Step 3 fails for FASTQ+64 formatted data

[isaacovercast]

FASTQ+64 files apparently don't work anymore:

Fails during step 3 with:

Fatal error: FASTQ quality value (70) above qmax (41)
By default, quality values range from 0 to 41.
To allow higher quality values, please use the option --fastq_qmax 70

Not sure when this stopped working but it could have been a while ago. I thought that cutadapt used to modify the quality score values based on the Phred_qscore_offset parameter, but it definitely isn't, and then vsearch is complaining downstream.

Related: https://github.com/torognes/vsearch/issues/194

The fix is to set --fastq_qmax and --fastq_qmin based on the offset value, and it'll have to be done for all the vsearch calls in step 3, I think.

[eaton-lab]

Don't we usually hand in derep'd data to vsearch where the scores are gone..., maybe this is only coming up in data sets that are doing pair-merging?

[isaacovercast]

Confirmed. This only happens for paired data. I tested with the sim data and SE with fasq+64 qscores works fine. It dies during pair-merging, good call.

On Wed, Nov 6, 2019 at 12:36 AM Eaton Lab notifications@github.com wrote:

Don't we usually hand in derep'd data to vsearch where the scores are gone..., maybe this is only coming up in data sets that are doing pair-merging?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/dereneaton/ipyrad/issues/368?email_source=notifications&email_token=ABNSXP5OC42YVXWZ6XHY5YTQSH7PTA5CNFSM4JIUQSNKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDEXCUI#issuecomment-550072657, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABNSXP6G3FUGKAYVDZAS5F3QSH7PTANCNFSM4JIUQSNA .

[eaton-lab]

Another person just ran into this bug. I'm gonna set the merge qmax to 100 and see if it fixes .

[isaacovercast]

I did this. Testing the fix in v.0.9.49.

[isaacovercast]

This is fixed. qmax is set to 93 in merge_pairs_with_vsearch to allow fastq+64 formatted data.