MuTect2 panel of normals – empty VCF file
I have successfully run MuTect2 on many paired tumor/normal samples. However, when I try creating a panel of normals (using the the recommended approach), I end up with empty VCF files.
Here is the tail of the output when running MuTect2 on a normal sample (restricted to chr7 for testing purposes):
INFO 15:02:19,928 ProgressMeter - chr7:159138663 0.0 4.8 h 15250.3 w 100.0% 4.8 h 0.0 s INFO 15:02:29,476 MuTect2 - Ran local assembly on 0 active regions INFO 15:02:29,488 ProgressMeter - done 1.59138663E8 4.8 h 107.0 s 100.0% 4.8 h 0.0 s INFO 15:02:29,490 ProgressMeter - Total runtime 17114.21 secs, 285.24 min, 4.75 hours INFO 15:02:29,492 MicroScheduler - 1997879 reads were filtered out during the traversal out of approximately 6720185 total reads (29.73%) INFO 15:02:29,494 MicroScheduler - -> 0 reads (0.00% of total) failing BadCigarFilter INFO 15:02:29,495 MicroScheduler - -> 1947653 reads (28.98% of total) failing DuplicateReadFilter INFO 15:02:29,497 MicroScheduler - -> 0 reads (0.00% of total) failing FailsVendorQualityCheckFilter INFO 15:02:29,498 MicroScheduler - -> 0 reads (0.00% of total) failing MalformedReadFilter INFO 15:02:29,500 MicroScheduler - -> 0 reads (0.00% of total) failing MappingQualityUnavailableFilter INFO 15:02:29,501 MicroScheduler - -> 19931 reads (0.30% of total) failing NotPrimaryAlignmentFilter INFO 15:02:29,503 MicroScheduler - -> 30295 reads (0.45% of total) failing UnmappedReadFilter
So MuTect2 is actually doing something (but I am not sure what it is). I think the Ran local assembly on 0 active regions part looks rather suspicious.
For completeness, here is the command executed:
java -Xmx16g -Djava.io.tmpdir=tmp -jar GenomeAnalysisTK.jar \ -T MuTect2 \ -nct 16 \ --reference_sequence ucsc.hg19.fasta \ --dbsnp dbsnp_138.hg19.vcf \ --cosmic hg19_cosmic_v54_120711.vcf \ --input_file:tumor test_sample.bam \ --artifact_detection_mode \ -L chr7 \ --out test_sample.vcf \ -et NO_ET \ -K my_key.key
Any ideas what could be wrong?