Genome-Wide Mapping of in Vivo Protein-DNA Interactions

1
Genome-Wide Mapping of in Vivo Protein-DNA
Interactions

• Johnson et al (Science 2007)
• Presented by Leo J. Lee

2
Outline

• Background on ChIP based methods to study
  protein-DNA interactions
• Salient features of ChIPSeq
• Overview of the experimental protocol
• Data analysis pipeline used in the paper
• Important biological findings/contributions
• General discussions

3
Protein-DNA interaction

• DNA is the information carrier of almost all
  living organisms.
• Protein is the major building block of life.
• Interaction between DNA and protein play vital
  roles in the development and normal function of
  living organisms, and disease if something goes
  wrong.
• An important mechanism of protein-DNA interaction
  is via direct binding, i.e., a protein binds to a
  particular fragment of the DNA.

4
Chromatin Immunoprecipitation (ChIP)

• ChIP is a method to investigate protein-DNA
  interaction in vivo.
• The output of ChIP is enriched fragments of DNA
  that were bound by a particular protein.
• The identity of DNA fragments need to be further
  determined by a second method.

5
ChIP-chip (or ChIP-on-chip)

• ChIP-chip uses microarray technology to determine
  the identity of DNA fragments produced by ChIP.
• Typically a control sample (genomic DNA without
  going through ChIP) is used to properly define
  relative enrichment of specific sequences in the
  ChIP DNA.
• It is the dominant high-throughput technique
  before the arrival of ChIPSeq.

6
ChIPSeq Workflow
ChIP
Size Selection(200-700bp for Exp 1 150-300bp
for Exp 2)
Solexa Sequencing
Mapping onto Genome
7
ChIPSeq vs. ChIP-chip

• The experimental design of ChIPSeq is
  considerably simpler.
• ChIPSeq typically can achieve higher genomic
  coverage than ChIP-chip (also depends on read
  length vs. probe length).
• The data from ChIPSeq is arguably cleaner and
  easier to process.
• Costs are comparable (?).

8
Nice things about NRSF (REST)

• Considerable knowledge on NRSF has been
  accumulated from previous studies, which provides
  a set of true positives and negatives.
• Yet there is still room to make new discoveries,
  as illustrated in the paper.
• The DNA motif bound by NRSF (called NRSE) is long
  and well-specified.
• There is a high-quality antibody that recognizes
  NRSF efficiently.

9
ChIPSeq Workflow
ChIP
Size Selection(200-700bp for Exp 1 150-300bp
for Exp 2)
Solexa Sequencing
Mapping onto Genome
10
Sequence Mapping Filtering

• Only sequence reads mapped to a unique position
  on the human genome are kept (about 50).
• Two mismatches were allowed to accommodate
  polymorphism (and sequencing error).
• The resulting sequence read distributions are
  processed by a peak locator algorithm to find the
  local concentration of sequence hits and its
  peak.
• A minimum five fold enrichment over the control
  sampled is required.

11
ChIPSeq Peak Locator Algorithm

• Merge enriched regions within 500bp of one
  another.
• Apply a triangular 5-point smoothing and identify
  the peak as the coordinate with the greatest
  number of overlapping reads.

12
Selecting a read count threshold

• A ROC curve was obtained by analyzing true
  positives and negatives.
• A sequence read threshold of 13 was selected to
  reach 98 specificity and 87 sensitivity.

13
Precision of ChIPSeq

• Evaluated against the center of high-scoring
  canonical NRSE motifs.
• 94 of these strong motifs fall within 50bp of
  the called experimental peak.

14
Comprehensiveness of ChIPSeq

• Virtually all strong canonical NRSE motif
  instances are detectably occupied.
• Most of the sites previously studies by
  transfection analysis are also detected.

15
Motif Visualization
16
Motif Discovery

• Two new kinds of motifs are discovered
• A noncanonical motif with variable spacing
  between the left and right half sites of the
  canonical motif
• Half-site motifs
• The enrichment of both kinds of motifs are highly
  statistically significant.
• The authors are able to tell a nice evolutionary
  story about them.

17
GO enrichment analysis

• As expected, NRSF-bound loci are highly enriched
  in gene ontology (GO) terms related to neurons
  and their development.
• A group of genes encoding transcription factors
  that are critical in driving islet cell
  development in pancreas are newly discovered.
• Sequence counts for this group are modest but
  comfortably above the threshold of 13.
• The authors are able to provide strong arguments
  on the significance of this discovery.

18
Discussions
What makes this a Science paper?