Working with input files with Q64 and Q33 (GATK IndelRealigner)

[woodydon]

Dear SEQanswers members,

I have two input fasta files from exome-seq. One is coded with Q64 and the other is coded with Q33 quality scores. I want to combine the two input fasta files and run bwa+GATK.

How do I combine them for IndelRealigner? I suppose that IndelRealigner needs all reads from both Q64 and Q33. Can I do IndelRealigner separately and then join them? Will this cause problems?

I have searched for many posts but can't find my answers. Please help me.

Thanks,
Woody

[Brian Bushnell]

Once they're in sam/bam format the original quality values get changed to ASCII-33 anyway. Just map them separately then combine the sam files.

[elkjournals1]

Please elaborate this issue and then i can see and Brian Bushnell is right please do one time this "Just map them separately then combine the sam files."............Best Of Luck

[woodydon]

Once they're in sam/bam format the original quality values get changed to ASCII-33 anyway. Just map them separately then combine the sam files.

Hi Brian,

Thanks for the response!

I combined the fastq files and ran BWA-MEM to generate a bam file. I used GATK's RealignerTargetCreator to create target regions for IndelRealigner. But, before IndelRealigner, I added --fix_misencoded_quality_scores to the RealignerTargetCreator step. GATK returned an error message that it can't handle a mixture of records with Q64 and Q33. Are you suggesting that I don't need to fix the quality scores?

It confused me.

Bests,
Woody

[woodydon]

Hi Brain,

I am sorry that I misunderstood your response. I will map them separately and merge the bam files for the following GATK steps. I will report if I succeed.

Thanks,
Woody

[woodydon]

Hi Brian and elkjournals1,

I found the solution for this after consulting with a GATK developer. Please refer to this post:

http://gatkforums.broadinstitute.org...aligner#latest

Bests,
Woody

[Brian Bushnell]

Thanks for posting that link. But, I disagree with the advice there. Let me clarify:

You should align the datasets separately, then merge the sam files. But when aligning the ASCII-64 files with bwa, you need to use the -I flag to tell it the qualities are in ASCII-64, since bwa does not autodetect qualities (as far as I know). Without that flag, the output will be incorrect, and simply fixing the quality values afterward will not solve it since the mapping will have been done according to incorrect quality values.

As long as the mapper is aware of the input quality format, it will convert it to ASCII-33 in the sam output regardless of the input, as that is required by the sam specification and is not something that varies by mapper.