Filtering unmapped reads from BAM files

Question

[Manuel Sokolov Ravasqueira]

I am applying the GATK best practices for variant calling pipeline.

By applying the picard addOrReplaceReadGroups I got the error: "invalid_mapping_quality", I read the post Final Solution For "Mapq Should Be 0 For Unmapped Read.". So decided to remove unmapped reads with samtools view -bF 4.

My question is since I will apply the GATK base recalibrator, is it better to set QC = 0 to the reads that are unmapped and how would I do this? Or removing them is the best practice?

Best Regards, Manuel

[Pierre Lindenbaum]

I'm not sure "Mapq Should Be 0 For Unmapped Read." will be a problem for gatk base recal as it will just ignore unampped reads.

https://github.com/broadinstitute/gatk/blob/master/src/main/java/org/broadinstitute/hellbender/engine/ReadWalker.java#L131C46-L131C66

valid start is required:

https://github.com/broadinstitute/gatk/blob/master/src/main/java/org/broadinstitute/hellbender/engine/filters/WellformedReadFilter.java#L65

[Manuel Sokolov Ravasqueira]

Thanks! Is there any flag for ignoring this error with picard addOrReplaceReadGroups? And for example for estimating TMB from BAM files would it be a problem?

[Pierre Lindenbaum]

try

VALIDATION_STRINGENCY=LENIENT

---

[Manuel Sokolov Ravasqueira]

I did this:

srun /mnt/beegfs/apptainer/images/picard_latest.sif java -XX:ParallelGCThreads=40 -jar 
/usr/picard/picard.jar AddOrReplaceReadGroups  -I "$SORTED_BAM" -O "$OUTPUT" -RGID 4 - 
RGLB lib1 -RGPL ILLUMINA -RGPU unit1 -RGSM 20 -VALIDATION_STRINGENCY "LENIENT"

It generated the output file but still showed error. I proceded with doing the next step picard MarkDuplicates with the same flag "-VALIDATION_STRINGENCY ", error was displayed again yet it generated output? Should I keep adding this flag for the rest of the pipeline whenever use picard?