auto-detect quality encoding

[macmanes]

I'm working with some old GA2 reads, get this error, which I suppose has to do with differences in Phred encoding. Would be awesome if Lighter could auto detect or at least allow me to specific Phred offset.

[2014-11-17 23:50:15] Bad quality threshold is "-"

[mourisl]

This is not an error message, it says the ascii symbol of the quality score that Lighter chooses as a threshold between good and bad quality. Lighter detects this quality score automatically by building a histogram of all the possible quality scores(described in the paper), so I think GA2 reads should be fine for Lighter.
But I should add user-specific quality score threshold.
Thanks.

[macmanes]

Thanks for the clarification! Yes, giving the user the ability to set the 'trusted' quality threshold might be useful, but probably could be an easily abused feature.

[bstamps]

Similar problem- I'm getting a segmentation fault when trying to run lighter. Is this an issue with lighter not detecting the correct phred offset, or an install issue?

lab@home:~/$ lighter -r R1.fastq -r R2.fastq -k 17 30000000 0.1 -od LighterOUT -t 4
[2014-11-29 14:04:36] =============Start====================
[2014-11-29 14:04:39] Bad quality threshold is "'"
Segmentation fault (core dumped)

Running a HiSeq instead of MiSeq dataset produces a similar result, but the Bad quality threshold error is "."

[mourisl]

I don't think the quality score is the culprit of the segfault.
I guess the reason is that you did not create the LighterOUT? I just uploaded a new version of Lighter that creates the directory specified in -od if it does not exists. Or you can create the directory first and then run Lighter.
If this is not the reason, could you please share the first few lines of R1.fastq?

Thanks

[bstamps]

Updated and running smoothly. Thank you.