Creating a PON in GATK

Question

[zizigolu]

Hi

I am trying to create a panel of normal in Mutect2 but I am getting error

[fi1d18@cyan01 ~]$ gatk --java-options "-Xmx24G"  Mutect2 -R ./hs37d5.fa -I /scratch/fi1d18/example_results/1631_WTSI-COLO_075_b/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_b.dupmarked.bam -tumor-sample /scratch/fi1d18/example_results/1631_WTSI-COLO_075_1pre/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_1pre.dupmarked.bam -O normal1_for_pon.vcf.gz


A USER ERROR has occurred: Bad input: Sample /scratch/fi1d18/example_results/1631_WTSI-COLO_075_1pre/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_1pre.dupmarked.bam is not in BAM header: [WTSI-COLO_075_b]

You know how to solve this error?

[Bastien Hervé]

What is your version of GATK ?

Seems like -tumor-sample is not a valid option

[zizigolu]

I guess version GATK/4.0

[Bastien Hervé]

Well this is the command line to create PON on GATK4

gatk Mutect2 -R ~/Documents/ref/hg38/Homo_sapiens_assembly38.fasta -I HG00190.bam -tumor HG00190 --disable-read-filter MateOnSameContigOrNoMappedMateReadFilter -L chr17plus.interval_list -O 3_HG00190.vcf.gz

-tumor-sample is not an option there

[zizigolu]

Sorry, what happened if simply I skip creating PON and call somatic variants by ?

[fi1d18@cyan01 ~]$ java -jar /home/local/software/GATK/3.7/source/GenomeAnalysisTK.jar -T MuTect2 -R /temp/hgig/fi1d18/hs37d5.fa -I:tumor /scratch/fi1d18/example_results/1631_WTSI-COLO_075_1pre/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_1pre.dupmarked.bam -I:normal /scratch/fi1d18/example_results/1631_WTSI-COLO_075_b/mapped_sample/HUMAN_1000Genomes_hs37d5_genomic_WTSI-COLO_075_b.dupmarked.bam -o out.vcf

[Bastien Hervé]

If you have multiples normal samples (meaning healthy tissues) it is good practise to create PON from those normals. The panel of normal will catch technical artifacts to improve your variant calling

You have to run Mutect2 tumor-only mode on every normal sample to create one VCF per normal sample. Then use CreateSomaticPanelOfNormals with the list of VCFs to create your PON

Sure you can skip this step but your results will be less significant with a lot of false positive