Error: logical subscript contains NAs (looks like could not annotate peak locations) about deq HOT 4 CLOSED

Hi Eduardo,

Can you run traceback() after the error occurs and send the details? As far as I can tell, the chromosome names shouldn't be an issue unless there's a mismatch between your alignment/peak files and your gtf -- does the gtf also use "1" rather than "chr1"?

Thanks!

from deq.

Yes, that's the problem: the gtf file has chromosome ids with "chr" in it, as in chr1. The alignments just show "1" in the case of chr1. I am processing the new alignments, run with new indices to correct the chromosome names, and see what happens. I'll let you know at the end of the day. Our day, that is :)

Thanks!

from deq.

OK, problem solved. The program ran to completion with no errors. Let me give some details that might help somebody else. This experiment was done in mouse, and I did the alignments with HISAT2. Instead of making my own indices, I download the indices prepared by the John Hopkins people (this is where the program comes from). Initially I downloaded the grcm38_tran indices (https://cloud.biohpc.swmed.edu/index.php/s/grcm38_tran/download). As I later found out, after getting the errors described above, the chromosome ids in these indices don't follow the standard format, chrXX, where XX is the chromosome number (or X or Y). Instead, the "chr" part is dropped. To run DEQ, you have to provide a gtf file for the appropriate genome. I used the ensembl file for mouse, mm10.ensGene.gtf (the refseq gtf, mm10.refGene.gtf, does not work). The chromosome ids in the gtf files are in the standard format, i.e., chrXX, as details above. So when DEQ tries to map the peaks to the appropriate region of the genome, it can't because the peak bedfiles will not have "chr" in the chromosome ids for the peak coordinates. This is what causes the error.

The solution was to make my own indices from the mm10 genomic sequences. There are other pre-made indices available (ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/data/mm10.tar.gz) but I have not tried them. The human hg38 (https://cloud.biohpc.swmed.edu/index.php/s/hg38_tran/download) or the rat rn6 indices (ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/data/rn6.tar.gz) don't have this problem.

I know this is a long message but I wanted to provide as much detail as possible. Thank you for your help.

Eduardo

from deq.

Thanks Eduardo!

from deq.