Incompatibility with masked reference genomes

Incompatibility with masked reference genomes

call-macs2 fails if a masked reference (where Ns are used to indicate repetitive regions) is used to build a custom genome reference database.

run_shell_cmd: PID=70151, CMD=bedtools intersect -a ${SAMPLE}.trim.merged.nodup.tn5.tagAlign.tmp1 -b ${SAMPLE}.trim.merged.nodup.tn5.pval0.01.300K.bfilt.narrowPeak.tmp2 -wa -u | wc -l

Traceback (most recent call last):
  File "/users/nicolerg/anaconda2/envs/encode-atac-seq-pipeline/bin/encode_macs2_atac.py", line 210, in <module>
    main()
  File "/users/nicolerg/anaconda2/envs/encode-atac-seq-pipeline/bin/encode_macs2_atac.py", line 200, in main
    frip_qc = frip( args.ta, bfilt_npeak, args.out_dir)
  File "/users/nicolerg/anaconda2/envs/encode-atac-seq-pipeline/bin/encode_frip.py", line 54, in frip
    write_txt(frip_qc, str(float(val1)/float(val2)))
ValueError: could not convert string to float: ***** WARNING: File rat_liver7_S12_L001_R1_001.trim.merged.nodup.tn5.tagAlign.tmp1 has inconsistent naming convention for record:
AABR07024382.1	100568	100639	N	1000	+

If it would be impractical to include compatibility with masked references, it would be helpful to specify that the pipeline is incompatible with masked references in the documentation. I understand that this is the function of the blacklist input, but blacklisted regions can be more difficult to define for less popular model organisms.

[leepc12]

This is actually a duplicate issue of #48 and will be fixed in the next release.

[akundaje]

Interesting. This is a bug/unintended behavior so we’d like to get to the bottom of it. I’ll ask @leepc12 to get in touch with you so we can get to the bottom of it and fix the issue.

On Sat, Nov 3, 2018, 12:05 PM nicolerg <[e-mail] wrote:

Yes, I have been able to run the pipeline with the UCSC soft-masked version of the rat genome (including contigs). This run was with a hard-masked version from Ensembl to see what the results looked like in comparison since there are some questions about the quality of the rat assembly, but it is not entirely necessary to use the hard-masked version. In fact it may be too limiting.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ENCODE-DCC/atac-seq-pipeline/issues/49#issuecomment-435613032, or mute the thread https://github.com/notifications/unsubscribe-auth/AAI7EWVUrI6RX1X9aH0HExgw2fgnEWNLks5urej_gaJpZM4YM5wz .