regions not sorted in file

[JavenCao]

Describe the bug

dnmtools dmr "${sample_diif_file_sort}" "${T_hmrfile_sort}" "${N_hmrfile_sort}" "${outfile_1}" "${outfile_2}"

reported errors in log files:

CpGs not sorted in: ****

regions not sorted in file: *****

To Reproduce
Please find my scripts outlines as below:

sort -k1,1 -k2,2n ${N_hmrfile} > ${N_hmrfile_sort}
sort -k1,1 -k2,2n ${T_hmrfile} > ${T_hmrfile_sort}
sort -k1,1 -k2,2n ${sample_diif_file} > ${sample_diif_file_sort}
dnmtools dmr "${sample_diif_file_sort}" "${T_hmrfile_sort}" "${N_hmrfile_sort}" "${outfile_1}" "${outfile_2}"

The hmr files and diff files can be generted without errors.

Generally, I follow count, sym, workflow for short-reads WGBS-bismark bam files.

My guess

I guess some CpGs/postions present in one file is not in the other file, creating the errors?

How could I fix it? Any suggestions?

Thanks

Yujie

[andrewdavidsmith]

Hi @JavenCao . I'm happy to help clear this up. If you can provide small files that reproduce the error, that's an almost guaranteed way for me to get a fix. If you can't do that, i can try some things. First, I suggest you try to use LC_ALL=C sort ... when you sort. See if that makes a difference. Even if it helps, it shouldn't be needed. Can you provide the version number for dnmtools? Also, dnmtools doesn't have "log files" of its own, only terminal output for error messages. So if you are looking into some log files then I suspect you might be using someone's pipeline. That's not a problem, but I need to know which if those error message came first and which tools produced them. You mention dnmtools dmr but in general (and I could be wrong) each command will stop immediately on any one failure, so I would not have expected two different error messages.

[JavenCao]

1. please find a small test file reproducing the error beow.
2. This test file reproduce the error "CpGs not sorted in". Since I have many samples, in other test, the errors of "CpGs not sorted in" appears. I guess the issue should be similar.

test.zip

[andrewdavidsmith]

Thanks. Setting the locale before sorting seems to make this example run to completion. I'll put an option inside the dmr command to sort the files if the user requests. Let me know if doing like this for both files helps:

mv CRC0001N-WGBS.wgbs.hmr.bed CRC0001N-WGBS.wgbs.hmr.bed.orig
LC_ALL=C sort -k 1,1 -k 2,2g -o CRC0001N-WGBS.wgbs.hmr.bed CRC0001N-WGBS.wgbs.hmr.bed.orig

And just as I type this I realize I tested on v1.4.3, but you are on 1.4.2. It would still be helpful if you could check on 1.4.2 anyway if you don't mind.

[JavenCao]

Thank you, Andrew. This solution works perfectly for version 1.4.2. All the samples have been processed smoothly without any issues.

I truly appreciate your quick response. I aspire to be a researcher like you—efficient and responsible.