Identifying Interaction Between Tfs As Regulators Or/And Targets

Question

[Dataminer]

Hi!

I have ChIP seq data of 5 TFs and I want to construct a transcriptional regulatory network between these TFs.

The approach I am using is, "If a TF is bound to a regulatory region of gene encoding one of the other TF, an edge is drawn between these two TF's, originating at the TF protein and pointing towards the bound gene". (Niu et al, 2011, Genome Research)

The window I had used to find regulatory region of gene bound to an TF is -250 bp +250 bp (is it too much or too less or if you will change this what will you keep as the window size? and why?)

Is there any other approach for the same which is simpler and more straight forward (for constructing network)?

Thank you for your time.

[Michael]

Hi, I made some minor edits, so please check. So the core question is, how far up/down-stream can a cis-regulatory site be located from the gene boundaries in eukaryotes? Then, 250 bp is clearly not enough, while 250kb is quite huge, but not going to tell you any concrete numbers, that will be attracting just more comments. (maybe +-10kb is good?)

[Casey Bergman]

see related question: http://biostar.stackexchange.com/questions/4791/how-to-assign-a-chip-chip-chip-seq-peak-to-a-target-gene

[Dataminer]

@ Casey: The tail question is important but the answer to main question, regarding the approach to construct network is more imp. to me. Anyways thank you.

Answer 1

[Marcin Cieslik]

We are mapping putative enhancer regions based on multipe epigenetic chip-seq signals. We find that the great majority 95%+ of putative enhancers is within 1Mb of the closest gene, 99% are withing 5Mb. The initial slope is pretty steep with 80% within the 100kb.

Regarding TFs binding in promoter regions anything between 2-5kb is fine, at least from a genome-wide analysis perspective.

A word of caution; these numbers are based on putative enhancers, still we observe significant enrichment of TF binding motifs and correlations with chipped TF.