collapse_isoforms_by_sam.py runtime error about cdna_cupcake

[thelingxichen]

I changed {1:.3f} to {1} in cupcake/tofu/branch/branch_simple2.py, and problem solved. Plus I am using minimap2 aligned bam.

[Magdoll]

@paprikachan ,

Even if that solves the problem, I think it's possible some records in your SAM file is not correct. This error happens when the SAM file is not sorted or is corrupted/incomplete.

--Liz

[thelingxichen]

@Magdoll I checked the Sam file, it's sorted but there are some unmapped reads at the end of file. So, the unmapped reads should be filtered out, right?

[thelingxichen]

@Magdoll
Hi, I have removed the ummapped reads and successfully completed the collapse and count steps with no error message.

While, when I run filter command:

filter_away_subset.py $prefix

Error occurs:

The number of PBID records in the files disagree! Sanity check failed.
# of PBIDs in /home/chenlingxi/mnt/chenlingxi/workspace/FS_Projects/WHOC.Sequel_PB.batch01/Isoform/cupcake_minimap2/OC001T/OC001T.hq_isoforms.collapsed.rep.fq: 14779
# of PBIDs in /home/chenlingxi/mnt/chenlingxi/workspace/FS_Projects/WHOC.Sequel_PB.batch01/Isoform/cupcake_minimap2/OC001T/OC001T.hq_isoforms.collapsed.gff: 14779
# of PBIDs in /home/chenlingxi/mnt/chenlingxi/workspace/FS_Projects/WHOC.Sequel_PB.batch01/Isoform/cupcake_minimap2/OC001T/OC001T.hq_isoforms.collapsed.abundance.txt: 12143

It seems there are problems in count step. I run count commands like:

get_abundance_post_collapse.py  $prefix $cluster_fn

whileI didn't append the cluster_id prefix before running count,
is this the cause of error?

Thanks a lot.

[Magdoll]

Hi @paprikachan ,

1. The unmapped reads should not have been a problem unless you didn't sort using the following

sort -k 3,3 -k 4,4n input.sam > input.sorted.sam

This sorts all unmapped reads in the beginning, not the end of the SAM file. Since you say the unmapped reads were at the end of the files, the sorting is probably the issue.

That said, since you removed the unmapped alignments from SAM file, I don't think they would affect the collapse results (because the collapse script simply ignores them as long as they are properly sorted to the beginning).

2. re: the discrepancy between the count file.

Can you check the number of lines in the group.txt file?

wc -l OC001T.hq_isoforms.group.txt

The number of lines in here should also be exactly 14779.

(side note: this sanity check is necessary, and I've now seen enough sanity check fails that I should figure out a way for you guys to not experience it, rather have the collapse or get_abundance step throw the proper warning if the outputs were not consistent.)

[thelingxichen]

Hi @Magdoll,

Thanks for your reply. For the discrepancy between the count files. I figure out why.

I use the old version of cluster, and run it with multiple bins. So I need to append bin_index into the cluster prefix in both cluster_report.csv and fastq id.