MF workshop 10 © Yaron Orenstein 1 Finding sequence motifs in PBM data Workshop Project Yaron Orenstein October 2010.
Post on 28-Dec-2015
219 Views
Preview:
Transcript
MF workshop 10 © Yaron Orenstein
1
Finding sequence motifs in PBM data
Workshop Project
Yaron OrensteinOctober 2010
MF workshop 10 © Yaron Orenstein
4
DNA Pre-mRNA
protein
transcription translation
Mature
mRNA
splicing
Gene: from DNA to protein
MF workshop 10 © Yaron Orenstein
5
DNA• DNA: a “string” over the alphabet of 4 bases (nucleotides): { A, C, G, T }
• Resides in chromosomes
• Complementary strands: A-T ; C-G
Forward/sense strand: AACTTGCG
Reverse-complement/anti-sense strand: TTGAACGC
• Directional: from 5’ to 3’: (upstream) AACTTGCGATACTCCTA (downstream)
5’ end 3’ end
MF workshop 10 © Yaron Orenstein
6
Gene structure (eukaryotes)
Transcription start site (TSS)
Promoter
Transcription (RNA polymerase)
DNA
Pre-mRNAExon ExonIntron
Splicing (spliceosome)
Mature mRNA
5’ UTR 3’ UTR
Start codon Stop codonCoding region
Translation (ribosome)
Protein
Coding strand
MF workshop 10 © Yaron Orenstein
7
Translation• Codon - a triplet of bases, codes a specific
amino acid (except the stop codons); many-to-1 relation
• Stop codons - signal termination of the protein synthesis process
http://ntri.tamuk.edu/cell/ribosomes.html
MF workshop 10 © Yaron Orenstein
8
Genome sequences• Many genomes have been sequences,
including those of viruses, microbes, plants and animals.
• Human: – 23 pairs of chromosomes– 3+ Gbps (bps = base pairs) , only ~3% are
genes– ~25,000 genes
• Yeast:– 16 chromosomes– 20 Mbps– 6,500 genes
MF workshop 10 © Yaron Orenstein
9
Regulation of Expression
• Each cell contains an identical copy of the whole genome - but utilizes only a subset of the genes to perform diverse, unique tasks
• Most genes are highly regulated – their expression is limited to specific tissues, developmental stages, physiological condition
• Main regulatory mechanism – transcriptional regulation
MF workshop 10 © Yaron Orenstein
10
•Transcription is regulated primarily by transcription factors (TFs) – proteins that bind to DNA subsequences, called binding sites (BSs)
•TFBSs are located mainly (not always!) in the gene’s promoter – the DNA sequence upstream the gene’s transcription start site (TSS)
•BSs of a particular TF share a common pattern, or motif
•Some TFs operate together – TF modules
TFTF
Gene5’ 3’
BSBSTSS
Transcriptional regulation
MF workshop 10 © Yaron Orenstein
11
•Consensus (“degenerate”) string:
TFBS motif models
gene 7
gene 9
gene 5
gene 3gene 2
gene 4
gene 6
gene 8
gene 10
gene 1AACTGT
CACTGTCACTCT
CACTGT
AACTGT
AC ACT
CGT
•Statistical models…•Motif logo representation
MF workshop 10 © Yaron Orenstein
12
Human G2+M cell-cycle genes:The CHR – NF-Y module
CDCA3 (trigger of mitotic entry 1)CTCAGCCAATAGGGTCAGGGCAGGGGGCGTGGCGGGAAGTTTGAAACT -18
CDCA8 (cell division cycle associated 8)TTGTGATTGGATGTTGTGGGA…[25bp]…TGACTGTGGAGTTTGAATTGG +23
CDC2 (cell division control protein 2 homolog)CTCTGATTGGCTGCTTTGAAAGTCTACGGGCTACCCGATTGGTGAATCCGGGGCCCTTTAGCGCGGTGAGTTTGAAACTGCT 0
CDC42EP4 (cdc42 effector protein 4)GCTTTCAGTTTGAACCGAGGA…[25bp]…CGACGGCCATTGGCTGCTGC -110
CCNB1 (G2/mitotic-specific cyclin B1)AGCCGCCAATGGGAAGGGAG…[30bp]…AGCAGTGCGGGGTTTAAATCT +45
CCNB2 (G2/mitotic-specific cyclin B2)TTCAGCCAATGAGAGT…[15bp]…GTGTTGGCCAATGAGAAC…[15bp]…GGGCCGCCCAATGGGGCGCAAGCGACGCGGTATTTGAATCCTGGA +10
BS’s are short, non-specific, hiding in both strands and at various locations along the promoters
TFs: NF-Y , CHR
MF workshop 10 © Yaron Orenstein
1313
Protein Binding Microarrays
Berger et al, Nat. Biotech 2006
• Generate an array of double-stranded DNA with all possible k-mers
• Detect TF binding to specific k-mers
MF workshop 10 © Yaron Orenstein
1515
PBM - implementation
• Use 60-mers (Agilent): 25nt constant primer + 35nt variable region
• De Bruijn seq of all 10-mers (410 long) split into 35nt long fragments with 9nt overlap
• ~40K probes• For each 8-mer, combine signals from
all probes that contain it (or differ in 1nt) to obtain its binding score
MF workshop 10 © Yaron Orenstein
16
The computational challenge
• Input: PBM data (sequences and binding scores) of one TF.
• Goal: Find a motif (PWM) that is the binding site of that TF.
• Intuition: sequences that match the motif (on one of the two possible strands!) are expected to have high binding scores.
MF workshop 10 © Yaron Orenstein
18
General goals• Research
- Learn about known solutions- Trial and error with training data
• Develop software from A-Z:– Design– Implementation (Optimization) – Execution & analysis of test data
• A taste of bioinformatics• Have fun• Get credit…
MF workshop 10 © Yaron Orenstein
19
The computational task
• Given a set of PBM data of different TFs.
• Find the binding site motif in PWM format of each TF.
• Main challenges:– Performance (time, memory)– Accuracy
MF workshop 10 © Yaron Orenstein
20
InputFile with 41,923 lines, each containing a
probe sequence of length 35 and binding intensity.
<sequence 35bp> \t <intensity> \n
MF workshop 10 © Yaron Orenstein
21
Input (II)• For the training data, an additional
PWM file will be supplied for each PBM data set.
A: <freq1> <freq2> … <freq10>
C: <freq1> … <freq10>
G: …
T: …
• Separated by \t and \n.• All lines must contain same number of
frequencies (10 is just an example).
MF workshop 10 © Yaron Orenstein
22
Input (III)
You will be given:1. 10 training sets (PBM data + PWM)2. 4 test sets (PBM data). You have to
provide the PWM.3. In the final project presentation, you
will be given an online test set (PBM data) and your software will be applied to it.
MF workshop 10 © Yaron Orenstein
23
Output1. A PWM file describing the binding
site found in the given PBM file.2. The PWM in motif logo format (i.e.
displayed on the screen).
The file logo.zip contains a java
package with the code that will easily display your motif.
bits = 2 - entropy
MF workshop 10 © Yaron Orenstein
Output (II)3. Show graphically how well your
motif predicts the binding intensity.
• One example (note it’s not PWM):
24
MF workshop 10 © Yaron Orenstein
25
Ranking 8-mers• One possible way to start: rank the 8-
mers in some way. Scores for example:1. Signal average.2. Signal median.
• You can think of other scores that incorporate more information, e.g. position in probe sequence.
• This is just an example. You can think of other ways to start.
MF workshop 10 © Yaron Orenstein
26
• Then, you can align the significant 8-mers.
• You may take into account the relative score.
• Don’t forget about the reverse complement!
• Example: Cebpb TF
Alignment procedure
MF workshop 10 © Yaron Orenstein
Enrichment scores
• To test how good your motif is, you can use an enrichment score.
• An enrichment score tests how good the motif distinguishes between high-ranking probes and the rest of the probes.
27
MF workshop 10 © Yaron Orenstein
Hypergeometric probability
28
drawn not drawn total
white k m − k m
black n − k N + k − n − m
N − m
total n N − n N
MF workshop 10 © Yaron Orenstein
Hypergeometric enrichment score
• Let B and T (T B) denote the BG and target sets, respectively, and let b and t denote the subset of probes from the BG and target set, respectively, that contain at least one occurrence of the motif.
29
min(| |,| |)
| |
| | | | | |
| |( | |,| |,| |,| |)
| |
| |
T b
i t
b B b
i T iHG tail B T b t
B
T
MF workshop 10 © Yaron Orenstein
Hypergeometric score (2)• The HG enrichment score computes
the probability of observing at least |t| target sequences with a motif occurrence, under the null hypothesis that the probes in the target set were drawn randomly, independently, and without replacement from the BG set.
• Code is provided in math.zip
30
MF workshop 10 © Yaron Orenstein
Wilcoxon-Mann-Whitney (WMW) enrichment score
• Foreground probes are all those containing a match, background are all the others.
• B and F are the sizes of background and foreground, respectively.
• ρB and ρF are the sums of the background and foreground ranks.
• Read more in supplementary info (Berger06).
31
FBFBarea FB 1
MF workshop 10 © Yaron Orenstein
Deciding the length of the motif
• Another challenge is to decide the length of the motif.
• Most binding site are 6-12 bp long.• You should consider the
information each position contains and decide on the length accordingly.
32
MF workshop 10 © Yaron Orenstein
Scoring your PWM
• One way to score your motif is by ranking the probe sequences according to a match score.
• You may use the given code for match score.
• Compare the ranking of the probes you got to the ranking according to binding intensities. There are different correlation score for that.
33
MF workshop 10 © Yaron Orenstein
Match Score between PWMs
• Already implemented for you:1.Euclidian Distance:
2.Pearson Correlation Coefficient
3.KL Divergence
34
MF workshop 10 © Yaron Orenstein
35
Implementation• Java (Eclipse) ; Linux (Other languages are
possible, but will not participate in bonus).• Input: one single argument PBM filename• Output: PWM file, PWM presented in logo
and graphical presentation of PWM matching distribution among probes.
• Packages for motif logo and statistical scores will be supplied
• Time performance will be measured• Reasonable documentation• Separate packages for data-structures,
scores, GUI, I/O, etc.
MF workshop 10 © Yaron Orenstein
Submission• Printed design document.• Printed code – for comments and
remarks.• Printed results document – for each test
set PWM logo + how good your result in terms of correlation to the probes ranks.
• 4 PWM files, e.g. Test_1.pwm (submitted by email).
• Executable for the online test.
36
MF workshop 10 © Yaron Orenstein
Grade• 20% for the design • 30% for the implementation (20% for modularity,
clarity, documentation, 10% for efficiency) • 30% for the performance and experimental results
(20% for the accuracy on the 4 test queries and 10% for the accuracy on the online test query)
• 20% for the final report and presentation • 10% bonus to the group with the most accurate
results • 10% bonus for the group with the fastest
implementation
37
MF workshop 10 © Yaron Orenstein
Bonus grading
• Accuracy will be determined using the provided code that compares two PWMs.
• We will take the average of runs on several different PBM data sets.
• Running time will be measured in java implementation, and the average will be taken.
38
MF workshop 10 © Yaron Orenstein
Schedule 1.First progress report 23/112.Design document 21/123.Final presentation 16/2
• We shall meet with each group on each of these dates – mark your calendars!
• Schedule can be made earlier if you are ready.
• You are always welcome to meet us. Contact us by email.
39
MF workshop 10 © Yaron Orenstein
40
Design document• Due in week 12 (21/12).• 3-5 pages (Word), Hebrew/English• Briefly describe main goal, input
and output of program• Describe main data structures,
algorithms, and scores.• Meet with me before submission.
MF workshop 10 © Yaron Orenstein
Reference
• Berger MF, Philippakis AA, Quershi AM, He FS, EstepIII PW, Bulyk ML. Compact, universal DNA microarrays to comprehensively determine transcription-factor binding site specificities. Nature biotechnology. 2006;338:1429-1435.
Very important! Read: the_brain.bwh.harvard.edu/UPBMseqn/suppl_methods.doc
• Chen X, Hughes TR, Morris Q. RankMotif++: a motif-search algorithm that accounts for relative ranks of K-mers in binding transcription factors. Bioinformatics. 2007 Jul 1;23(13):i72-79.
41
top related