secondary collapse fails on down/sub-sampled fastq

[sand60]

Hi Daniel,
I am unable to run collapse on subset of downsampled fastq, it fails without any useful message. I used seqtk to downsample. All other tools can use those fastq without any errors,.
Thanks
Sukhinder

[dnbaker]

seqtk probably mangled the fastq and/or metadata somehow. Can you post a small sample of your seqtk-processed fastq?

[sand60]

attached is sample fastq with 5000 reads,.thanks
seqtk_sample_5k_R1.fastq.gz

[dnbaker]

1. Your reads are different lengths. BMFtools requires uniform read length.
2. seqtk has mangled the names such that there is no comment field, which might cause some problems.

The first is probably your real problem.

[sand60]

1 has usually worked for me,.and since the only difference during failing was seqtk. I think it is 2nd. Is that comment filed is mandatory for bmftools? I can try adding it back,.because I can see that is the only difference from my original fastq which worked. Thanks

[sand60]

and that "Description" field is optional for fastq format,.so not sure why bmf must want it,.

[dnbaker]

What release are you using? Something related to this issue was fixed recently. #98 (comment)

[sand60]

I have this, I guess I should re-pull?
commit 7279788
Merge: b40d9d6 cca362d
Author: Brett Kennedy [email protected]
Date: Wed Feb 22 15:33:22 2017 -0700
Merge pull request #95 from dnbh/master
Bug fixes, portability.

[dnbaker]

If you update to master, it should handle problem 2. And you shouldn't be working with varying read lengths as that assumption is made throughout the code base, but if you think it's working, then you can do what you want.

[sand60]

Should this fix it? Do I need to do anything different to implement the change? Thanks

commit f9483ea
Merge: 7279788 7a6a3e4
Author: Brett Kennedy [email protected]
Date: Mon May 15 11:33:59 2017 -0600
Merge pull request #99 from dnbh/master
Fix issue 98

[dnbaker]

Yes, that commit does have the fix. (See

BMFtools/lib/mseq.cpp

Line 33 in f9483ea

std::strcpy(ret->comment, seq->comment.s ? seq->comment.s: "");

).