I have DNase-seq data (two SAM files or two BED files) in two different cell lines (no control data in any of the cell line) from ENCODE. I would like to have enriched/significantly differentially open chromatin location. In short comparison between accessible regions between: cell-A/cell-B.
Is there any tool that would do that for me? Suggested/recommended command-line option settings for that tool would also help.
thanks
•
written
by
Shumon Shumon •
What you can do is to use a peak caller between the two files. You will do this twice, once to find enriched regions in one file that are not in the other and viceversa. MACS is one of the tools that will do the job. You just need to convert your SAM files to BAM.
Normally, peak callers try to find the enriched genomic regions of a sample, but since the distrubution of reads is not homogeneous througout the genome they use a so called input file to distinguish spurious enrichments from real enrichments. In other words, peak callers assess enrichments given a background signal. However, you can replace this background signal for comparison with another comparable sample (based in the same genome assembly) that is not an input. In that case the peak caller will identify enrichments that are statistically significant in one of the samples with respect to the other sample.
MACS is not really an appropriate tool for calling DNase peaks. I use F-seq. If this is ENCODE data, there should be peaks or hotspots determined by the lab that generated the data available on the browser. You can just download the peak sets and do intersections on Galaxy.
This is a very delayed followup, but I just found this: "MACS and F-SEQ are considered among the best reported peak callers for the DNaseI-Seq" at http://chipster.csc.fi/material/chipseq/tutorials/hashem/DNase-seqIntroduction.pdf
Posting it now in case anyone else is researching DNase-Seq peak callers.
I second this - I have just completed MACS2 peak calling on ATAC-seq data (similar to DNase-seq) with good results
I would recommend loading the regions of interest (open regions from peak caller in BED format) into R as GRanges objects then doing and overlap and seeing which reads are unique for each condition. This is of course assuming you know R. I do not know of any tools that will give you this data without some programming.
maybe this article can be interesting for you: http://www.ncbi.nlm.nih.gov/pubmed/23118738
Madrigal P and Krajewski P (2012) Current bioinformatic approaches to identify DNase I hypersensitive sites and genomic footprints from DNase-seq data. Front. Gene. 3:230. doi: 10.3389/fgene.2012.00230
Log in to answer this question.
