suggestion: coverageBed

40 views
Skip to first unread message

Jason Tsai

unread,
Feb 23, 2012, 3:28:15 PM2/23/12
to bedtools-discuss
Hello!

We love bedtools ! :)

One suggestion: we have been using coverageBed to compute the "mean/
median depth" the genes/region of interest on the genome from parsing
the result of coverageBed -d -abam xxx.bam -b xxx.gff option.

It would be extremely nice, if, a similar option is created like -d or
-hist, to output the median and 95% CI coverage of the region of
interest (per gff file). So if we have a 10bp region with
1,2,3,4,5,6,7,8,9,10 X depth of coverage, this option can report
0.5,5,9.5 X.

Or top 5 most occurring depth of coverage per gff line? So a bit like
parsing the -hist output.

I think this would be useful for various parties.

Best,
Jason





Aaron Quinlan

unread,
Feb 27, 2012, 1:21:17 PM2/27/12
to bedtools...@googlegroups.com
Hi Jason,

Thanks for using the tools and for the suggestion. The challenge with these types of options is supporting every threshold that a user may want. In your case, you are interested in the 95% CI, but other might like different thresholds. I appreciate the utility of such an option, but prefer that these metrics be computed directly from the raw counts.

One compromise would be to add options to the "groupby" command for calculating CI. This way, one could pipe coverageBed -d or -hist output to groupby and calculate the metrics one is interested in.

Does this sound reasonable?

Best,
Aaron

Reply all
Reply to author
Forward
0 new messages