Chip Seq Peak Annotation ... .. Missing Genes In Annotation

Question

[Dataminer]

Hi!

I am facing a very unique problem, I have a list of regions (in a bed file) and using GREAT and HOMER I annotated these regions to the near by genes. The problem is I do not get genes for few regions in a window of 500 base pairs ( that is fine, if the gene is not present), when I look up for these regions in UCSC browser I do find these regions present over a UCSC and RefSeq gene.

As an example, following are the regions

chr1     9747627            9748127   for this region PIK3CD can be seen in UCSC
chr1      6659871             6660371    for this  region KLHL21

I have nearly 300 regions like this.

I am using hg19

Have you ever encountered such a problem? If yes how you have solved it? If no, what will be your approach?

Thank you,

[Sukhi Singh]

I just figured its hg19, please update in post for other people

Answer 1

[Sukhi Singh]

Are you annotating using same genome build, with every new build the co-ordinates are shifted, might be a reason. (hg18 vs hg19)

I would download the list of genes from UCSC (refseq genes) and manually intersect them, the thing you are saying should not happen, might a small parameter problem.

[Dataminer]

Yes, I am using hg19.

[Sukhi Singh]

Can you show your homer command used for annotation

[Dataminer]

annotatePeaks.pl region.txt hg19 > annotated_peaks

[Sukhi Singh]

Yeah, its fine, I just tested my annotation with homer, results are little different. Like it just annotated in summary, that 68 peaks are in promoters but I am positive for >300 peaks. Good idea to ask the author or you can manually look into the python code, how its happening.

[Dataminer]

I think the safest way is to take the intersection of file containing Chr TSS TSE & GeneSymbol with my peak file.

[Sukhi Singh]

Yes, thats what I said, by that you will be sure, what you are doing !!