MuTect --- with more stringent parameter settings I got more mutations -- Is this possible

Question

[haiying.kong]

I ran MuTect twice, once with default parameter settings, and the second time with stringent parameter settings.

default:

/usr/bin/java -Xmx4g -Djava.io.tmpdir=tmp -jar ${MuTect}mutect-1.1.7.jar \
--input_file:normal ${BQSR_dir}${Blood}.recal.bam --input_file:tumor ${BQSR_dir}${Nevus}.recal.bam \
--out ${MuTect_dir}${Blood}${Nevus}_call_stats.out \
--coverage_file ${MuTect_dir}${Blood}${Nevus}coverage.wig.txt \
--vcf ${MuTect_dir}${Blood}${Nevus}.vcf \
--analysis_type MuTect --reference_sequence ${GATK_hg} --cosmic ${COSMIC} --dbsnp ${dbSNP} --intervals ${Intervals} \
--disable_auto_index_creation_and_locking_when_reading_rods

stringent:

/usr/bin/java -Xmx4g -Djava.io.tmpdir=tmp -jar ${MuTect}mutect-1.1.7.jar \
--input_file:normal ${BQSR_dir}${Blood}.recal.bam --input_file:tumor ${BQSR_dir}${Nevus}.recal.bam \
--out ${MuTect_dir}${Blood}${Nevus}_call_stats.out \
--coverage_file ${MuTect_dir}${Blood}${Nevus}coverage.wig.txt \
--vcf ${MuTect_dir}${Blood}${Nevus}.vcf \
--analysis_type MuTect --reference_sequence ${GATK_hg} --cosmic ${COSMIC} --dbsnp ${dbSNP} --intervals ${Intervals} \
--disable_auto_index_creation_and_locking_when_reading_rods \
--min_qscore 30 --max_alt_alleles_in_normal_count 0

Only the last line is added in the stringent running.

To my surprise, I got about 30 more mutations on the second running. Total 9 pairs of samples. It this possible?

This probably possible for small number of mutations, but we are doing melanoma, with large number of mutations (over 2000 from 9 patient). Can this still happen? Or does this indicate any mistake I made?

In both runs, I used MuTect-1.1.7.

The data is whole exome sequence data from matched tumor and normal samples.

Please let me know if I need to provide more information.

[igor]

Are you counting both "PASS" and "REJECT" mutations?

[haiying.kong]

No, only PASS. I think it is possible. I checked all IGVs, and found that if I set phred score threshold as 30, and a lot of mutated alleles in normal are ignored, and some locations can be identified as somatic mutations. I am saying some somatic mutations identified with stringent criteria are not identified in default, because in default, some alleles in normal are considered that are ignored in stringent.

[Amitm]

hi haiying.kong, Not directly related to your query, but I was wondering how much was the coverage for the tumor-normal pairs. I am asking as it would be relevant to know if this (what you are observing above) happens at very high/ low depth samples.

Also, it could be helpful if you look into the STDOUT of MuTect (where it spits out more details for each call: PASS or REJECT) and check if those extra 30 variants were called in the default settings as well (but were REJECTed). If yes, then you could look in the reason assigned for rejection. Its the second last col. in STDOUT (failure_reasons)