Identification Of Genomic Regions Where Multiple Tf Binds.

Question

[Dataminer]

Hi!

I have peak called data of 8 transcription factors (using MACS on BED files).

The format of each file is:
Chr Chr_Start Chr_Stop

Basically three columns.

I want to find the regions where atleast 4 TF bind (Any 4).

Note: I already have a union of these regions in a file and have counted tags for each TF in these region.

Thank you,

[Dataminer]

The Answer is here: http://biostar.stackexchange.com/questions/13548/bedtools-compare-multiple-bed-files

Answer 1

[Hanif Khalak]

Several previous BioStar entries address the basic "overlapping intervals" problem:

• Finding an genomic intervals shared for a given individual as called by two or more algorithms

• What is the best library for handling interval logic?

• Fast interval intersection methodologies

• Quick programming challenge: calculate common and unique regions from a list of chromosome segments

Answer 2

[Ian Simpson]

Well one of the first things you need to decide is how you define a 'region'. Fixed size, minimum TF density etc. If you can (albeit fairly arbitrarily) decide this it's simply a case of windowing across the sequences and keeping running totals for the TFs in the bins. You can then summarise the window counts across the 8 and only keep the ones where the sum is greater than 4.

If I were doing this I would hack together a quick Perl script to do the job. I wouldn't think this would take too long to do if you're familiar with scripting.

[Hanif Khalak]

@Ian: I like a good Perl hack myself - still, interval logic is best dealt with through a library/module. It's not quite as sinister as regex for XML, but I've tried it from scratch and there are a number of gotchas that make anything quick/throw-away prone to error