Finding histone modification enrichment specific loci using modENCODE data

Question

[dr_bantz]

Posted this on SEQanswers but no luck so far...

I have a list of interesting genomc regions (intergenic regions, a few 100 bases long) and I want to compare enrichment of various histone modifications between these regions, using previously published ChIP data. One way of doing this is to download specific data sets from modENCODE, convert them to bed files then use simple grep-type commands (probably in python) to retrieve the enrichment value for each locus. While this works in principle, it's somewhat slow. I know that there is a tool called modMINE that lets you get this sort of information through database queries, but I didn't have any luck when searching for specific histone modifications.

Is there a better way of doing this?

Answer 1

[Fidel]

One way to do what you want is to call peaks (or see if peaks are available) and then use bedtools fisher to compute a p-value.

Other option is to cluster the regions based on the histone marks signal. I recently clustered modEncode data (not only histone marks) for a set of about 3000 regions. The results were quite encouraging. I did this using deepTools heatmapper

[dr_bantz]

Very helpful - thanks! When you did the clustering analysis, did you download the ChIP tracks as wig/bed files?

[Fidel]

I downloaded the wig.gz files and converted them to bigwig files using USCS wigToBigWig. However, you need to be aware that some files prefix the chromosome name with 'chr' as in chr1 while other files do not do that (at least this happens for drosophila).