GATK licensing moves to direct-through-Broad model -- read about it on the GATK blog

DepthOfCoverage producing "extremely high quality score" error

igorigor Posts: 31Member
edited April 2013 in Ask the GATK team

I am running GATK DepthOfCoverage one a bunch of samples (sequenced using Illumina MiSeq). For one of the samples, I am getting the following error:

ERROR MESSAGE: SAM/BAM file SAMFileReader{<file.bam>} appears to be using the wrong encoding for quality scores: we encountered an extremely high quality score of 61; please see the GATK --help documentation for options related to this error

I found a suggestion that I should be using fixMisencodedQuals flag in this case. However, if I do that, "the engine will simply subtract 31 from every quality score as it is read in". That will fix this error, but then all my quality scores will be incorrect.

Furthermore, the BAM file I am using was last recalibrated with GATK. Why is GATK producing files with invalid quality scores?

What should I do?

Any help would be much appreciated.

Post edited by igor on
Tagged:

Best Answer

Answers

  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 7,530Administrator, GATK Developer admin

    Hi Igor,

    It is unlikely that the GATK "produced" the bad quals. Most likely, Base Recalibrator just took the quals in your data as they were, and recalibrated (adjusted them upwards or downwards) as appropriate.

    Ideally you should check your original fastq files and see what the quals encoding is in there. If they're in the old (deprecated) encoding, then you can simply use the fixMisencodedQuals flag -- the resulting quals will all be correct. The new encoding is just like the old one but offset by 31. If not, let me know and I will try to help you figure out what's happening.

    Geraldine Van der Auwera, PhD

  • igorigor Posts: 31Member

    These are definitely the new encodings. I've never gotten this error before and it occurs for only one of the files in the batch (they were all processed the same way at the same time).

    With a different set of intervals, there is no error, so I think this might only be happening at very few positions for some reason.

  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 7,530Administrator, GATK Developer admin

    Actually, let me qualify that -- newer versions of GATK shouldn't be producing bad quals, but some of the older ones did, so if your files were processed with an older version, then it is possible that it was the GATK's fault.

    For running DoC the quals don't matter so you can safely ignore the error message. In the docs you will find a flag to override the error.

    Geraldine Van der Auwera, PhD

  • igorigor Posts: 31Member
    edited April 2013

    I am using GATK 2.3-6 for both recalibration and DoC. I am surprised if the qualities are problematic prior to realignment or recalibration, why didn't GATK complain then (which is why I assumed the faulty qualities were introduced by GATK)?

    Which flag should I be using? Would it affect minimum base quality setting?

    Post edited by igor on
  • blueskypyblueskypy Posts: 229Member

    I got the same error while using RealignerTargetCreator in GATK v2.5-2. However, fastqc output shows that the encoding of my bam file is Illumina 1.5 and the quality score is from 0 to 40.

  • CarneiroCarneiro Posts: 274Administrator, GATK Developer admin

    can you take a look at a read from your bam file to see the qual encoding? It may be a bug on fastqc.

  • blueskypyblueskypy Posts: 229Member

    Here is a read from the fastq file, does it help in resolving that error?

                    @A807HEABXX:3:1:1345:2117#TTAGGCAT/1
                    GTTGCTGTCTTCCTGCTTGCATTCTGAGTCTGGCATCCTTTCTGTTCCTGGGCTTAACCATGCTTTTCCTGCCTTCAGACTTTTGTATCA
                    +
                    ggfgggggeefggggggdgcgedfdfafbdegefeeggggegggccggg\gdee`\acbcdaadaeebc^MKRYYc^dbcZccccO\]]a
    
  • CarneiroCarneiro Posts: 274Administrator, GATK Developer admin

    these are illumina new encodings. Probably a bug with FastQC.

  • blueskypyblueskypy Posts: 229Member

    hi, Mauricio, thanks for the reply! fastqc says the encoding is illumina 1.5, do you mean that is wrong? are those quality scores "extremely high"? Thanks!

  • CarneiroCarneiro Posts: 274Administrator, GATK Developer admin

    If I remember correctly 1.5 is the letter encoded quality scores (starting at 59). They have reverted back to the sanger standard after 1.8. So yeah, you need to reencode your quality scores before using this file with the GATK.

  • CarneiroCarneiro Posts: 274Administrator, GATK Developer admin

    the flag is -allowPotentiallyMisencodedQuals

Sign In or Register to comment.