MuTect2 still discard many reads, how to fix?

Question

[Sharon]

Hello everyone

MuTect2 from GATK3 discarded 87.33% of the reads. This is during preparing panel of normals. I will use sp1.vcf in the PON creation !

sp1.vcf: sp1.bam
        java -jar ${GATK}/GenomeAnalysisTK.jar \
        -T MuTect2 \
        -R ${hg38}.fasta \
        -I:tumor sp1.bam \
        --dbsnp ${DBSNP} \
        --cosmic ${COSMIC} \
        --artifact_detection_mode \
        -o sp1.vcf

Does this seem okay to you? Is there anything I can do to fix non primary reads (42.77%) and duplicate reads (44.56%)?

INFO  00:57:17,188 MicroScheduler - 158885515 reads were filtered out during the traversal out of approximately 181931389 total reads (87.33%) 
INFO  00:57:17,190 MicroScheduler -   -> 9210 reads (0.01% of total) failing BadCigarFilter 
INFO  00:57:17,192 MicroScheduler -   -> 81066454 reads (44.56% of total) failing DuplicateReadFilter 
INFO  00:57:17,193 MicroScheduler -   -> 0 reads (0.00% of total) failing FailsVendorQualityCheckFilter 
INFO  00:57:17,195 MicroScheduler -   -> 0 reads (0.00% of total) failing MalformedReadFilter 
INFO  00:57:17,197 MicroScheduler -   -> 0 reads (0.00% of total) failing MappingQualityUnavailableFilter 
INFO  00:57:17,198 MicroScheduler -   -> 77809851 reads (42.77% of total) failing NotPrimaryAlignmentFilter 
INFO  00:57:17,200 MicroScheduler -   -> 0 reads (0.00% of total) failing UnmappedReadFilter 
------------------------------------------------------------------------------------------
Done. ------------------------------------------------------------------------------------------

This is how STAR log summary looks:

            UNIQUE READS:
               Uniquely mapped reads number |       29463720
                    Uniquely mapped reads % |       72.35%
                      Average mapped length |       271.00
                   Number of splices: Total |       21158459
        Number of splices: Annotated (sjdb) |       21067563
                   Number of splices: GT/AG |       20952149
                   Number of splices: GC/AG |       126813
                   Number of splices: AT/AC |       5622
           Number of splices: Non-canonical |       73875
                  Mismatch rate per base, % |       0.48%
                     Deletion rate per base |       0.02%
                    Deletion average length |       1.27
                    Insertion rate per base |       0.01%
                   Insertion average length |       1.84
                         MULTI-MAPPING READS:
    Number of reads mapped to multiple loci |       9507878
         % of reads mapped to multiple loci |       23.35%
    Number of reads mapped to too many loci |       355263
         % of reads mapped to too many loci |       0.87%
                              UNMAPPED READS:
   % of reads unmapped: too many mismatches |       0.00%
             % of reads unmapped: too short |       2.76%
                 % of reads unmapped: other |       0.67%
                              CHIMERIC READS:
                   Number of chimeric reads |       0
                        % of chimeric reads |       0.00%

Answer 1

[dariober]

If you are sure you want to retain non primary and duplicates, I think you can add to mutect the option --disable_read_filter NotPrimaryAlignmentFilter --disable_read_filter DuplicateReadFilter (not tested). See some docs here.

[Sharon]

Thanks dariober. I don't know it I should retain or not !

[igor]

If you don't know, then you should probably not retain. There is a reason why they are not retained by default. You can consult GATK Best Practices for more info: https://software.broadinstitute.org/gatk/best-practices/workflow?id=11146