 00:00:32,030
This course has many numbers.

11
00:00:32,030 --> 00:00:35,410
We'll explain all the
differences and similarities.

12
00:00:35,410 --> 00:00:39,540
But briefly, there are three
undergrad course numbers,

13
00:00:39,540 --> 00:00:44,770
which are all similar in
content, 7.36, 20.390, 6.802.

14
00:00:44,770 --> 00:00:49,470
And then, there are
four graduate versions.

15
00:00:49,470 --> 00:00:55,410
The 7.91, 20.490, and HST
versions are all very similar,

16
00:00:55,410 --> 00:00:57,030
basically identical.

17
00:00:57,030 --> 00:01:01,245
But the 6.874 has some
additional AI content

18
00:01:01,245 --> 00:01:04,420
that we'll discuss in a moment.

19
00:01:04,420 --> 00:01:06,720
So make sure that
you are registered

20
00:01:06,720 --> 00:01:11,160
for the appropriate
version of this course.

21
00:01:11,160 --> 00:01:15,710
And please interrupt with
questions at any time.

22
00:01:15,710 --> 00:01:22,260
The main goal today is to give
an overview of the course, both

23
00:01:22,260 --> 00:01:24,420
the content as well as
the mechanics of how

24
00:01:24,420 --> 00:01:25,503
the course will be taught.

25
00:01:25,503 --> 00:01:28,310
And we want to make sure
that everything is clear.

26
00:01:28,310 --> 00:01:31,070
So this course is
taught by myself

27
00:01:31,070 --> 00:01:35,135
and Chris Burge from Biology,
Professor Fraenkel from BE,

28
00:01:35,135 --> 00:01:38,960
and Professor Gifford from EECS.

29
00:01:38,960 --> 00:01:45,450
We have three TAs, Peter
Freese and Collette Picard,

30
00:01:45,450 --> 00:01:51,570
from Computational and Systems
Biology, and Tahin, from EECS.

31
00:01:51,570 --> 00:01:54,620
All the TAs have expertise
in computational biology

32
00:01:54,620 --> 00:01:59,480
as well as other
quantitative areas

33
00:01:59,480 --> 00:02:02,429
like math, statistics,
computer science.

34
00:02:02,429 --> 00:02:04,845
So in addition to the lectures
by the regular instructors,

35
00:02:04,845 --> 00:02:09,320
we will also have guest
lectures by George Church,

36
00:02:09,320 --> 00:02:12,640
from Harvard, toward
the end of the semester.

37
00:02:12,640 --> 00:02:14,640
Doug Lauffenburger
will give a lecture

38
00:02:14,640 --> 00:02:18,530
in the regulatory network
section of the course.

39
00:02:18,530 --> 00:02:21,350
And we'll have a guest
lecture from Ron Weiss

40
00:02:21,350 --> 00:02:23,220
on synthetic biology.

41
00:02:23,220 --> 00:02:25,770
And just a note
that today's lecture

42
00:02:25,770 --> 00:02:27,580
and all the lectures
this semester

43
00:02:27,580 --> 00:02:32,950
are being recorded by AMPS,
by MIT's OpenCourseWare.

44
00:02:32,950 --> 00:02:35,850
So the videos, after a
little bit of editing,

45
00:02:35,850 --> 00:02:38,972
will eventually end
up on OpenCourseWare.

46
00:02:38,972 --> 00:02:39,930
What are these courses?

47
00:02:39,930 --> 00:02:45,860
So these course numbers are
the graduate level versions,

48
00:02:45,860 --> 00:02:49,170
which are survey courses
in computational biology.

49
00:02:49,170 --> 00:02:51,310
Our target audience
is graduate students

50
00:02:51,310 --> 00:02:54,935
who have a solid
background and comfort--

51
00:02:54,935 --> 00:02:56,740
or, a solid
background in biology,

52
00:02:56,740 --> 00:03:01,130
and also, a comfort with
quantitative approaches.

53
00:03:01,130 --> 00:03:03,230
We don't assume that
you've programmed before,

54
00:03:03,230 --> 00:03:05,200
but there will be
some programming

55
00:03:05,200 --> 00:03:06,970
content on the homeworks.

56
00:03:06,970 --> 00:03:12,430
And you will therefore need to
learn some Python programming.

57
00:03:12,430 --> 00:03:16,180
And the TAs will help with
that component of the course.

58
00:03:16,180 --> 00:03:20,240
We also have some online
tutorials on Python programming

59
00:03:20,240 --> 00:03:22,210
that are available.

60
00:03:22,210 --> 00:03:24,830
The undergrad course
numbers-- this

61
00:03:24,830 --> 00:03:26,940
is an upper level
undergraduate survey

62
00:03:26,940 --> 00:03:28,730
course in computational biology.

63
00:03:28,730 --> 00:03:33,200
And our target audience are
upper level undergraduates

64
00:03:33,200 --> 00:03:35,182
with solid biology
background and comfort

65
00:03:35,182 --> 00:03:36,390
with quantitative approaches.

66
00:03:36,390 --> 00:03:39,350
So there's one key difference
between the graduate

67
00:03:39,350 --> 00:03:42,540
and undergraduate versions,
which I'll come to in a moment.

68
00:03:42,540 --> 00:03:45,600
So the goal of this course
is to develop understanding

69
00:03:45,600 --> 00:03:49,070
of foundational methods in
computational biology that

70
00:03:49,070 --> 00:03:53,490
will enable you to
contextualize and understand

71
00:03:53,490 --> 00:03:55,990
a good portion of research
literature in a growing field.

72
00:03:55,990 --> 00:03:58,922
So if you pick up Science, or
Nature, or PLOS Computational

73
00:03:58,922 --> 00:04:03,780
Biology and you want
to read those papers

74
00:04:03,780 --> 00:04:05,500
and understand them,
after this course,

75
00:04:05,500 --> 00:04:06,960
you will have a better chance.

76
00:04:06,960 --> 00:04:09,770
We're not guaranteeing you'll be
able to understand all of them.

77
00:04:09,770 --> 00:04:15,480
But you'll be able to recognize,
perhaps, the category of paper,

78
00:04:15,480 --> 00:04:18,850
the class, and perhaps,
some of the algorithms

79
00:04:18,850 --> 00:04:20,579
that are involved.

80
00:04:20,579 --> 00:04:23,940
And for the graduate
version, another goal

81
00:04:23,940 --> 00:04:27,350
is to help you gain exposure
to research in this field.

82
00:04:27,350 --> 00:04:30,440
So it's actually
possible to do a smaller

83
00:04:30,440 --> 00:04:32,820
scale computational biology
research project on your own,

84
00:04:32,820 --> 00:04:36,700
on your laptop-- perhaps,
on ATHENA-- with relatively

85
00:04:36,700 --> 00:04:39,840
limited computational
resources and potentially

86
00:04:39,840 --> 00:04:42,190
even discover something new.

87
00:04:42,190 --> 00:04:44,720
And so we want to give
you that experience.

88
00:04:44,720 --> 00:04:47,110
And that's through
the project component

89
00:04:47,110 --> 00:04:49,400
that we'll say more
about in a moment.

90
00:04:49,400 --> 00:04:52,860
So just to make sure that
everyone's in the right class--

91
00:04:52,860 --> 00:04:55,840
this is not a systems
biology class.

92
00:04:55,840 --> 00:04:57,340
There are some more
focused systems

93
00:04:57,340 --> 00:04:58,900
biology classes
offered on campus.

94
00:04:58,900 --> 00:05:01,380
But we will cover
some topics that

95
00:05:01,380 --> 00:05:04,390
are important for
analyzing complex systems.

96
00:05:04,390 --> 00:05:08,170
This is also certainly not
a synthetic biology course.

97
00:05:08,170 --> 00:05:12,050
Some of the systems methods are
also used in synthetic biology.

98
00:05:12,050 --> 00:05:14,430
And there will be this
guest lecturer I mentioned,

99
00:05:14,430 --> 00:05:18,030
Ron Weiss, which will
cover synthetic biology.

100
00:05:18,030 --> 00:05:20,140
It's also not an
algorithms class.

101
00:05:20,140 --> 00:05:22,880
We don't assume that
you have experience

102
00:05:22,880 --> 00:05:25,290
in designing or
analyzing algorithms.

103
00:05:25,290 --> 00:05:28,220
We'll discuss various
bioinformatics algorithms.

104
00:05:28,220 --> 00:05:30,830
And you'll have the
opportunity to implement

105
00:05:30,830 --> 00:05:36,050
at least one bioinformatics
algorithm on your homework.

106
00:05:36,050 --> 00:05:39,910
But algorithms and not really
the center of the course.

107
00:05:39,910 --> 00:05:41,620
And there's one
exception to this,

108
00:05:41,620 --> 00:05:44,670
which is, those of you
who are taking 6.874

109
00:05:44,670 --> 00:05:46,480
will go to a special
recitation that

110
00:05:46,480 --> 00:05:50,350
will cover more advanced
algorithm content.

111
00:05:50,350 --> 00:05:53,190
And there will be special
homework problems for you

112
00:05:53,190 --> 00:05:53,690
as well.

113
00:05:53,690 --> 00:05:58,240
So that course really does
have more algorithm content.

114
00:05:58,240 --> 00:06:00,280
So the plan for today
is that I will just

115
00:06:00,280 --> 00:06:03,750
do a brief, anecdotal history
of computational and systems

116
00:06:03,750 --> 00:06:04,250
biology.

117
00:06:04,250 --> 00:06:07,770
This is to set the stage and
the context for the class.

118
00:06:07,770 --> 00:06:10,620
And then we'll spend
a significant amount

119
00:06:10,620 --> 00:06:14,600
of time reviewing the course
mechanics, organization,

120
00:06:14,600 --> 00:06:15,130
and content.

121
00:06:15,130 --> 00:06:17,730
Because as you'll see, it's
a little bit complicated.

122
00:06:17,730 --> 00:06:22,560
But it'll make sense once
we go over it, hopefully.

123
00:06:22,560 --> 00:06:26,660
So this is my take
on computational

124
00:06:26,660 --> 00:06:27,650
and systems biology.

125
00:06:27,650 --> 00:06:29,910
Again, it's not a
scholarly overview.

126
00:06:29,910 --> 00:06:32,770
It doesn't hit everything
important that happened.

127
00:06:32,770 --> 00:06:35,870
It just gives you
a flavor of what

128
00:06:35,870 --> 00:06:41,860
was happening in computational
biology decade by decade.

129
00:06:41,860 --> 00:06:46,060
So first of all,
where does this field

130
00:06:46,060 --> 00:06:49,470
fall in the academic
scheme of things?

131
00:06:49,470 --> 00:06:53,110
So I consider
computational biology

132
00:06:53,110 --> 00:06:55,550
to be actually part of biology.

133
00:06:55,550 --> 00:07:01,110
So in the way that genetics or
biochemistry are disciplines

134
00:07:01,110 --> 00:07:03,920
that have strategies
for understanding

135
00:07:03,920 --> 00:07:07,080
biological questions, so
does computational biology.

136
00:07:07,080 --> 00:07:09,210
You can use it to
understand a variety

137
00:07:09,210 --> 00:07:10,790
of computational
questions in gene

138
00:07:10,790 --> 00:07:15,000
regulation, many other areas.

139
00:07:15,000 --> 00:07:18,430
There is also, some
people make a distinction

140
00:07:18,430 --> 00:07:21,500
that bioinformatics is
more about building tools

141
00:07:21,500 --> 00:07:25,811
whereas computational biology
is more about using tools,

142
00:07:25,811 --> 00:07:26,310
for example.

143
00:07:26,310 --> 00:07:29,130
Although many people
don't-- it's very blurry--

144
00:07:29,130 --> 00:07:33,199
and don't try to-- people
use them in various ways.

145
00:07:33,199 --> 00:07:34,990
And then, you could
think of bioinformatics

146
00:07:34,990 --> 00:07:39,280
as being embedded in the larger
field of informatics, where you

147
00:07:39,280 --> 00:07:43,340
include tools for management
and analysis of data in general.

148
00:07:43,340 --> 00:07:47,570
And it's certainly true that
many of the core concepts

149
00:07:47,570 --> 00:07:51,170
and algorithms in bioinformatics
come from the field,

150
00:07:51,170 --> 00:07:53,980
come from computer science,
come from other branches

151
00:07:53,980 --> 00:07:57,710
of engineering, from statistics,
mathematics, and so forth.

152
00:07:57,710 --> 00:07:59,710
So it's really a cross
disciplinary field.

153
00:07:59,710 --> 00:08:03,120
And then, synthetic biology
cuts across in the sense

154
00:08:03,120 --> 00:08:05,404
that it's really an
engineering discipline.

155
00:08:05,404 --> 00:08:07,070
Because you're designing
and engineering

156
00:08:07,070 --> 00:08:09,820
synthetic molecular
cellular systems.

157
00:08:09,820 --> 00:08:11,700
But you can also use
synthetic biology

158
00:08:11,700 --> 00:08:16,220
to help understand natural
biological systems, of course.

159
00:08:16,220 --> 00:08:19,030
All right, so what was
happening, decade by decade?

160
00:08:19,030 --> 00:08:22,910
So in the '70s, there
were not genome sequences

161
00:08:22,910 --> 00:08:26,027
available or large sequence
databases of any sort.

162
00:08:26,027 --> 00:08:28,360
Except there were starting
to be some protein sequences.

163
00:08:28,360 --> 00:08:31,030
And early computational
biologists

164
00:08:31,030 --> 00:08:36,440
focused on comparing proteins,
understanding their function,

165
00:08:36,440 --> 00:08:38,419
structure, and evolution.

166
00:08:38,419 --> 00:08:40,840
And so in order to
compare proteins,

167
00:08:40,840 --> 00:08:44,810
you need a protein-- an amino
acid substitution matrix--

168
00:08:44,810 --> 00:08:47,730
a matrix that describes
how often one amino acid is

169
00:08:47,730 --> 00:08:48,750
substituted for another.

170
00:08:48,750 --> 00:08:50,990
And Margaret Dayhoff
was a pioneer

171
00:08:50,990 --> 00:08:52,630
in developing these
sorts of matrices.

172
00:08:52,630 --> 00:08:55,980
And some of the matrices she
developed, the PAM series,

173
00:08:55,980 --> 00:08:56,870
are still used today.

174
00:08:56,870 --> 00:09:00,720
And we'll discuss those
matrices early next week.

175
00:09:00,720 --> 00:09:04,310
So in terms of asking
evolutionary questions,

176
00:09:04,310 --> 00:09:09,400
two big thinkers were Russ
Doolittle and Carl Woese,

177
00:09:09,400 --> 00:09:14,930
analyzing both ribosomal RNA
sequences to study evolution.

178
00:09:14,930 --> 00:09:18,780
And Carl Woese realized,
looking at these RNA alignments,

179
00:09:18,780 --> 00:09:21,777
that actually, the prokaryotes,
which had been-- there

180
00:09:21,777 --> 00:09:23,360
was this big split
between prokaryotes

181
00:09:23,360 --> 00:09:25,530
and eukaryotes was
sort of a false split--

182
00:09:25,530 --> 00:09:27,430
that actually,
there was a subgroup

183
00:09:27,430 --> 00:09:30,750
of single-celled
anuclear organisms

184
00:09:30,750 --> 00:09:32,520
that were closer
to the eukaryotes--

185
00:09:32,520 --> 00:09:33,740
and named them the Archaea.

186
00:09:33,740 --> 00:09:36,890
So a whole kingdom of life
was recognized, really,

187
00:09:36,890 --> 00:09:38,350
by sequence analysis.

188
00:09:38,350 --> 00:09:42,040
And Russ Doolittle also
did a lot of analysis

189
00:09:42,040 --> 00:09:45,030
approaching sequences and came
up with this molecular clock

190
00:09:45,030 --> 00:09:48,860
idea, or contributed
to that idea,

191
00:09:48,860 --> 00:09:52,390
to actually build-- instead
of systematics being based

192
00:09:52,390 --> 00:09:57,770
on phenotypic characteristics,
do it on a molecular level.

193
00:09:57,770 --> 00:10:00,780
So in the '80s, the
databases started to expand.

194
00:10:00,780 --> 00:10:04,000
Sequence alignment and
search became more important.

195
00:10:04,000 --> 00:10:07,830
And various people
developed fast algorithms

196
00:10:07,830 --> 00:10:12,530
to compare protein and DNA
sequences and align them.

197
00:10:12,530 --> 00:10:15,430
So the FASTA program
was widely used.

198
00:10:15,430 --> 00:10:17,820
BLAST-- several of the authors
of BLAST are shown here--

199
00:10:17,820 --> 00:10:21,580
David Lipman, Pearson, Webb
Miller, Stephen Altschul.

200
00:10:21,580 --> 00:10:25,610
The statistics for knowing
when a BLAST search result is

201
00:10:25,610 --> 00:10:28,110
significant were developed
by Karlin and Altschul.

202
00:10:28,110 --> 00:10:31,505
And there was also progress
in gapped alignment,

203
00:10:31,505 --> 00:10:35,340
in particularly,
Smith-Waterman, shown above.

204
00:10:35,340 --> 00:10:37,700
Also progress in RNA
secondary structure prediction

205
00:10:37,700 --> 00:10:40,140
from Nusinov and Zuker.

206
00:10:40,140 --> 00:10:44,730
We'll talk about all of these
algorithms during the course.

207
00:10:44,730 --> 00:10:48,100
And there was also development
of literature databases.

208
00:10:48,100 --> 00:10:49,665
I always liked this picture.

209
00:10:49,665 --> 00:10:51,040
Many of you probably
used PubMed.

210
00:10:51,040 --> 00:10:56,400
But Al Gore was well coached
here, by these experts,

211
00:10:56,400 --> 00:10:57,530
in how to use it.

212
00:10:57,530 --> 00:11:01,480
And then, in the '90s,
computational biology

213
00:11:01,480 --> 00:11:02,736
really started to expand.

214
00:11:02,736 --> 00:11:04,360
It was driven partly
by the development

215
00:11:04,360 --> 00:11:09,540
of a microarrays, the first
genome sequences, and questions

216
00:11:09,540 --> 00:11:12,030
like how to identify
domains in a protein,

217
00:11:12,030 --> 00:11:15,520
how to identify
genes in the genome.

218
00:11:15,520 --> 00:11:18,340
It was recognized that
this family of models

219
00:11:18,340 --> 00:11:22,260
from electrical engineering,
the hidden Markov model,

220
00:11:22,260 --> 00:11:25,780
were quite useful for these sort
of sequence labeling problems.

221
00:11:25,780 --> 00:11:28,800
That was really pioneered
by Anders Krogh here,

222
00:11:28,800 --> 00:11:30,220
and David Haussler.

223
00:11:30,220 --> 00:11:32,700
And a variety of
algorithms were developed

224
00:11:32,700 --> 00:11:36,610
that performed
these useful tasks.

225
00:11:36,610 --> 00:11:39,150
There was also
important progress

226
00:11:39,150 --> 00:11:43,630
in the earliest comparative
genomic approaches,

227
00:11:43,630 --> 00:11:45,475
since you have--
the first genomes

228
00:11:45,475 --> 00:11:47,870
were sequenced in the mid
'90s, of free living organisms.

229
00:11:47,870 --> 00:11:50,510
And so you could then start
to compare these genomes

230
00:11:50,510 --> 00:11:51,350
and learn a lot.

231
00:11:51,350 --> 00:11:54,040
We'll talk a little bit
about comparative genomics.

232
00:11:54,040 --> 00:11:59,180
And there was important progress
on predicting protein structure

233
00:11:59,180 --> 00:12:00,950
from primary sequence.

234
00:12:00,950 --> 00:12:03,220
Particularly, David Baker
made notable progress

235
00:12:03,220 --> 00:12:05,190
on this Rosetta algorithm.

236
00:12:05,190 --> 00:12:07,750
So it's a biophysics
field, but it's

237
00:12:07,750 --> 00:12:12,310
very much part of
computational biology as well.

238
00:12:12,310 --> 00:12:17,380
So in the 2000s, definitely,
genome sequencing

239
00:12:17,380 --> 00:12:21,370
became very fashionable,
as you can see here.

240
00:12:21,370 --> 00:12:28,030
And the genomes of
now larger organisms,

241
00:12:28,030 --> 00:12:31,450
including human, it became
possible to sequence them.

242
00:12:31,450 --> 00:12:33,540
And then, this
introduced a huge host

243
00:12:33,540 --> 00:12:36,849
of computational challenges
in assembling the genomes,

244
00:12:36,849 --> 00:12:38,390
annotating the
genomes, and so forth.

245
00:12:38,390 --> 00:12:41,510
And we'll hear from
Professor Gifford

246
00:12:41,510 --> 00:12:44,510
about some genome
assembly topics.

247
00:12:44,510 --> 00:12:47,750
And annotation will
come up throughout.

248
00:12:51,014 --> 00:12:52,430
Actually, let's
just mention, this

249
00:12:52,430 --> 00:12:56,779
is Jim Kent, who's the guru
who did the first human genome

250
00:12:56,779 --> 00:12:58,570
assembly-- at least,
that was widely used--

251
00:12:58,570 --> 00:13:00,950
and also was involved in UCSC.

252
00:13:00,950 --> 00:13:05,810
And here, Ewan Birney
has started Ensembl

253
00:13:05,810 --> 00:13:08,904
and continues to run it today.

254
00:13:08,904 --> 00:13:10,820
You know who these other
people are, probably.

255
00:13:10,820 --> 00:13:11,830
OK.

256
00:13:11,830 --> 00:13:12,330
All right.

257
00:13:12,330 --> 00:13:19,200
So in another phase
of the last decade,

258
00:13:19,200 --> 00:13:24,070
I would say that much
of biological research

259
00:13:24,070 --> 00:13:27,570
became more high throughput
than it was before.

260
00:13:27,570 --> 00:13:29,270
So molecular biology
had traditionally,

261
00:13:29,270 --> 00:13:31,670
in the '80s and
'90s, mostly focused

262
00:13:31,670 --> 00:13:35,420
on analysis of individual
gene or protein products.

263
00:13:35,420 --> 00:13:38,970
But now it became possible,
and in widespread use,

264
00:13:38,970 --> 00:13:42,380
that you could measure the
expression of all the genes,

265
00:13:42,380 --> 00:13:45,270
in theory-- using
microarrays, for example--

266
00:13:45,270 --> 00:13:48,330
and you could start
to profile all

267
00:13:48,330 --> 00:13:49,900
of the transcripts
in the cell, all

268
00:13:49,900 --> 00:13:53,670
of the proteins in the
cell, and so forth.

269
00:13:53,670 --> 00:13:55,690
And then, a variety
of groups started

270
00:13:55,690 --> 00:13:57,510
to use some of these
high throughput data

271
00:13:57,510 --> 00:14:00,850
to study various challenges
in gene expression

272
00:14:00,850 --> 00:14:04,710
to understand how transcription
works, how splicing works,

273
00:14:04,710 --> 00:14:07,940
how microRNAs work, translation,
epigenetics, and so forth.

274
00:14:07,940 --> 00:14:14,920
And you'll hear updates on some
of that work in this course.

275
00:14:14,920 --> 00:14:17,000
Bioimage informatics,
particularly

276
00:14:17,000 --> 00:14:21,000
for developmental
biology, became popular.

277
00:14:21,000 --> 00:14:25,230
It continues to be
a new emerging area.

278
00:14:25,230 --> 00:14:31,980
Systems biology was also really
born around 2000, roughly.

279
00:14:31,980 --> 00:14:35,610
A very prominent example
would be the development

280
00:14:35,610 --> 00:14:38,890
of the first gene
regulatory network

281
00:14:38,890 --> 00:14:43,020
models that describe sea urchin
development, here, my Eric

282
00:14:43,020 --> 00:14:48,565
Davidson, as well as a whole
variety of models of other gene

283
00:14:48,565 --> 00:14:50,510
networks in the cell
that control things like

284
00:14:50,510 --> 00:14:53,990
cell proliferation,
apoptosis, et cetera.

285
00:14:56,660 --> 00:15:01,450
At the same time, a new
field of synthetic biology

286
00:15:01,450 --> 00:15:05,470
was born with the development
of some of the first

287
00:15:05,470 --> 00:15:09,610
completely artificial gene
networks that would then

288
00:15:09,610 --> 00:15:14,220
program cells to perform
desired behavior.

289
00:15:14,220 --> 00:15:16,810
So an example would be this
so-called repressilator,

290
00:15:16,810 --> 00:15:20,050
where you have a network of
three transcription factors.

291
00:15:20,050 --> 00:15:22,100
Each represses the other.

292
00:15:22,100 --> 00:15:24,127
And then, one of
them represses GFP.

293
00:15:24,127 --> 00:15:25,460
And you put these into bacteria.

294
00:15:25,460 --> 00:15:29,740
And it causes oscillations
in GFP, expressions

295
00:15:29,740 --> 00:15:33,400
that are described by these
differential equations here.

296
00:15:33,400 --> 00:15:36,856
And some of the
modeling approaches

297
00:15:36,856 --> 00:15:38,230
used in synthetic
biology will be

298
00:15:38,230 --> 00:15:43,910
covered by Professor Fraenkel
and Lauffenburger later.

299
00:15:43,910 --> 00:15:44,410
All right.

300
00:15:44,410 --> 00:15:48,840
So late 2000s, early
2010s, it's still too early

301
00:15:48,840 --> 00:15:53,130
to say, for sure, what the most
important developments will be.

302
00:15:53,130 --> 00:15:57,020
But certainly, in
the late 2000s,

303
00:15:57,020 --> 00:15:59,020
next gen sequencing--
which now probably

304
00:15:59,020 --> 00:16:00,936
should be called second
generation sequencing,

305
00:16:00,936 --> 00:16:04,280
since there may be
future generations--

306
00:16:04,280 --> 00:16:08,980
really started to transform
a whole wide variety

307
00:16:08,980 --> 00:16:12,927
of applications in biology,
from making genome sequencing--

308
00:16:12,927 --> 00:16:15,010
instead of having to be
done in the genome center,

309
00:16:15,010 --> 00:16:18,640
now an individual lab can easily
do microbial genome sequencing.

310
00:16:18,640 --> 00:16:21,560
And when needed,
it's possible, also,

311
00:16:21,560 --> 00:16:25,140
to do genome sequencing
of larger organisms.

312
00:16:25,140 --> 00:16:28,850
Transcriptome sequencing
is now routine.

313
00:16:28,850 --> 00:16:31,290
We'll hear about that.

314
00:16:31,290 --> 00:16:34,550
There are applications
for mapping protein-DNA

315
00:16:34,550 --> 00:16:36,510
interactions genome
wide, including

316
00:16:36,510 --> 00:16:38,750
both sequence specific
transcription factors

317
00:16:38,750 --> 00:16:41,710
as well as more general
factors like histones,

318
00:16:41,710 --> 00:16:43,930
protein-RNA interactions--
a method called

319
00:16:43,930 --> 00:16:47,860
CLIP-Seq-- methods for mapping
all the translated messages,

320
00:16:47,860 --> 00:16:51,320
the methylated sites in
the genome, open chromatin,

321
00:16:51,320 --> 00:16:52,730
and so forth.

322
00:16:52,730 --> 00:16:55,250
So many people contributed
to this, obviously.

323
00:16:55,250 --> 00:16:57,530
I'm just mentioning,
Barbara Wold

324
00:16:57,530 --> 00:17:03,240
was a pioneer in both
RNA-Seq as well as ChIP-Seq.

325
00:17:03,240 --> 00:17:07,380
And some of the sequencing
technologies that came out here

326
00:17:07,380 --> 00:17:09,390
are shown here.

327
00:17:09,390 --> 00:17:12,800
And we will discuss those
at the beginning of lecture

328
00:17:12,800 --> 00:17:13,960
on Thursday.

329
00:17:13,960 --> 00:17:16,089
So I encourage you to
read this review here,

330
00:17:16,089 --> 00:17:19,990
by Metzger, which covers
many of the newer sequencing

331
00:17:19,990 --> 00:17:21,219
technologies.

332
00:17:21,219 --> 00:17:22,510
And they're pretty interesting.

333
00:17:22,510 --> 00:17:24,800
As you'll see, there's
some interesting tricks,

334
00:17:24,800 --> 00:17:29,600
interesting chemistry and
image analysis tricks.

335
00:17:29,600 --> 00:17:30,100
All right.

336
00:17:30,100 --> 00:17:33,540
So that was not very scholarly.

337
00:17:33,540 --> 00:17:37,960
But if you want a proper
history, then this guy,

338
00:17:37,960 --> 00:17:41,120
Hallam Stevens, who was a
History of Science PhD student

339
00:17:41,120 --> 00:17:43,150
at Harvard and
recently graduated,

340
00:17:43,150 --> 00:17:46,250
wrote this history
of bioinformatics.

341
00:17:49,000 --> 00:17:49,910
OK.

342
00:17:49,910 --> 00:17:51,290
So let's look at the syllabus.

343
00:17:51,290 --> 00:17:55,665
So also posted on the
[INAUDIBLE] site is a syllabus.

344
00:17:55,665 --> 00:17:57,430
It looks like this.

345
00:17:57,430 --> 00:18:00,430
This is quite an
information-rich document.

346
00:18:00,430 --> 00:18:03,280
It has all the lecture
titles, all the due dates

347
00:18:03,280 --> 00:18:04,940
of all the problem
sets, and so forth.

348
00:18:04,940 --> 00:18:08,710
So please print yourself a
copy and familiarize yourself

349
00:18:08,710 --> 00:18:09,510
with it.

350
00:18:09,510 --> 00:18:13,270
So we'll just try to
look at a high level

351
00:18:13,270 --> 00:18:15,920
first, and then zoom
in to the details.

352
00:18:15,920 --> 00:18:20,600
So at a high level, if you
look in this column here,

353
00:18:20,600 --> 00:18:25,681
we've broken the course
into six different topics.

354
00:18:25,681 --> 00:18:26,180
OK?

355
00:18:26,180 --> 00:18:29,430
So there's Genomic
Analysis I, that I'll

356
00:18:29,430 --> 00:18:34,080
be teaching, which is more
classical computational

357
00:18:34,080 --> 00:18:38,980
biology, you could say-- local
alignment, global alignment,

358
00:18:38,980 --> 00:18:39,910
and so forth.

359
00:18:39,910 --> 00:18:42,295
Then, Genomic Analysis II,
which Professor Gifford will

360
00:18:42,295 --> 00:18:44,770
be teaching, covers
some newer methods

361
00:18:44,770 --> 00:18:46,390
that are required
when you're doing

362
00:18:46,390 --> 00:18:49,580
a lot of second
generation sequencing--

363
00:18:49,580 --> 00:18:51,550
the standard algorithms
are not fast enough,

364
00:18:51,550 --> 00:18:55,960
you need better
algorithms, and so forth.

365
00:18:55,960 --> 00:19:02,180
And then, I will come back and
give a few lectures on modeling

366
00:19:02,180 --> 00:19:03,300
biological function.

367
00:19:03,300 --> 00:19:06,630
This will have to do with
sequence motifs, hidden Markov

368
00:19:06,630 --> 00:19:09,217
models, and RNA
secondary structure.

369
00:19:09,217 --> 00:19:10,800
Professor Fraenkel
will then do a unit

370
00:19:10,800 --> 00:19:13,190
on proteomics and
protein structure.

371
00:19:13,190 --> 00:19:16,080
And then, there will
be an extended unit

372
00:19:16,080 --> 00:19:17,777
on regulatory networks.

373
00:19:17,777 --> 00:19:19,360
Different types of
regulatory networks

374
00:19:19,360 --> 00:19:24,480
will be covered, with most
of the lectures by Ernest,

375
00:19:24,480 --> 00:19:27,020
one by David, and one by Doug.

376
00:19:27,020 --> 00:19:30,660
And then, we'll finish up
with computational genetics,

377
00:19:30,660 --> 00:19:32,440
by David.

378
00:19:32,440 --> 00:19:37,380
And there will also be some
guest lecturers, one of them

379
00:19:37,380 --> 00:19:39,180
interspersed in
regulatory networks,

380
00:19:39,180 --> 00:19:43,780
and then two at the end, from
Ron Weiss and George Church.

381
00:19:43,780 --> 00:19:49,090
So I just wanted to point out
that in all of these topics,

382
00:19:49,090 --> 00:19:52,299
we will include some discussion
of motivating question.

383
00:19:52,299 --> 00:19:53,840
So, what are the
biological questions

384
00:19:53,840 --> 00:19:56,760
that we're seeking to address
with these approaches.

385
00:19:56,760 --> 00:19:59,590
And there will also
be some discussion

386
00:19:59,590 --> 00:20:00,990
of the experimental method.

387
00:20:00,990 --> 00:20:02,940
So for example,
in the first unit,

388
00:20:02,940 --> 00:20:04,640
it's heavy on sequence analysis.

389
00:20:04,640 --> 00:20:08,240
So we'll talk about
how sequencing is done,

390
00:20:08,240 --> 00:20:12,440
and then, quite a bit
about the interaction

391
00:20:12,440 --> 00:20:14,810
between the
experimental technology

392
00:20:14,810 --> 00:20:18,630
and the computational
analysis, which often involves

393
00:20:18,630 --> 00:20:20,470
statistical methods for
estimating the error

394
00:20:20,470 --> 00:20:23,740
rate of the experimental
method, and things like that.

395
00:20:23,740 --> 00:20:25,690
So the emphasis is on
the computational part,

396
00:20:25,690 --> 00:20:31,751
but we'll have some
discussion about experiments.

397
00:20:31,751 --> 00:20:32,750
Everyone with me so far?

398
00:20:32,750 --> 00:20:33,590
Any questions?

399
00:20:36,440 --> 00:20:39,290
OK.

400
00:20:39,290 --> 00:20:40,090
All right.

401
00:20:40,090 --> 00:20:42,050
So what are some of these
motivating questions

402
00:20:42,050 --> 00:20:43,216
that we'll be talking about?

403
00:20:43,216 --> 00:20:48,560
So what are the instructions
encoded in our genomes?

404
00:20:48,560 --> 00:20:50,760
You can think of the
genome as a book.

405
00:20:50,760 --> 00:20:52,610
But it's in this very
strange language.

406
00:20:52,610 --> 00:20:57,450
And we need to understand the
rules, the code that underlies

407
00:20:57,450 --> 00:21:01,730
a lot of research
in gene expression.

408
00:21:01,730 --> 00:21:03,640
How are chromosomes organized?

409
00:21:03,640 --> 00:21:07,840
What genes are present-- so
tools for annotating genomes.

410
00:21:07,840 --> 00:21:10,220
What regulatory
circuitry is encoded?

411
00:21:10,220 --> 00:21:13,470
You'd like to be able to
eventually look at a genome,

412
00:21:13,470 --> 00:21:15,590
understand all the
regulatory elements,

413
00:21:15,590 --> 00:21:18,960
and be able to predict
that there's some feedback

414
00:21:18,960 --> 00:21:23,650
circuit there that responds to--
a particular stimulation that

415
00:21:23,650 --> 00:21:26,360
responds to light, or
nutrient deprivation,

416
00:21:26,360 --> 00:21:28,195
or whatever it might be.

417
00:21:28,195 --> 00:21:30,320
Can the transcriptome be
predicted from the genome?

418
00:21:30,320 --> 00:21:33,040
This is a longstanding question.

419
00:21:33,040 --> 00:21:37,220
The translatome, if you
will-- well, let's say,

420
00:21:37,220 --> 00:21:40,670
the proteome can be predicted
from the transcriptome

421
00:21:40,670 --> 00:21:43,570
in the sense that we
have a genetic code

422
00:21:43,570 --> 00:21:46,850
and we can look
up those triplets.

423
00:21:46,850 --> 00:21:49,340
So there's a dream
that we would be

424
00:21:49,340 --> 00:21:53,550
able to model other
steps in gene expression

425
00:21:53,550 --> 00:21:57,780
with the precision with which
the genetic code predicts

426
00:21:57,780 --> 00:22:00,340
translation-- that we'd
be able to predict where

427
00:22:00,340 --> 00:22:02,590
the polymerase would start
transcribing, where it will

428
00:22:02,590 --> 00:22:08,100
finish transcribing, how a
transcript will be spliced,

429
00:22:08,100 --> 00:22:13,800
et cetera-- all the other
steps in gene expression.

430
00:22:13,800 --> 00:22:17,470
And that motivates a lot
of work in the field.

431
00:22:17,470 --> 00:22:19,530
Can protein function be
predicted from sequence?

432
00:22:19,530 --> 00:22:23,240
So this is a very
classical problem.

433
00:22:23,240 --> 00:22:26,680
But there are a number of new
and interesting developments

434
00:22:26,680 --> 00:22:30,900
as resolved from a lot of
this high throughput data

435
00:22:30,900 --> 00:22:34,060
generation, both in
nucleic acid sequencing

436
00:22:34,060 --> 00:22:36,830
as well as in proteomics.

437
00:22:36,830 --> 00:22:39,520
Can evolutionary history be
reconstructed from sequence?

438
00:22:39,520 --> 00:22:43,292
Again, this has been a
longstanding goal of the field.

439
00:22:43,292 --> 00:22:45,000
And a lot of progress
has been made here.

440
00:22:45,000 --> 00:22:48,400
And now most evolutionary
classifications

441
00:22:48,400 --> 00:22:52,070
are actually based on molecular
sequence at some level.

442
00:22:52,070 --> 00:22:56,440
And new species are often
defined based on sequence.

443
00:22:58,990 --> 00:22:59,800
OK.

444
00:22:59,800 --> 00:23:01,210
Other motivating questions.

445
00:23:01,210 --> 00:23:03,717
So what would you
need to measure

446
00:23:03,717 --> 00:23:05,800
if you wanted to discover
the causes of a disease,

447
00:23:05,800 --> 00:23:07,866
the mechanisms of
existing drugs,

448
00:23:07,866 --> 00:23:09,490
metabolic pathways
in a micro-organism?

449
00:23:09,490 --> 00:23:11,980
So this is a systems
biology question.

450
00:23:11,980 --> 00:23:13,350
You've got a new bug.

451
00:23:13,350 --> 00:23:16,030
It causes some disease.

452
00:23:16,030 --> 00:23:17,030
What should you measure?

453
00:23:17,030 --> 00:23:18,110
Should you sequences its genome?

454
00:23:18,110 --> 00:23:19,693
Should you sequence
its transcriptome?

455
00:23:19,693 --> 00:23:21,540
Should you do proteomics?

456
00:23:21,540 --> 00:23:23,260
What type of proteomics?

457
00:23:23,260 --> 00:23:28,330
Should you perturb the system in
some way and do a time series?

458
00:23:28,330 --> 00:23:31,050
What are the most
efficient ways?

459
00:23:31,050 --> 00:23:33,530
What information should
be gathered, and in what

460
00:23:33,530 --> 00:23:34,150
quantities?

461
00:23:34,150 --> 00:23:37,810
And how should that information
be integrated in order

462
00:23:37,810 --> 00:23:40,895
to come up with an
understanding of the physiology

463
00:23:40,895 --> 00:23:42,640
of that organisms
so that, then, you

464
00:23:42,640 --> 00:23:44,900
can know where to
intervene, what

465
00:23:44,900 --> 00:23:46,470
would be suitable drug targets?

466
00:23:50,200 --> 00:23:50,700
Yeah.

467
00:23:50,700 --> 00:23:51,770
What kind of modeling
would help you

468
00:23:51,770 --> 00:23:53,540
to use the data to
design new therapies,

469
00:23:53,540 --> 00:23:56,920
or even, in a synthetic
biology context,

470
00:23:56,920 --> 00:24:01,435
to re-engineer organisms
for new purposes?

471
00:24:01,435 --> 00:24:04,890
So microbes to
generate-- to produce

472
00:24:04,890 --> 00:24:09,820
fuel, for example, or
other useful products.

473
00:24:09,820 --> 00:24:11,670
What can we currently measure?

474
00:24:11,670 --> 00:24:15,119
What does each type of
data mean individually?

475
00:24:15,119 --> 00:24:16,660
What are the strengths
and weaknesses

476
00:24:16,660 --> 00:24:20,010
of each of the types of
high throughput approaches

477
00:24:20,010 --> 00:24:21,580
that we have?

478
00:24:21,580 --> 00:24:26,136
And how do we integrate all
the data we have on a system

479
00:24:26,136 --> 00:24:28,010
to understand the
functioning of that system?

480
00:24:28,010 --> 00:24:30,600
So these are some of
the questions that

481
00:24:30,600 --> 00:24:35,920
motivate the latter topics
on regulatory networks.

482
00:24:35,920 --> 00:24:36,840
OK.

483
00:24:36,840 --> 00:24:40,862
So let's now zoom in
and look more closely

484
00:24:40,862 --> 00:24:41,820
at the course syllabus.

485
00:24:41,820 --> 00:24:44,440
I've just broken
it into two halves,

486
00:24:44,440 --> 00:24:48,500
just so it's more readable.

487
00:24:48,500 --> 00:24:52,010
So today we're going over,
obviously, course mechanics

488
00:24:52,010 --> 00:24:52,510
mostly.

489
00:24:57,160 --> 00:25:01,120
On Thursday, we'll cover
both some DNA sequencing

490
00:25:01,120 --> 00:25:03,970
technologies and we'll talk
about local alignment on BLAST.

491
00:25:03,970 --> 00:25:05,100
More on that in a bit.

492
00:25:10,060 --> 00:25:16,790
The 6.8047 recitation--
6.874, thank you-- recitation

493
00:25:16,790 --> 00:25:19,250
will be on Friday.

494
00:25:19,250 --> 00:25:21,460
The other recitations
will start next week.

495
00:25:21,460 --> 00:25:27,022
And then, as you can see, we'll
move through the other topics.

496
00:25:27,022 --> 00:25:28,480
So each of the
instructors is going

497
00:25:28,480 --> 00:25:31,520
to briefly review their topic.

498
00:25:31,520 --> 00:25:34,210
So I won't go through
all the titles here.

499
00:25:34,210 --> 00:25:37,490
But please note,
on the left side

500
00:25:37,490 --> 00:25:41,270
here, that their assignment
due dates are marked.

501
00:25:41,270 --> 00:25:41,950
OK?

502
00:25:41,950 --> 00:25:45,530
And they're all due at
noon on the indicated day.

503
00:25:45,530 --> 00:25:48,210
And so some of these
are problem sets.

504
00:25:48,210 --> 00:25:49,910
So for example,
problem set 1 will

505
00:25:49,910 --> 00:25:53,410
be due on Thursday,
February 20 at noon.

506
00:25:53,410 --> 00:25:56,770
And some of the
other assignments

507
00:25:56,770 --> 00:26:00,200
relate to the project component
of the course, which we're

508
00:26:00,200 --> 00:26:02,610
going to talk more
about in a moment.

509
00:26:02,610 --> 00:26:04,970
In particular,
we're going to ask

510
00:26:04,970 --> 00:26:10,860
you to submit a brief
statement of your background

511
00:26:10,860 --> 00:26:14,630
and your research interests
related to forming teams.

512
00:26:14,630 --> 00:26:16,700
So the projects are
going to be done

513
00:26:16,700 --> 00:26:19,290
in teams of one
to five students.

514
00:26:19,290 --> 00:26:23,910
And in order to
facilitate especially

515
00:26:23,910 --> 00:26:26,100
cross disciplinary
teams-- we'd love

516
00:26:26,100 --> 00:26:28,340
if you interact
with, maybe, students

517
00:26:28,340 --> 00:26:31,440
in a different grad
program, or whatever--

518
00:26:31,440 --> 00:26:33,820
you'll post your background.

519
00:26:33,820 --> 00:26:36,350
You know, I'm a
first year BE student

520
00:26:36,350 --> 00:26:39,200
and I have a background
in Perl programming,

521
00:26:39,200 --> 00:26:42,470
but never done Python, or
whatever-- something like that.

522
00:26:42,470 --> 00:26:47,310
And then, I'm interested
in doing systems biology

523
00:26:47,310 --> 00:26:50,290
modeling in microbial systems,
or something like that.

524
00:26:50,290 --> 00:26:52,460
And then you can match
up your interests

525
00:26:52,460 --> 00:26:54,000
with others and form teams.

526
00:26:54,000 --> 00:26:57,870
And then you'll come up
with your own project ideas

527
00:26:57,870 --> 00:26:59,710
so that the team
and initial idea

528
00:26:59,710 --> 00:27:04,090
will be due here, February 25th.

529
00:27:04,090 --> 00:27:08,010
Then you'll need to do
some aims and so forth.

530
00:27:08,010 --> 00:27:11,220
So the project components
here, these are only for those

531
00:27:11,220 --> 00:27:13,870
taking the grad
version of the course.

532
00:27:13,870 --> 00:27:15,770
We'll make that clear later.

533
00:27:15,770 --> 00:27:16,310
OK.

534
00:27:16,310 --> 00:27:20,710
So after the first
three topics here,

535
00:27:20,710 --> 00:27:26,170
taught by myself and David,
there will be an exam.

536
00:27:26,170 --> 00:27:27,140
More on that later.

537
00:27:27,140 --> 00:27:30,990
And then there will be three
more topics, mostly taught

538
00:27:30,990 --> 00:27:34,110
by Ernest.

539
00:27:34,110 --> 00:27:38,529
And notice there are additional
assignments here related

540
00:27:38,529 --> 00:27:40,320
to the project-- so,
to research strategy--

541
00:27:40,320 --> 00:27:45,490
and the final written report,
additional problems sets.

542
00:27:45,490 --> 00:27:48,350
We'll have a guest
lecturer here.

543
00:27:48,350 --> 00:27:50,750
This will be Ron Weiss.

544
00:27:50,750 --> 00:27:53,840
Then, there will
be the second exam.

545
00:27:53,840 --> 00:27:56,105
Exams are non-cumulative,
so the second exam

546
00:27:56,105 --> 00:28:02,370
will just cover these three
topics here predominantly.

547
00:28:02,370 --> 00:28:06,300
And then there will be
another guest lecturer.

548
00:28:06,300 --> 00:28:08,290
This would be
George Church here.

549
00:28:08,290 --> 00:28:10,480
And then notice,
here, presentation.

550
00:28:10,480 --> 00:28:15,170
So those who are doing
the project component,

551
00:28:15,170 --> 00:28:19,820
those teams will be given--
assigned a time to present,

552
00:28:19,820 --> 00:28:23,490
to the class, the results
of their research.

553
00:28:23,490 --> 00:28:28,600
And you'll be graded--
the presentation will

554
00:28:28,600 --> 00:28:31,439
be part of the overall
project grade assigned

555
00:28:31,439 --> 00:28:32,230
by the instructors.

556
00:28:32,230 --> 00:28:35,640
But you'll also-- we'll
also ask all the students

557
00:28:35,640 --> 00:28:38,720
in the class to send comments
on the presentations.

558
00:28:38,720 --> 00:28:41,850
So you may find that you
get helpful suggestions

559
00:28:41,850 --> 00:28:44,410
about interpreting your data
from other people and so forth.

560
00:28:44,410 --> 00:28:48,170
So that'll be a
required component

561
00:28:48,170 --> 00:28:51,800
of the course for all students,
to attend the presentations

562
00:28:51,800 --> 00:28:53,210
and comment on them.

563
00:28:53,210 --> 00:28:56,820
And we hope that
will be a lot of fun.

564
00:28:56,820 --> 00:28:57,320
OK.

565
00:28:57,320 --> 00:28:59,220
So is this the
right course for me?

566
00:28:59,220 --> 00:29:02,350
So I just wanted to
let you know you're

567
00:29:02,350 --> 00:29:05,880
fortunate to have a rich
selection of courses

568
00:29:05,880 --> 00:29:09,080
in computational systems,
synthetic biology here at MIT.

569
00:29:09,080 --> 00:29:10,760
I've listed many of them.

570
00:29:10,760 --> 00:29:14,510
Probably not all,
but the ones that I'm

571
00:29:14,510 --> 00:29:16,820
aware of that are
available on-campus.

572
00:29:19,860 --> 00:29:22,750
757 is really only for
biology grad students.

573
00:29:22,750 --> 00:29:26,030
But the other courses listed
here are generally open.

574
00:29:26,030 --> 00:29:29,180
Some are more geared for
graduate students, some more

575
00:29:29,180 --> 00:29:29,990
undergrads.

576
00:29:29,990 --> 00:29:31,210
Some are more specialized.

577
00:29:31,210 --> 00:29:35,080
So for example, Jeff Gore's
systems biology course,

578
00:29:35,080 --> 00:29:37,030
it's more focused
on systems biology

579
00:29:37,030 --> 00:29:41,307
whereas our course covers both
computational and systems.

580
00:29:41,307 --> 00:29:43,770
So keep that in mind.

581
00:29:43,770 --> 00:29:45,270
Make sure you're
in the right place,

582
00:29:45,270 --> 00:29:48,050
that this is what you want.

583
00:29:48,050 --> 00:29:48,910
OK.

584
00:29:48,910 --> 00:29:54,000
A few notes on the textbook--
so there is a textbook.

585
00:29:54,000 --> 00:29:56,400
It's not required.

586
00:29:56,400 --> 00:29:58,080
It's called Understanding
Bioinformatics

587
00:29:58,080 --> 00:30:00,000
by Zvelebil and Baum.

588
00:30:00,000 --> 00:30:03,270
It's quite good
on certain topics.

589
00:30:03,270 --> 00:30:07,590
But it really only covers
about, maybe, a third of what

590
00:30:07,590 --> 00:30:09,410
we cover in the course.

591
00:30:09,410 --> 00:30:12,890
So there is good content
on local alignment,

592
00:30:12,890 --> 00:30:14,690
global alignment,
scoring matrices--

593
00:30:14,690 --> 00:30:17,070
the topics of the
next couple lectures.

594
00:30:17,070 --> 00:30:20,220
And I'll point you
to those chapters.

595
00:30:20,220 --> 00:30:23,130
But it's very
important to emphasize

596
00:30:23,130 --> 00:30:26,740
that the content of
the course is really

597
00:30:26,740 --> 00:30:30,410
what happens in lecture,
and on the homeworks,

598
00:30:30,410 --> 00:30:34,290
and to some extent, what
happens in recitation.

599
00:30:34,290 --> 00:30:39,270
And the textbook is just there
as a backup, if you will,

600
00:30:39,270 --> 00:30:41,990
or for those who
would like to get more

601
00:30:41,990 --> 00:30:46,860
background on the
topic or want to read

602
00:30:46,860 --> 00:30:50,610
a different description
of that topic.

603
00:30:50,610 --> 00:30:55,870
So you decide whether you
want to purchase the textbook

604
00:30:55,870 --> 00:30:56,370
or not.

605
00:30:56,370 --> 00:30:59,330
It's available at the
Coop or through Amazon.

606
00:30:59,330 --> 00:31:00,630
Shop around.

607
00:31:00,630 --> 00:31:01,895
You can find it.

608
00:31:01,895 --> 00:31:02,520
It's paperback.

609
00:31:05,030 --> 00:31:08,500
Pretty good general reference
on a variety of topics,

610
00:31:08,500 --> 00:31:11,881
but it doesn't really have
much on systems biology.

611
00:31:11,881 --> 00:31:12,380
All right.

612
00:31:12,380 --> 00:31:15,210
Another important reference
that was developed specifically

613
00:31:15,210 --> 00:31:19,230
for this course a few years
ago is the probability

614
00:31:19,230 --> 00:31:20,610
and statistics primer.

615
00:31:20,610 --> 00:31:24,350
So you'll notice that
some of the homeworks,

616
00:31:24,350 --> 00:31:27,410
particularly in the earlier
parts of the course,

617
00:31:27,410 --> 00:31:30,630
will have significant
probability and statistics.

618
00:31:30,630 --> 00:31:35,800
And we assume that you have
some background in this area.

619
00:31:35,800 --> 00:31:37,120
Many of you do.

620
00:31:37,120 --> 00:31:40,840
If you don't, you'll
need to pick that up.

621
00:31:40,840 --> 00:31:44,710
And this primer was written
to provide those topics,

622
00:31:44,710 --> 00:31:50,270
in probability especially,
that are foundational

623
00:31:50,270 --> 00:31:52,910
and most relevant to
computational biology.

624
00:31:52,910 --> 00:31:57,540
So for example, there
are some concepts

625
00:31:57,540 --> 00:32:02,600
like p-value, probability
density function, probability

626
00:32:02,600 --> 00:32:07,010
mass function, cumulative
distribution function,

627
00:32:07,010 --> 00:32:10,180
and then, common distributions,
exponential distribution,

628
00:32:10,180 --> 00:32:13,905
Poisson distribution,
extreme value distribution.

629
00:32:16,940 --> 00:32:20,800
If those are mostly sounding
familiar to you, that's good.

630
00:32:20,800 --> 00:32:24,600
If they're familiar, but
you couldn't-- you really

631
00:32:24,600 --> 00:32:27,920
don't-- you get binomial and
Poisson confused or something,

632
00:32:27,920 --> 00:32:33,100
then, definitely, you want
to consult this primer.

633
00:32:33,100 --> 00:32:39,860
So I think, looking at the
lectures and the homeworks,

634
00:32:39,860 --> 00:32:42,550
it should probably
become pretty clear

635
00:32:42,550 --> 00:32:45,500
which aspects are
going to be relevant.

636
00:32:45,500 --> 00:32:48,410
And I'll try to point
those out when possible.

637
00:32:48,410 --> 00:32:51,490
And you can also
consult your TAs

638
00:32:51,490 --> 00:32:54,631
if you're having trouble with
the probability and statistics

639
00:32:54,631 --> 00:32:55,130
content.

640
00:32:55,130 --> 00:32:57,280
So we are going to
focus, here, on, really,

641
00:32:57,280 --> 00:33:00,340
the computational biology,
bioinformatics content.

642
00:33:00,340 --> 00:33:04,792
And we might briefly review
a concept from probability,

643
00:33:04,792 --> 00:33:06,500
like, maybe, conditional
probability when

644
00:33:06,500 --> 00:33:07,666
we talk about Markov chains.

645
00:33:07,666 --> 00:33:09,620
But we're not going to
spend a lot of time.

646
00:33:09,620 --> 00:33:10,560
So if that's the
first time you've

647
00:33:10,560 --> 00:33:13,200
seen conditional probability,
you might be a little bit lost.

648
00:33:13,200 --> 00:33:18,760
So you'd be better off
reading about it in advance.

649
00:33:18,760 --> 00:33:20,270
OK?

650
00:33:20,270 --> 00:33:22,329
Questions?

651
00:33:22,329 --> 00:33:22,870
No questions?

652
00:33:22,870 --> 00:33:23,370
All right.

653
00:33:23,370 --> 00:33:26,900
Maybe it's that the video
is intimidating people.

654
00:33:26,900 --> 00:33:28,420
OK.

655
00:33:28,420 --> 00:33:31,120
All right.

656
00:33:31,120 --> 00:33:33,920
The TAs know a lot about
probability and statistics

657
00:33:33,920 --> 00:33:36,260
and will be able to help you.

658
00:33:36,260 --> 00:33:36,910
OK.

659
00:33:36,910 --> 00:33:38,820
So homework-- so I
apologize, the font

660
00:33:38,820 --> 00:33:40,270
is a little bit small here.

661
00:33:40,270 --> 00:33:42,040
So I'll try to state it clearly.

662
00:33:42,040 --> 00:33:46,080
So there are going
to be five problem

663
00:33:46,080 --> 00:33:51,640
sets that are roughly
one per topic.

664
00:33:51,640 --> 00:33:54,962
Except you'll see p set two
covers topics two and three.

665
00:33:54,962 --> 00:33:56,420
So it might be a
little bit longer.

666
00:34:01,450 --> 00:34:08,090
The way we handle students who
have to travel-- so many of you

667
00:34:08,090 --> 00:34:08,810
might be seniors.

668
00:34:08,810 --> 00:34:10,768
You might be interviewing
for graduate schools.

669
00:34:10,768 --> 00:34:13,949
Or you might have other
conflicts with the course.

670
00:34:13,949 --> 00:34:15,820
So rather than
doing that on a case

671
00:34:15,820 --> 00:34:20,050
by case basis, which, we've
found, gets very complicated

672
00:34:20,050 --> 00:34:23,000
and is not necessarily fair,
the way we've set it up

673
00:34:23,000 --> 00:34:25,620
is that the total number of
points available on the five

674
00:34:25,620 --> 00:34:28,040
homeworks is 120.

675
00:34:28,040 --> 00:34:28,639
OK?

676
00:34:28,639 --> 00:34:33,870
But the maximum score
that you can get is 100.

677
00:34:33,870 --> 00:34:40,929
So if you, for example,
were to get 90% on all five

678
00:34:40,929 --> 00:34:43,350
of the homeworks,
that would be 90%

679
00:34:43,350 --> 00:34:46,380
of 120, which would
be 108 points.

680
00:34:46,380 --> 00:34:48,980
You would get the
full 100-- you'd

681
00:34:48,980 --> 00:34:50,800
get 100% on your homework.

682
00:34:50,800 --> 00:34:53,070
That would be a perfect
score on the homework.

683
00:34:53,070 --> 00:34:53,570
OK?

684
00:34:53,570 --> 00:34:56,130
But because of that-- because
there's more points available

685
00:34:56,130 --> 00:35:02,060
than you need-- we don't
allow you to drop homeworks,

686
00:35:02,060 --> 00:35:04,730
or to do an alternate
assignment, or something

687
00:35:04,730 --> 00:35:06,690
like that.

688
00:35:06,690 --> 00:35:08,510
So the way it works
is you can basically

689
00:35:08,510 --> 00:35:12,441
miss-- as long as
you do well on, say,

690
00:35:12,441 --> 00:35:14,190
four of the homeworks--
you could actually

691
00:35:14,190 --> 00:35:16,540
miss one without
much of a penalty.

692
00:35:16,540 --> 00:35:22,090
For example, if each of the
homeworks were worth 24 points

693
00:35:22,090 --> 00:35:24,770
and you got a perfect
score on four of them,

694
00:35:24,770 --> 00:35:26,800
that would be 96 points.

695
00:35:26,800 --> 00:35:29,300
You would have an almost
perfect score on your homework

696
00:35:29,300 --> 00:35:31,380
and you could miss
that fifth homework.

697
00:35:31,380 --> 00:35:32,920
Now of course, we
don't encourage

698
00:35:32,920 --> 00:35:35,430
you to skip that homework.

699
00:35:35,430 --> 00:35:38,620
We think the homeworks are
useful and are a good way

700
00:35:38,620 --> 00:35:42,627
to solidify the information
you've gotten from lecture,

701
00:35:42,627 --> 00:35:43,710
and reading, and so forth.

702
00:35:43,710 --> 00:35:45,780
So It's good to do them.

703
00:35:45,780 --> 00:35:48,850
And doing the homeworks
will help you and perhaps

704
00:35:48,850 --> 00:35:50,160
prepare you for the exams.

705
00:35:50,160 --> 00:35:53,555
But that's the way we
handle the homework policy.

706
00:35:53,555 --> 00:35:55,930
Now I should also mention that
not all the homeworks will

707
00:35:55,930 --> 00:35:57,820
be the same number of points.

708
00:35:57,820 --> 00:36:00,770
We'll apportion the
points in proportion

709
00:36:00,770 --> 00:36:03,836
to the difficulty and length
of the homework assignment.

710
00:36:03,836 --> 00:36:05,710
So for example, the
first homework assignment

711
00:36:05,710 --> 00:36:07,790
is a little bit easier
than the others.

712
00:36:07,790 --> 00:36:11,140
So it's going to have
somewhat fewer points.

713
00:36:11,140 --> 00:36:13,990
So late assignments--
so all the homeworks

714
00:36:13,990 --> 00:36:16,410
are due at noon on
the indicated day.

715
00:36:16,410 --> 00:36:19,670
And if it's within
24 hours after that,

716
00:36:19,670 --> 00:36:21,850
you'll be eligible
for 50% credit.

717
00:36:21,850 --> 00:36:27,330
And beyond that, you don't
get any points, in part

718
00:36:27,330 --> 00:36:30,976
because the TAs will be posting
the answers to the homeworks.

719
00:36:30,976 --> 00:36:32,260
OK?

720
00:36:32,260 --> 00:36:36,290
And we want to be able to post
them promptly so that you'll

721
00:36:36,290 --> 00:36:38,480
get the answers while
those problems are

722
00:36:38,480 --> 00:36:39,667
fresh in your mind.

723
00:36:39,667 --> 00:36:40,750
Questions about homeworks?

724
00:36:44,150 --> 00:36:44,940
OK.

725
00:36:44,940 --> 00:36:45,560
Good.

726
00:36:45,560 --> 00:36:46,976
So collaboration
on problem sets--

727
00:36:46,976 --> 00:36:52,101
so we want you to
do the problem sets.

728
00:36:52,101 --> 00:36:53,350
You can do them independently.

729
00:36:53,350 --> 00:36:57,490
You can work with a friend
on them, or even in a group,

730
00:36:57,490 --> 00:36:58,820
discuss them together.

731
00:36:58,820 --> 00:37:01,065
But write up your
solutions independently.

732
00:37:01,065 --> 00:37:03,480
You don't learn
anything by copying

733
00:37:03,480 --> 00:37:05,190
someone else's solution.

734
00:37:05,190 --> 00:37:12,890
And if the TAs see duplicate
or near identical solutions,

735
00:37:12,890 --> 00:37:15,050
both of those
homeworks will get a 0.

736
00:37:15,050 --> 00:37:15,550
OK?

737
00:37:15,550 --> 00:37:16,800
And this occasionally happens.

738
00:37:16,800 --> 00:37:19,910
We don't want this
to happen to you.

739
00:37:19,910 --> 00:37:21,790
So just avoid that.

740
00:37:21,790 --> 00:37:26,860
So discuss together, but write
up your solutions separately.

741
00:37:26,860 --> 00:37:29,340
Similarly, with
programming, if you

742
00:37:29,340 --> 00:37:31,590
have a friend who's a more
experienced programmer

743
00:37:31,590 --> 00:37:34,990
than you are, by
all means, ask them

744
00:37:34,990 --> 00:37:39,270
for advice, general things, how
should I structure my program,

745
00:37:39,270 --> 00:37:43,070
do you know of a function
that generates a loop,

746
00:37:43,070 --> 00:37:45,200
or whatever it is that you need.

747
00:37:45,200 --> 00:37:50,350
But don't share code
with anyone else.

748
00:37:50,350 --> 00:37:51,434
OK?

749
00:37:51,434 --> 00:37:52,350
That would be a no-no.

750
00:37:52,350 --> 00:37:55,240
So write up your
code independently.

751
00:37:55,240 --> 00:38:00,430
And again, the graders will
be looking for identical code.

752
00:38:00,430 --> 00:38:02,280
And that will be thrown out.

753
00:38:02,280 --> 00:38:05,270
And so we don't want
to have any misconduct

754
00:38:05,270 --> 00:38:07,630
of that type occurring.

755
00:38:07,630 --> 00:38:08,130
All right.

756
00:38:08,130 --> 00:38:11,030
So recitations-- there are
three recitation sessions

757
00:38:11,030 --> 00:38:14,080
offered each week,
Wednesday at 4:00 by Peter,

758
00:38:14,080 --> 00:38:17,920
Thursday at 4:00 by Colette,
Friday at 4:00 by Tahin.

759
00:38:17,920 --> 00:38:21,990
And that is a special
recitation that's

760
00:38:21,990 --> 00:38:25,550
required for the 6.874
students-- David?

761
00:38:25,550 --> 00:38:29,090
Yes-- and has
additional AI content.

762
00:38:29,090 --> 00:38:31,320
So anyone is welcome to
go to that recitation.

763
00:38:31,320 --> 00:38:36,450
But those who are
taking 6.874 must go.

764
00:38:36,450 --> 00:38:39,180
For students registered for the
other versions of the course,

765
00:38:39,180 --> 00:38:42,660
going to recitation is optional
but strongly recommended.

766
00:38:42,660 --> 00:38:46,780
Because the TAs will go over
material from the lectures,

767
00:38:46,780 --> 00:38:49,040
material that's helpful
for the homeworks

768
00:38:49,040 --> 00:38:53,090
or for studying for exams--
in the first weeks, Python,

769
00:38:53,090 --> 00:38:55,020
probability as well.

770
00:38:55,020 --> 00:38:59,260
So go to the
recitations, particularly

771
00:38:59,260 --> 00:39:01,890
if you're having
trouble in the course.

772
00:39:01,890 --> 00:39:04,750
So Tahin's recitation
starts this week.

773
00:39:04,750 --> 00:39:08,020
And Peter and Colette's
will start next week.

774
00:39:08,020 --> 00:39:08,550
Question.

775
00:39:08,550 --> 00:39:08,880
Yes.

776
00:39:08,880 --> 00:39:11,046
AUDIENCE: Are they covering
the same material, Peter

777
00:39:11,046 --> 00:39:12,310
and Colette?

778
00:39:12,310 --> 00:39:14,840
PROFESSOR: Peter and Colette
will cover similar material

779
00:39:14,840 --> 00:39:16,830
and Tahin will cover
different material.

780
00:39:24,730 --> 00:39:28,900
Python instruction-- so
the first problem set,

781
00:39:28,900 --> 00:39:30,900
which will be
posted this evening,

782
00:39:30,900 --> 00:39:32,860
doesn't have any
programming on it.

783
00:39:32,860 --> 00:39:35,440
But if you don't
have programming,

784
00:39:35,440 --> 00:39:38,010
you need to start
learning it very soon.

785
00:39:38,010 --> 00:39:42,600
And so there will be a
significant programming problem

786
00:39:42,600 --> 00:39:44,010
on p set two.

787
00:39:44,010 --> 00:39:46,720
And we'll be posting that
problem soon-- this week,

788
00:39:46,720 --> 00:39:47,700
some time.

789
00:39:47,700 --> 00:39:50,800
And you'll want to
look at that and gauge,

790
00:39:50,800 --> 00:39:53,560
how much Python
do I need to learn

791
00:39:53,560 --> 00:39:55,690
to at least do that problem.

792
00:39:55,690 --> 00:39:57,210
So what is this
project component

793
00:39:57,210 --> 00:39:59,950
that we've been hearing about?

794
00:39:59,950 --> 00:40:02,580
So here's a more
concrete description.

795
00:40:02,580 --> 00:40:04,740
So again, this is only
for the graduate versions

796
00:40:04,740 --> 00:40:05,780
of the course.

797
00:40:05,780 --> 00:40:09,840
So students will--
basically, we've

798
00:40:09,840 --> 00:40:12,940
structured it so you
work incrementally

799
00:40:12,940 --> 00:40:16,240
toward the final
research project

800
00:40:16,240 --> 00:40:19,720
and so that we can offer
feedback and help along the way

801
00:40:19,720 --> 00:40:20,700
if needed.

802
00:40:20,700 --> 00:40:24,650
So the first assignment
will be next week.

803
00:40:24,650 --> 00:40:27,407
I think it's on Tuesday.

804
00:40:27,407 --> 00:40:29,490
I'll have more instructions
on Thursday's lecture.

805
00:40:29,490 --> 00:40:33,130
But all the students
registered for the grad version

806
00:40:33,130 --> 00:40:34,994
will submit their
background and interests

807
00:40:34,994 --> 00:40:36,410
for posting on the
course website.

808
00:40:36,410 --> 00:40:38,250
And then you can
look at those and try

809
00:40:38,250 --> 00:40:42,020
to find other students who
have, ideally, similar interests

810
00:40:42,020 --> 00:40:44,620
but perhaps somewhat
different backgrounds.

811
00:40:44,620 --> 00:40:47,949
It's particularly
helpful to have

812
00:40:47,949 --> 00:40:49,240
a strong biologist on the team.

813
00:40:49,240 --> 00:40:53,260
And perhaps, a strong
programmer would help as well.

814
00:40:53,260 --> 00:40:57,710
So then, you will
choose your teams

815
00:40:57,710 --> 00:41:02,320
and submit a project title
and one-paragraph summary,

816
00:41:02,320 --> 00:41:04,660
the basic idea of your project.

817
00:41:04,660 --> 00:41:08,720
Now we are not providing a
menu of research projects.

818
00:41:08,720 --> 00:41:11,690
It's your choice--
whatever you want

819
00:41:11,690 --> 00:41:14,511
to do, as long as it's related
to computational and systems

820
00:41:14,511 --> 00:41:15,010
biology.

821
00:41:15,010 --> 00:41:20,060
So it could be analysis of
some publicly available data.

822
00:41:20,060 --> 00:41:22,600
Could be analysis of
some data that you

823
00:41:22,600 --> 00:41:24,620
got during your rotation.

824
00:41:24,620 --> 00:41:27,560
Or for those of you who are
already in labs that you're

825
00:41:27,560 --> 00:41:29,510
actually working
on, it's totally

826
00:41:29,510 --> 00:41:32,580
fine and encouraged that the
project be something that's

827
00:41:32,580 --> 00:41:38,980
related to your main PhD work
if you've started on that.

828
00:41:38,980 --> 00:41:44,280
And it could also be more in
the modeling, some modeling

829
00:41:44,280 --> 00:41:47,780
with MATLAB or something, if
you're familiar with that.

830
00:41:47,780 --> 00:41:49,840
But, a variety of possibilities.

831
00:41:49,840 --> 00:41:51,980
We'll have more
information on this later.

832
00:41:51,980 --> 00:41:53,900
But we want you to form teams.

833
00:41:53,900 --> 00:41:58,920
The teams can work independently
or with up to four friends

834
00:41:58,920 --> 00:42:00,220
in teams of five.

835
00:42:00,220 --> 00:42:02,807
And if people want
to have a giant team

836
00:42:02,807 --> 00:42:04,390
to do some really
challenging project,

837
00:42:04,390 --> 00:42:07,110
then you can come
and discuss with us.

838
00:42:07,110 --> 00:42:11,010
And we'll see if
that would work.

839
00:42:11,010 --> 00:42:14,830
So then there's
the initial title

840
00:42:14,830 --> 00:42:16,372
and one-paragraph summary.

841
00:42:16,372 --> 00:42:18,080
We'll give you a little
feedback on that.

842
00:42:18,080 --> 00:42:21,900
And then you'll submit an
actual specific aims document--

843
00:42:21,900 --> 00:42:25,060
so with actual,
NIH-style, specific aims--

844
00:42:25,060 --> 00:42:29,400
the goal is to understand
whether this organism has

845
00:42:29,400 --> 00:42:36,280
operons or not, or some actual
scientific question-- and a bit

846
00:42:36,280 --> 00:42:38,180
about how you will
undertake that.

847
00:42:38,180 --> 00:42:43,730
And then you'll submit a longer
two-page research strategy,

848
00:42:43,730 --> 00:42:45,540
which will include,
specifically,

849
00:42:45,540 --> 00:42:47,120
we will use these data.

850
00:42:47,120 --> 00:42:49,720
We will use this software,
these statistical

851
00:42:49,720 --> 00:42:52,010
approaches-- that sort of thing.

852
00:42:52,010 --> 00:42:54,960
And then, toward the end of
the semester, a final written

853
00:42:54,960 --> 00:42:58,080
report will be due
that'll be five pages.

854
00:42:58,080 --> 00:42:59,740
You'll work on it
together, but it'll

855
00:42:59,740 --> 00:43:01,750
need to be clear who did what.

856
00:43:01,750 --> 00:43:04,130
You'll need an author
contribution statement.

857
00:43:04,130 --> 00:43:05,480
So and so did this analysis.

858
00:43:05,480 --> 00:43:08,687
So and so wrote this
section-- that sort of thing.

859
00:43:08,687 --> 00:43:10,270
And then, as I
mentioned before, there

860
00:43:10,270 --> 00:43:12,930
will be oral presentations,
by each team,

861
00:43:12,930 --> 00:43:16,335
on the last two course sessions.

862
00:43:16,335 --> 00:43:18,610
OK?

863
00:43:18,610 --> 00:43:20,390
Questions about the projects?

864
00:43:20,390 --> 00:43:23,787
There's more information on the
course info document online.

865
00:43:26,821 --> 00:43:27,320
OK?

866
00:43:27,320 --> 00:43:29,850
Good.

867
00:43:29,850 --> 00:43:36,560
Yes, so for those taking 6.874,
in addition to the project,

868
00:43:36,560 --> 00:43:39,700
there will also be
additional AI problems

869
00:43:39,700 --> 00:43:42,620
on both the p
sets-- and the exam?

870
00:43:42,620 --> 00:43:43,120
David?

871
00:43:43,120 --> 00:43:43,350
Yes.

872
00:43:43,350 --> 00:43:43,849
OK.

873
00:43:47,090 --> 00:43:49,964
And they're optional for others.

874
00:43:49,964 --> 00:43:50,840
All right.

875
00:43:50,840 --> 00:43:51,870
Good.

876
00:43:51,870 --> 00:43:52,370
OK.

877
00:43:52,370 --> 00:43:55,470
So how are we going
to do the exam?

878
00:43:55,470 --> 00:43:58,282
So as I mentioned, there's
two 80-minute exams.

879
00:43:58,282 --> 00:43:59,240
They're non-cumulative.

880
00:43:59,240 --> 00:44:00,490
So the first exam
covers, basically,

881
00:44:00,490 --> 00:44:01,448
the first three topics.

882
00:44:01,448 --> 00:44:04,369
The second exam is on
the last three topics.

883
00:44:04,369 --> 00:44:05,160
They're 80 minutes.

884
00:44:05,160 --> 00:44:08,880
They're during
normal class time.

885
00:44:08,880 --> 00:44:11,960
There's no final exam.

886
00:44:11,960 --> 00:44:15,450
The grading-- so for those
taking the undergrad version,

887
00:44:15,450 --> 00:44:20,390
the homeworks will count 36%
out of the maximum 100 points.

888
00:44:20,390 --> 00:44:24,000
And the exams will count 62%.

889
00:44:24,000 --> 00:44:25,880
And then, this
peer review, where

890
00:44:25,880 --> 00:44:27,650
there's two days
where you go, and you

891
00:44:27,650 --> 00:44:29,340
listen to presentations,
and you submit

892
00:44:29,340 --> 00:44:32,300
comments online counts 2%.

893
00:44:32,300 --> 00:44:35,970
For the graduate
Bio BE HST versions,

894
00:44:35,970 --> 00:44:43,240
it's 30% homeworks, 48% exams,
20% project, 2% peer review.

895
00:44:43,240 --> 00:44:49,800
For the EECS version, 6.874,
25% homework, 48% exams, 20%

896
00:44:49,800 --> 00:44:52,870
project, and then, 5% for
these extra AI related

897
00:44:52,870 --> 00:44:55,830
problems, and 2% peer review.

898
00:44:55,830 --> 00:44:57,440
And those should add up to 100.

899
00:44:57,440 --> 00:45:02,020
And then, in addition, we
will reward 1% extra credit

900
00:45:02,020 --> 00:45:05,320
for outstanding
class participation--

901
00:45:05,320 --> 00:45:10,405
so questions,
comments during class.

902
00:45:10,405 --> 00:45:12,385
OK.

903
00:45:12,385 --> 00:45:13,870
All right.

904
00:45:13,870 --> 00:45:15,690
So a few announcements
about topic one.

905
00:45:15,690 --> 00:45:21,180
And then, each of us will review
the topics that are coming up.

906
00:45:21,180 --> 00:45:22,855
So p set one will
be posted tonight.

907
00:45:22,855 --> 00:45:25,040
It's due February 20th, at noon.

908
00:45:25,040 --> 00:45:28,112
It involves basic microbiology,
probability, and statistics.

909
00:45:28,112 --> 00:45:29,820
It'll give you some
experience with BLAST

910
00:45:29,820 --> 00:45:31,760
and some of the
statistics associated.

911
00:45:31,760 --> 00:45:35,390
P set two will be
posted later this week.

912
00:45:35,390 --> 00:45:38,590
So you don't need to, obviously,
start on p set two yet.

913
00:45:38,590 --> 00:45:40,210
It's not due for several weeks.

914
00:45:40,210 --> 00:45:42,670
But definitely, look at
the programming problem

915
00:45:42,670 --> 00:45:45,020
to give you an idea
of what's involved

916
00:45:45,020 --> 00:45:49,750
and what to focus on when
you're reviewing your Python.

917
00:45:49,750 --> 00:45:52,110
Mentioned the
probability/stats primer.

918
00:45:52,110 --> 00:45:55,110
Sequencing technology--
so for Thursday's lecture,

919
00:45:55,110 --> 00:45:59,260
it will be very helpful if you
read the review, by Metzger,

920
00:45:59,260 --> 00:46:01,240
on next gen sequencing
technologies.

921
00:46:01,240 --> 00:46:02,770
It's pretty well written.

922
00:46:02,770 --> 00:46:06,190
Covers Illumina,
454, PACBIO, and

923
00:46:06,190 --> 00:46:11,000
a few other interesting
sequence technologies.

924
00:46:11,000 --> 00:46:12,680
Other background
reading-- so we'll

925
00:46:12,680 --> 00:46:15,200
be talking about local
alignment, global alignment,

926
00:46:15,200 --> 00:46:18,430
statistics, and similarly
matrices for the next two

927
00:46:18,430 --> 00:46:19,310
lectures.

928
00:46:19,310 --> 00:46:21,340
And chapters four and
five of the textbook

929
00:46:21,340 --> 00:46:23,760
provide a pretty good
background on these topics.

930
00:46:23,760 --> 00:46:27,332
I encourage you to take a look.

931
00:46:27,332 --> 00:46:28,220
OK.

932
00:46:28,220 --> 00:46:30,390
So I'm just going to
briefly review my lectures.

933
00:46:30,390 --> 00:46:33,260
And then we'll have David
and Ernest do the same.

934
00:46:33,260 --> 00:46:35,990
So sequencing
technologies will be

935
00:46:35,990 --> 00:46:37,300
the beginning of lecture two.

936
00:46:37,300 --> 00:46:43,170
And then we'll talk about
local ungapped sequence

937
00:46:43,170 --> 00:46:45,670
alignment-- in
particular, BLAST.

938
00:46:45,670 --> 00:46:49,140
So BLAST is something
like the Google search

939
00:46:49,140 --> 00:46:51,150
engine of bioinformatics,
if you will.

940
00:46:51,150 --> 00:46:53,860
It's one of the most
widely used tools.

941
00:46:53,860 --> 00:46:56,940
And it's important to understand
something about how it works,

942
00:46:56,940 --> 00:46:58,440
and in particular,
how to evaluate

943
00:46:58,440 --> 00:47:01,260
the significance of BLAST
hits, which are described

944
00:47:01,260 --> 00:47:07,110
by this extreme value
distribution here.

945
00:47:07,110 --> 00:47:11,900
And then, lecture three, we'll
talk about global alignment

946
00:47:11,900 --> 00:47:14,750
and introducing gaps
into sequence alignments.

947
00:47:14,750 --> 00:47:18,680
We'll talk about some
dynamic programming

948
00:47:18,680 --> 00:47:21,530
algorithms-- Needleman-Wunsch,
Smith-Waterman.

949
00:47:21,530 --> 00:47:23,930
And in lecture four,
we'll talk about

950
00:47:23,930 --> 00:47:27,840
comparative genomic analysis
of gene regulation--

951
00:47:27,840 --> 00:47:33,440
so using sequence similarity
across genomes two infer

952
00:47:33,440 --> 00:47:36,060
location of regulatory elements
such as microRNA target

953
00:47:36,060 --> 00:47:39,190
sites, other things like that.

954
00:47:39,190 --> 00:47:39,690
All right.

955
00:47:39,690 --> 00:47:44,000
So I think this is now-- oh
sorry, a few more lectures.

956
00:47:44,000 --> 00:47:48,570
Then, in the next unit,
modeling biological function,

957
00:47:48,570 --> 00:47:51,160
I'll talk about the
problem of motif finding--

958
00:47:51,160 --> 00:47:56,640
so searching a set of sequences
for a common subsequence,

959
00:47:56,640 --> 00:48:00,756
or similar subsequences,
that possess

960
00:48:00,756 --> 00:48:02,130
a particular
biological function,

961
00:48:02,130 --> 00:48:04,124
like binding to a protein.

962
00:48:04,124 --> 00:48:05,540
It's often a complex
search space.

963
00:48:05,540 --> 00:48:06,998
We'll talk about
the Gibbs sampling

964
00:48:06,998 --> 00:48:08,760
algorithm and some alternatives.

965
00:48:08,760 --> 00:48:11,410
And then, in lecture
10, I'll talk

966
00:48:11,410 --> 00:48:13,960
about Markov and hidden
Markov models, which

967
00:48:13,960 --> 00:48:18,720
have been called the Legos
of bioinformatics, which

968
00:48:18,720 --> 00:48:23,400
can be used to model a
variety of linear sequence

969
00:48:23,400 --> 00:48:25,290
labeling problems.

970
00:48:25,290 --> 00:48:30,200
And then, in the last
lecture of that unit,

971
00:48:30,200 --> 00:48:32,600
I'll talk a little
bit about protein--

972
00:48:32,600 --> 00:48:34,610
I'm sorry, about RNA--
secondary structure--

973
00:48:34,610 --> 00:48:39,970
so the base pairing of
RNAs-- predicting it

974
00:48:39,970 --> 00:48:44,170
from thermodynamic
tools, as well as

975
00:48:44,170 --> 00:48:46,620
comparative genomic approaches.

976
00:48:46,620 --> 00:48:50,190
And you'll learn
about the mfold tool

977
00:48:50,190 --> 00:48:52,390
and how you can use
a diagram like that

978
00:48:52,390 --> 00:48:56,350
to infer that this RNA may have
different possible structures

979
00:48:56,350 --> 00:48:59,360
that I can fold into,
like those shown.

980
00:48:59,360 --> 00:49:00,060
All right.

981
00:49:00,060 --> 00:49:02,820
So I'm going to pass
it off to David here.

982
00:49:11,270 --> 00:49:14,492
PROFESSOR: Thanks
very much, Chris.

983
00:49:14,492 --> 00:49:15,170
All Right.

984
00:49:15,170 --> 00:49:16,480
So I'm David Gifford.

985
00:49:16,480 --> 00:49:18,220
And I'm delighted to be here.

986
00:49:18,220 --> 00:49:20,560
It's really a
wonderfully exciting time

987
00:49:20,560 --> 00:49:22,920
in computational biology.

988
00:49:22,920 --> 00:49:24,620
And one of the reasons
it's so exciting

989
00:49:24,620 --> 00:49:29,450
is shown on this slide, which
is the production of DNA base

990
00:49:29,450 --> 00:49:31,620
sequence per
instrument over time.

991
00:49:31,620 --> 00:49:36,290
And as you can see, it's
just amazingly more efficient

992
00:49:36,290 --> 00:49:37,620
as time goes forward.

993
00:49:37,620 --> 00:49:40,570
And if you think about the
reciprocal of this curve,

994
00:49:40,570 --> 00:49:44,960
the cost per base is basically
becoming extraordinarily low.

995
00:49:44,960 --> 00:49:47,730
And this kind of
instrument allows

996
00:49:47,730 --> 00:49:49,910
us to produce
hundreds of millions

997
00:49:49,910 --> 00:49:53,751
of sequence reads for
a single experiment.

998
00:49:53,751 --> 00:49:56,000
And thus do we not only need
new computational methods

999
00:49:56,000 --> 00:49:59,086
to handle this kind
of a big data problem,

1000
00:49:59,086 --> 00:50:00,460
but we need
computational methods

1001
00:50:00,460 --> 00:50:03,730
to represent the results
in computational models.

1002
00:50:03,730 --> 00:50:06,760
And so we have multiple
challenges computationally.

1003
00:50:06,760 --> 00:50:09,710
Because modern biology
really can't be done

1004
00:50:09,710 --> 00:50:12,990
outside of a
computational framework.

1005
00:50:12,990 --> 00:50:16,690
And to summarise the
way people have adapted

1006
00:50:16,690 --> 00:50:18,950
these high throughput DNA
sequencing instruments,

1007
00:50:18,950 --> 00:50:21,010
I built this small
figure for you.

1008
00:50:21,010 --> 00:50:23,325
And you can see that--
Professor Burge will

1009
00:50:23,325 --> 00:50:26,490
be talking about DNA
sequencing next time.

1010
00:50:26,490 --> 00:50:28,310
And obviously, you
can use DNA sequencing

1011
00:50:28,310 --> 00:50:30,800
to sequence your own
genomes, or the genomes

1012
00:50:30,800 --> 00:50:33,850
of your favorite pet,
or whatever you like.

1013
00:50:33,850 --> 00:50:36,060
And so we'll talk about,
in lecture six, how

1014
00:50:36,060 --> 00:50:38,300
to actually do
genome sequencing.

1015
00:50:38,300 --> 00:50:41,590
And one of the challenges
in doing genome sequencing

1016
00:50:41,590 --> 00:50:44,380
is how to actually find
what you have sequenced.

1017
00:50:44,380 --> 00:50:48,700
And we'll be talking about how
to map sequence reads as well.

1018
00:50:48,700 --> 00:50:50,170
Another way to
use DNA sequencing

1019
00:50:50,170 --> 00:50:53,130
is to take the RNA species
present in a single cell,

1020
00:50:53,130 --> 00:50:55,590
or in a population of
cells, and convert them

1021
00:50:55,590 --> 00:50:58,680
into DNA using
reverse transcriptase.

1022
00:50:58,680 --> 00:51:02,140
Then we can sequence the DNA
and understand the RNA component

1023
00:51:02,140 --> 00:51:03,810
of the cell, which,
of course, either

1024
00:51:03,810 --> 00:51:07,260
can be used as messenger
RNA to code for protein,

1025
00:51:07,260 --> 00:51:09,917
or for structural RNAs, or
for non-coding RNAs that

1026
00:51:09,917 --> 00:51:12,250
have other kinds of functions
associated with chromatin.

1027
00:51:14,780 --> 00:51:16,700
In lecture seven,
we'll be talking

1028
00:51:16,700 --> 00:51:19,440
about protein/DNA interactions,
which Professor Burge already

1029
00:51:19,440 --> 00:51:21,620
mentioned-- the idea
that we can actually

1030
00:51:21,620 --> 00:51:24,010
locate all the regulatory
factors associated

1031
00:51:24,010 --> 00:51:27,910
with the genome using a single
high throughput experiment.

1032
00:51:27,910 --> 00:51:30,750
We do this by isolating the
proteins and their associated

1033
00:51:30,750 --> 00:51:33,400
DNA fragments and
sequencing the DNA fragments

1034
00:51:33,400 --> 00:51:37,010
using this DNA
sequencing technology.

1035
00:51:37,010 --> 00:51:40,820
So briefly, the first
thing that we'll

1036
00:51:40,820 --> 00:51:44,160
look at in lecture five is,
given a reference genome

1037
00:51:44,160 --> 00:51:47,250
sequence and a basket
of DNA sequence reads,

1038
00:51:47,250 --> 00:51:49,710
how do we build
an efficient index

1039
00:51:49,710 --> 00:51:51,970
so that we can either
map or align those

1040
00:51:51,970 --> 00:51:54,420
reads back to the
reference genome.

1041
00:51:54,420 --> 00:51:56,840
That's a very important
and fundamental problem.

1042
00:51:56,840 --> 00:51:59,510
Because if I give you a
basket of 200 million reads,

1043
00:51:59,510 --> 00:52:03,280
we need to build its
alignment very, very rapidly,

1044
00:52:03,280 --> 00:52:05,510
and quickly, and
accurately, especially

1045
00:52:05,510 --> 00:52:08,380
in the context of
repetitive elements.

1046
00:52:08,380 --> 00:52:11,300
Because a genome obviously
has many repeats in it.

1047
00:52:11,300 --> 00:52:13,520
We need to consider how
our indexing and searching

1048
00:52:13,520 --> 00:52:16,550
algorithms are going to handle
those sorts of elements.

1049
00:52:16,550 --> 00:52:18,910
In the next lecture,
we'll talk about how

1050
00:52:18,910 --> 00:52:22,075
to actually sequence a
genome and assemble it.

1051
00:52:22,075 --> 00:52:24,450
So the fundamental way that
we approach genome sequencing

1052
00:52:24,450 --> 00:52:26,370
given today's
sequencing instruments

1053
00:52:26,370 --> 00:52:29,450
is that we take intact
chromosomes at the top, which,

1054
00:52:29,450 --> 00:52:31,770
of course, are hundreds
of millions of bases long,

1055
00:52:31,770 --> 00:52:33,980
and we shatter them into pieces.

1056
00:52:33,980 --> 00:52:37,160
And then we sequence
size selected pieces

1057
00:52:37,160 --> 00:52:39,340
in a sequencing instrument.

1058
00:52:39,340 --> 00:52:41,690
And then we need to put the
jigsaw puzzle back together

1059
00:52:41,690 --> 00:52:43,870
with a computational assembler.

1060
00:52:43,870 --> 00:52:46,740
So we'll be talking about
assembly algorithms, how

1061
00:52:46,740 --> 00:52:50,390
they work, and furthermore,
how resolve ambiguities

1062
00:52:50,390 --> 00:52:53,310
as we put that puzzle
together, which often arise

1063
00:52:53,310 --> 00:52:56,280
in the context of
repetitive sequence.

1064
00:52:56,280 --> 00:52:58,480
And in the next lecture,
in lecture seven,

1065
00:52:58,480 --> 00:53:02,140
we'll be looking at how to
actually take those little DNA

1066
00:53:02,140 --> 00:53:04,460
molecules that are
associated with proteins

1067
00:53:04,460 --> 00:53:08,810
and analyze them to figure out
where particular proteins are

1068
00:53:08,810 --> 00:53:11,610
bound to the genome and how they
might regulate target genes.

1069
00:53:11,610 --> 00:53:13,970
Here we see two
different occurrences

1070
00:53:13,970 --> 00:53:16,762
of OCT4 binding events
binding proximally

1071
00:53:16,762 --> 00:53:18,595
to the SOX2 gene, which
they are regulating.

1072
00:53:21,250 --> 00:53:29,730
So that's the beginning of our
analysis of DNA sequencing.

1073
00:53:29,730 --> 00:53:32,940
And then we'll look at RNA
sequencing-- once again,

1074
00:53:32,940 --> 00:53:34,610
with going through
a DNA intermediate--

1075
00:53:34,610 --> 00:53:36,770
and ask the question,
how can we look

1076
00:53:36,770 --> 00:53:39,830
at the expression
of particular genes

1077
00:53:39,830 --> 00:53:43,620
by mapping RNA sequence
reads back onto the genome.

1078
00:53:43,620 --> 00:53:46,910
And there are two fundamental
questions we can address here,

1079
00:53:46,910 --> 00:53:51,860
which is, what is the level
of expression of a given gene,

1080
00:53:51,860 --> 00:53:55,860
and secondarily, what
isoforms are being expressed.

1081
00:53:55,860 --> 00:53:58,210
The second sets of reads,
you see up on the screen,

1082
00:53:58,210 --> 00:54:01,930
are split reads that
cross splice junctions.

1083
00:54:01,930 --> 00:54:05,760
And so by looking at how
reads align to the genome,

1084
00:54:05,760 --> 00:54:09,170
we can figure out which
particular axons are included

1085
00:54:09,170 --> 00:54:13,750
or excluded from a
particular transcript.

1086
00:54:13,750 --> 00:54:17,400
So that's the beginning of
the high throughput biology

1087
00:54:17,400 --> 00:54:19,970
genomic analysis module.

1088
00:54:19,970 --> 00:54:22,870
And I'll be returning to
talk, later in the term,

1089
00:54:22,870 --> 00:54:26,130
about computational genetics,
which, really, is a way

1090
00:54:26,130 --> 00:54:27,790
to summarize everything
we're learning

1091
00:54:27,790 --> 00:54:30,980
in the course into
an applicable way

1092
00:54:30,980 --> 00:54:34,250
to ask fundamental questions
about genome function, which

1093
00:54:34,250 --> 00:54:38,000
Professor Burge
talked about earlier.

1094
00:54:38,000 --> 00:54:43,900
Now we all have about 3
billion bases in our genomes.

1095
00:54:43,900 --> 00:54:46,350
And as you know, you
differ from the individual

1096
00:54:46,350 --> 00:54:49,950
sitting next to you in about
one in every 1,000 base pairs,

1097
00:54:49,950 --> 00:54:51,770
on average.

1098
00:54:51,770 --> 00:54:53,670
And so one question
is, how do we actually

1099
00:54:53,670 --> 00:54:57,880
interpret these differences
between genomes.

1100
00:54:57,880 --> 00:55:00,900
And how can we build
accurate computational models

1101
00:55:00,900 --> 00:55:05,410
that allow us to infer
function from genome sequence?

1102
00:55:05,410 --> 00:55:08,170
And that's a pretty
big challenge.

1103
00:55:08,170 --> 00:55:13,100
So we'll start by
asking questions about,

1104
00:55:13,100 --> 00:55:14,740
what parts of the
genome are active

1105
00:55:14,740 --> 00:55:17,440
and how could we annotate them.

1106
00:55:17,440 --> 00:55:21,030
So we can use, once again,
different kinds of sequencing

1107
00:55:21,030 --> 00:55:24,325
based assays to identify the
regions of the genome that

1108
00:55:24,325 --> 00:55:27,260
are active in any
given cellular state.

1109
00:55:27,260 --> 00:55:29,680
And furthermore, if we
look at different cells,

1110
00:55:29,680 --> 00:55:31,420
we can tell which
parts of the genome

1111
00:55:31,420 --> 00:55:33,250
are differentially active.

1112
00:55:33,250 --> 00:55:35,670
Here you can see
the active chromatin

1113
00:55:35,670 --> 00:55:38,240
during the
differentiation of an ESL

1114
00:55:38,240 --> 00:55:41,500
into a terminal type over
a 50 kilobase window.

1115
00:55:41,500 --> 00:55:44,090
And the regions of the genome
that are shaded in yellow

1116
00:55:44,090 --> 00:55:48,000
represent regions that
are differentially active.

1117
00:55:48,000 --> 00:55:52,090
And so, using this kind of
DNA seq. data and other data,

1118
00:55:52,090 --> 00:55:54,860
we can automatically
annotate the genome

1119
00:55:54,860 --> 00:55:56,840
with where the
regulatory elements are

1120
00:55:56,840 --> 00:56:00,430
and begin to understand what the
regulatory code of the genome

1121
00:56:00,430 --> 00:56:02,560
is.

1122
00:56:02,560 --> 00:56:05,600
So once we understand what
parts of the genome are active,

1123
00:56:05,600 --> 00:56:07,350
we can ask questions
about, how do they

1124
00:56:07,350 --> 00:56:10,720
contribute to some
overall phenotype.

1125
00:56:10,720 --> 00:56:14,240
And our next
lecture, lecture 19,

1126
00:56:14,240 --> 00:56:16,410
we'll be looking
at how we can build

1127
00:56:16,410 --> 00:56:19,530
a model of a
quantitative trait based

1128
00:56:19,530 --> 00:56:22,145
upon multiple loci and the
particular alleles that

1129
00:56:22,145 --> 00:56:23,870
are present at that loci.

1130
00:56:23,870 --> 00:56:27,040
So here you see an
example of a bunch

1131
00:56:27,040 --> 00:56:29,090
of different
quantitative trait loci

1132
00:56:29,090 --> 00:56:31,240
that are contributing to
the growth rate of yeast

1133
00:56:31,240 --> 00:56:33,410
in a given condition.

1134
00:56:33,410 --> 00:56:37,750
So part of our exploration
during this term

1135
00:56:37,750 --> 00:56:40,070
will be to develop
computational methods

1136
00:56:40,070 --> 00:56:43,130
to automatically identify
regions of the genome that

1137
00:56:43,130 --> 00:56:47,760
control such traits and to
assign them significance.

1138
00:56:47,760 --> 00:56:51,010
And finally, we'd like to
put all this together and ask

1139
00:56:51,010 --> 00:56:52,970
a very fundamental
question, which

1140
00:56:52,970 --> 00:56:57,290
is, how do we assign
variations in the human genome

1141
00:56:57,290 --> 00:57:00,500
to differential risk
for human disease.

1142
00:57:00,500 --> 00:57:02,880
And associated questions
are, how could we

1143
00:57:02,880 --> 00:57:06,230
assign those variants to what
best therapy would be applied

1144
00:57:06,230 --> 00:57:08,120
to the disease--
what therapeutics

1145
00:57:08,120 --> 00:57:10,370
might be used, for example.

1146
00:57:10,370 --> 00:57:13,890
And here we have
a bunch of results

1147
00:57:13,890 --> 00:57:15,976
from genome wide
association studies,

1148
00:57:15,976 --> 00:57:17,725
starting at the top
with bipolar disorder.

1149
00:57:17,725 --> 00:57:21,410
And at the bottom
is type 2 diabetes.

1150
00:57:21,410 --> 00:57:23,450
And looking along
the chromosomes,

1151
00:57:23,450 --> 00:57:27,100
we're asking which
locations along the genome

1152
00:57:27,100 --> 00:57:29,300
have variants that
are highly associated

1153
00:57:29,300 --> 00:57:32,640
with these particular diseases
in these so-called Manhattan

1154
00:57:32,640 --> 00:57:33,140
plots.

1155
00:57:33,140 --> 00:57:36,030
Because the things that
stick up look like buildings.

1156
00:57:36,030 --> 00:57:39,920
And so these sorts of studies
are yielding very interesting

1157
00:57:39,920 --> 00:57:45,444
insights into variants that are
associated with human disease.

1158
00:57:45,444 --> 00:57:47,360
And the next step, of
course, is to figure out

1159
00:57:47,360 --> 00:57:51,720
how to actually prove that
these variants are causal,

1160
00:57:51,720 --> 00:57:54,020
and also, to look
at mechanisms where

1161
00:57:54,020 --> 00:57:56,550
we might be able to address
what kinds of therapeutics

1162
00:57:56,550 --> 00:58:00,370
might be applied to deal
with these diseases.

1163
00:58:00,370 --> 00:58:01,930
So those are my two units.

1164
00:58:01,930 --> 00:58:05,090
Once again, high throughput
genomic analysis,

1165
00:58:05,090 --> 00:58:07,870
and secondarily,
computational genetics.

1166
00:58:07,870 --> 00:58:11,026
And finally, if you have
any questions about 6.874,

1167
00:58:11,026 --> 00:58:12,150
I'll be here after lecture.

1168
00:58:12,150 --> 00:58:13,850
Feel free to ask me.

1169
00:58:13,850 --> 00:58:14,350
Ernest.

1170
00:58:24,294 --> 00:58:25,970
PROFESSOR: Thank
you very much, Dave.

1171
00:58:25,970 --> 00:58:26,800
All right.

1172
00:58:26,800 --> 00:58:28,500
So in the preceding
lectures, you'll

1173
00:58:28,500 --> 00:58:30,550
have heard a lot about
the amazing things

1174
00:58:30,550 --> 00:58:32,842
we can learn from
nucleic acid sequencing.

1175
00:58:32,842 --> 00:58:35,300
And what we're going to do in
the latter part of the course

1176
00:58:35,300 --> 00:58:37,034
is look at other
modalities in the cell.

1177
00:58:37,034 --> 00:58:38,450
Obviously, there's
a lot that goes

1178
00:58:38,450 --> 00:58:41,100
on inside of cells
that's not taking place

1179
00:58:41,100 --> 00:58:42,830
at the level of DNA/RNA.

1180
00:58:42,830 --> 00:58:45,520
And so we're going to start
to look at proteins, protein

1181
00:58:45,520 --> 00:58:49,680
interactions, and ultimately,
protein interaction networks.

1182
00:58:49,680 --> 00:58:51,890
So we'll start with
the small scale,

1183
00:58:51,890 --> 00:58:54,180
looking at intermolecular
interactions

1184
00:58:54,180 --> 00:58:56,530
of the biophysics-- the
fundamental biophysics

1185
00:58:56,530 --> 00:58:57,866
of a protein structure.

1186
00:58:57,866 --> 00:59:00,240
Then we'll start to look at
protein-protein interactions.

1187
00:59:00,240 --> 00:59:03,560
And as I said, the final
level will be networks.

1188
00:59:03,560 --> 00:59:05,850
There's been an amazing
advance in our ability

1189
00:59:05,850 --> 00:59:07,830
to predict protein structure.

1190
00:59:07,830 --> 00:59:10,260
So it's always been the dream
of computational biology

1191
00:59:10,260 --> 00:59:12,750
to be able to go from
the sequence of a gene

1192
00:59:12,750 --> 00:59:15,400
to the structure of the
corresponding protein.

1193
00:59:15,400 --> 00:59:17,400
And ever since Anfinsen,
we know that, at least,

1194
00:59:17,400 --> 00:59:20,035
that should be theoretically
possible for a lot of proteins.

1195
00:59:20,035 --> 00:59:22,160
But it's been computationally
virtually intractable

1196
00:59:22,160 --> 00:59:23,351
until quite recently.

1197
00:59:23,351 --> 00:59:24,850
And a number of
different approaches

1198
00:59:24,850 --> 00:59:27,150
have allowed us to
predict protein structure.

1199
00:59:27,150 --> 00:59:28,680
So this slide
shows, for example--

1200
00:59:28,680 --> 00:59:31,120
I don't know which one's
which-- but in blue, perhaps,

1201
00:59:31,120 --> 00:59:33,810
prediction, and in red, the
true structure, or the other way

1202
00:59:33,810 --> 00:59:34,120
around.

1203
00:59:34,120 --> 00:59:35,995
But you can see, it
doesn't matter very much.

1204
00:59:35,995 --> 00:59:39,320
We're getting extremely accurate
predictions of small protein

1205
00:59:39,320 --> 00:59:40,637
structure.

1206
00:59:40,637 --> 00:59:42,220
There are a variety
of approaches here

1207
00:59:42,220 --> 00:59:43,800
that live on a spectrum.

1208
00:59:43,800 --> 00:59:47,270
On the one hand, there's
the computational approach

1209
00:59:47,270 --> 00:59:50,870
that tries to make special
purpose hardware to carry out

1210
00:59:50,870 --> 00:59:52,890
the calculations for
protein structure.

1211
00:59:52,890 --> 00:59:54,380
And that's been
wildly successful.

1212
00:59:54,380 --> 00:59:56,390
On the other end of
the extreme, there's

1213
00:59:56,390 --> 00:59:58,450
been the crowd sourcing
approach to have gamers

1214
00:59:58,450 --> 00:59:59,926
try to predict
protein structure.

1215
00:59:59,926 --> 01:00:01,425
And that turns out
to be successful.

1216
01:00:01,425 --> 01:00:04,244
And there's a lot of interesting
computational approaches

1217
01:00:04,244 --> 01:00:05,160
in the middle as well.

1218
01:00:05,160 --> 01:00:09,080
So we'll explore some of
these different strategies.

1219
01:00:09,080 --> 01:00:11,790
Then, the ability to go from
just seeing a single protein

1220
01:00:11,790 --> 01:00:13,970
structure, how these
proteins interact.

1221
01:00:13,970 --> 01:00:16,150
So now there are
pretty good algorithms

1222
01:00:16,150 --> 01:00:18,220
predicting protein-protein
interactions as well.

1223
01:00:18,220 --> 01:00:20,220
And that allows us, then,
to figure out not only

1224
01:00:20,220 --> 01:00:22,170
how these proteins
function individually,

1225
01:00:22,170 --> 01:00:25,460
but how they begin to
function as a network.

1226
01:00:25,460 --> 01:00:27,250
Now one of the things
that we've already--

1227
01:00:27,250 --> 01:00:29,541
going to be touched on in
the early parts of the course

1228
01:00:29,541 --> 01:00:32,767
are protein-DNA interactions
through sequencing approaches.

1229
01:00:32,767 --> 01:00:34,850
We want to now look at
them at a regulatory level.

1230
01:00:34,850 --> 01:00:38,910
And could you reconstruct
the regulation of a genome

1231
01:00:38,910 --> 01:00:41,010
by predicting the protein-DNA?

1232
01:00:41,010 --> 01:00:42,790
And there's been a
lot of work here.

1233
01:00:42,790 --> 01:00:45,410
We talked earlier
about Eric Davidson's

1234
01:00:45,410 --> 01:00:48,230
pioneering approaches, a lot
of interesting computational

1235
01:00:48,230 --> 01:00:51,790
approaches as well, that go from
the relatively simple models we

1236
01:00:51,790 --> 01:00:54,710
saw on those earlier slides to
these very complicated networks

1237
01:00:54,710 --> 01:00:55,750
that you see here.

1238
01:00:55,750 --> 01:00:57,860
We'll look at what
these networks actually

1239
01:00:57,860 --> 01:01:01,670
tell us, how much information
is really encoded in them.

1240
01:01:01,670 --> 01:01:04,280
They are certainly
pretty, whatever they are.

1241
01:01:04,280 --> 01:01:06,405
We'll look at other kinds
of interactions networks.

1242
01:01:06,405 --> 01:01:09,830
We'll look at genetic
interaction networks as well,

1243
01:01:09,830 --> 01:01:11,350
and perhaps, some other kinds.

1244
01:01:11,350 --> 01:01:13,244
And finally, we'll look
at computable models.

1245
01:01:13,244 --> 01:01:14,410
And what do we mean by that?

1246
01:01:14,410 --> 01:01:16,590
We mean model that
make some kind of very

1247
01:01:16,590 --> 01:01:18,770
specific prediction,
whether it's

1248
01:01:18,770 --> 01:01:21,470
a Boolean prediction-- this gene
will be on or off-- or perhaps,

1249
01:01:21,470 --> 01:01:23,340
an even more
quantitative prediction.

1250
01:01:23,340 --> 01:01:25,140
So we'll look at
logic based modeling,

1251
01:01:25,140 --> 01:01:28,460
and probably, Bayesian
networks as well.

1252
01:01:28,460 --> 01:01:32,190
So that's been a very
whirlwind tour of the course.

1253
01:01:32,190 --> 01:01:33,580
Just to go back
to the mechanics,

1254
01:01:33,580 --> 01:01:35,990
make sure you're signed
up for the right course.

1255
01:01:35,990 --> 01:01:37,970
The undergraduate
versions of the course

1256
01:01:37,970 --> 01:01:40,080
do not have a project.

1257
01:01:40,080 --> 01:01:42,810
And they certainly do not have
the artificial intelligence

1258
01:01:42,810 --> 01:01:43,601
problems.

1259
01:01:43,601 --> 01:01:46,100
Then, there are the graduate
versions that have the project,

1260
01:01:46,100 --> 01:01:48,580
but do not have the AI,
and finally, the 6.874,

1261
01:01:48,580 --> 01:01:49,480
which has both.

1262
01:01:49,480 --> 01:01:52,740
So please be sure you're
in the right class

1263
01:01:52,740 --> 01:01:56,440
so you get credit for
the right assignments.

1264
01:01:56,440 --> 01:01:59,105
And finally, if you've got
any questions about the course

1265
01:01:59,105 --> 01:02:00,480
mechanics, we have
a few minutes.

1266
01:02:00,480 --> 01:02:02,690
We can talk about it here,
in front of everybody.

1267
01:02:02,690 --> 01:02:05,340
And then, we'll all be available
after class, a little bit,

1268
01:02:05,340 --> 01:02:06,780
to answer questions.

1269
01:02:06,780 --> 01:02:09,110
So please, any questions?

1270
01:02:09,110 --> 01:02:09,610
Yes.

1271
01:02:09,610 --> 01:02:11,290
AUDIENCE: Can
undergrads do a project?

1272
01:02:11,290 --> 01:02:12,915
PROFESSOR: Can
undergrads do a project?

1273
01:02:12,915 --> 01:02:15,790
Well, they can sign up for the
graduate version, absolutely.

1274
01:02:15,790 --> 01:02:16,931
Other questions?

1275
01:02:16,931 --> 01:02:17,430
Yes.

1276
01:02:17,430 --> 01:02:22,100
AUDIENCE: Yeah, I actually can't
make either of those sessions.

1277
01:02:22,100 --> 01:02:24,100
PROFESSOR: Well, send an
email to the staff list

1278
01:02:24,100 --> 01:02:26,088
and we'll see what we can do.

1279
01:02:28,932 --> 01:02:29,880
Other questions?

1280
01:02:33,210 --> 01:02:34,133
Yes.

1281
01:02:34,133 --> 01:02:36,971
AUDIENCE: 6.877
doesn't exist anymore--

1282
01:02:36,971 --> 01:02:39,514
Computational Evolutionary
Biology, the class.

1283
01:02:39,514 --> 01:02:41,930
PROFESSOR: Oh, the thing we
listed as alternative classes?

1284
01:02:41,930 --> 01:02:42,430
Sorry.

1285
01:02:42,430 --> 01:02:43,570
PROFESSOR: We'll fix that.

1286
01:02:43,570 --> 01:02:44,278
AUDIENCE: Really?

1287
01:02:44,278 --> 01:02:46,206
I would be so happy.

1288
01:02:46,206 --> 01:02:48,140
[LAUGHTER]

1289
01:02:48,140 --> 01:02:50,260
PROFESSOR: We'll delete
it from our slides, yes.

1290
01:02:54,890 --> 01:02:58,450
We are powerful, but
not that powerful.

1291
01:02:58,450 --> 01:03:01,810
Other questions,
comments, critiques?

1292
01:03:01,810 --> 01:03:02,740
Yes, in the back.

1293
01:03:02,740 --> 01:03:05,210
AUDIENCE: Are each of the
exams equally weighted?

1294
01:03:05,210 --> 01:03:07,293
PROFESSOR: Are each of the
exams equally weighted?

1295
01:03:07,293 --> 01:03:09,745
Yes, they are.

1296
01:03:09,745 --> 01:03:10,245
Yes.

1297
01:03:10,245 --> 01:03:13,944
AUDIENCE: If we're in 6.874,
we have the additional AI

1298
01:03:13,944 --> 01:03:14,908
problems.

1299
01:03:14,908 --> 01:03:17,318
Does that mean that we have
more questions on the exam,

1300
01:03:17,318 --> 01:03:20,922
but just as much
time to do them?

1301
01:03:20,922 --> 01:03:22,880
PROFESSOR: The question
was, if you're in 6.874

1302
01:03:22,880 --> 01:03:25,250
and you have the additional
AI questions on the exam,

1303
01:03:25,250 --> 01:03:27,190
does that mean you
have more questions,

1304
01:03:27,190 --> 01:03:28,420
but the same amount of time.

1305
01:03:28,420 --> 01:03:32,900
And I believe the
answer to that is yes.

1306
01:03:32,900 --> 01:03:34,600
PROFESSOR: We may
revisit that question.

1307
01:03:34,600 --> 01:03:36,975
PROFESSOR: We will revisit
that question at a later date.

1308
01:03:40,530 --> 01:03:43,180
But course six students
are just so much smarter

1309
01:03:43,180 --> 01:03:45,576
than everyone else, right?

1310
01:03:45,576 --> 01:03:46,076
Yes.

1311
01:03:46,076 --> 01:03:47,814
AUDIENCE: Can we switch between
different versions of the class

1312
01:03:47,814 --> 01:03:48,446
by the add deadline?

1313
01:03:48,446 --> 01:03:49,720
PROFESSOR: I'm sorry,
could you say that again?

1314
01:03:49,720 --> 01:03:51,928
AUDIENCE: Can we switch
between versions of the class

1315
01:03:51,928 --> 01:03:52,920
by the add deadline?

1316
01:03:52,920 --> 01:03:55,086
PROFESSOR: Can you switch
between different versions

1317
01:03:55,086 --> 01:03:56,780
of class by the
add/drop deadline?

1318
01:03:56,780 --> 01:03:58,400
In principle, yes.

1319
01:03:58,400 --> 01:04:00,810
But if you haven't
done the work for that,

1320
01:04:00,810 --> 01:04:03,110
then you will be in trouble.

1321
01:04:03,110 --> 01:04:04,910
And there may not be
a smooth mechanism

1322
01:04:04,910 --> 01:04:06,690
for making up missed
work that late.

1323
01:04:06,690 --> 01:04:08,780
Because the add/drop
deadline is rather late.

1324
01:04:08,780 --> 01:04:11,200
So I'd encourage you,
if you're considering

1325
01:04:11,200 --> 01:04:13,300
doing the more intensive
version or not,

1326
01:04:13,300 --> 01:04:15,290
sign up for the more
intensive version.

1327
01:04:15,290 --> 01:04:17,380
You can always drop back.

1328
01:04:17,380 --> 01:04:20,080
Once we form groups
though, for the projects,

1329
01:04:20,080 --> 01:04:21,869
then, obviously,
there's an aspect

1330
01:04:21,869 --> 01:04:23,160
of letting down your teammates.

1331
01:04:23,160 --> 01:04:26,210
So think carefully, now, about
which one you want to join.

1332
01:04:26,210 --> 01:04:28,457
It'll be hard to
switch between them.

1333
01:04:28,457 --> 01:04:30,290
But we'll do whatever
the registrar requires

1334
01:04:30,290 --> 01:04:33,010
us to do in terms
of allowing it.

1335
01:04:33,010 --> 01:04:35,776
Other questions?

1336
01:04:35,776 --> 01:04:36,275
Yes.

1337
01:04:36,275 --> 01:04:40,443
AUDIENCE: If the course sites,
can we undergrads access the AI

1338
01:04:40,443 --> 01:04:42,910
problems just for
fun, to look at them?

1339
01:04:42,910 --> 01:04:43,960
PROFESSOR: Yes.

1340
01:04:43,960 --> 01:04:45,952
If the course sites
remain separate,

1341
01:04:45,952 --> 01:04:48,035
will undergrads be able
to access the AI problems?

1342
01:04:48,035 --> 01:04:49,150
The answer is yes.

1343
01:04:51,781 --> 01:04:52,280
Yes.

1344
01:04:52,280 --> 01:04:54,272
AUDIENCE: Should
students in 6.874

1345
01:04:54,272 --> 01:04:56,270
attend both presentations?

1346
01:04:56,270 --> 01:04:57,755
PROFESSOR: Should
students in 6.874

1347
01:04:57,755 --> 01:04:58,850
attend both presentations?

1348
01:04:58,850 --> 01:04:59,750
Not necessary.

1349
01:04:59,750 --> 01:05:01,270
You're welcome to
both if you want.

1350
01:05:01,270 --> 01:05:04,410
But the course six recitation
should be sufficient.

1351
01:05:07,049 --> 01:05:07,715
Other questions?

1352
01:05:10,600 --> 01:05:11,100
Yes.

1353
01:05:11,100 --> 01:05:14,214
AUDIENCE: So what's covered
in the normal recitations?

1354
01:05:14,214 --> 01:05:16,380
PROFESSOR: What's covered
in the normal recitations?

1355
01:05:16,380 --> 01:05:18,710
It'll be reinforcing material
that's in the lectures.

1356
01:05:18,710 --> 01:05:23,750
So the goal is not introduce any
new material in the course 20,

1357
01:05:23,750 --> 01:05:27,120
course seven recitations.

1358
01:05:27,120 --> 01:05:30,439
Just to clarify, not to
introduce any new material.

1359
01:05:30,439 --> 01:05:31,105
Other questions?

1360
01:05:34,501 --> 01:05:35,000
OK.

1361
01:05:35,000 --> 01:05:35,350
Great.

1362
01:05:35,350 --> 01:05:36,766
And we'll hang
around a little bit

1363
01:05:36,766 --> 01:05:40,550
to answer any remaining
questions when they come up.