Is possible tumor-only samples?

[byeongill]

Hi
My data are WGS and not contain any normal samples, consist of tumor and pre-tumor tissues.
For variants calling, used GATK-haplotypecaller.
I understood UNMASC need UMN (pools of unmatched normals).

Is it possible to use UNMASC for my data?

thanks !

[pllittle]

Hi byeongill,

Thank you for your interest in UNMASC. Do you have any previously sequenced normal WGS samples that underwent similar sequencing protocols and bioinformatic workflows as your tumors? And may I ask are your pre-tumor tissues considered normal? If yes to either question, I believe you could use UNMASC if your variant caller requires both tumor AND normal bams. Having a set of normal read counts paired to each tumor variant call for UNMASC helps screen and classify tumor variants as somatic, germline, artifact.

Hope this helps!

[byeongill]

Thanks,

Although the pre-tumor samples are not normal, i will try it as normal (6 samples out of 86 samples).
Variant calling were performed GATK-haplotypecaller (Not use joint calling)

The main function of "run_UNMASC" need vcf as data.frame which consist of Qscore, nRD, nAD, tRD, tAD etc.
I can know nRD and nAD in paired tumor samples through matching to variants of pre-tumors.
How do i know nRD and nAD in tumor-only samples ?

[pllittle]

Hi @byeongill,

Thank you for your questions.

For the input data.frame, because we're performing tumor-only variant calling meaning the matched normal is unavailable so UNMASC replaces the matched normal with multiple unmatched normals. I looked over GATK-haplotypecaller and it wouldn't be suitable for UNMASC. Whichever somatic variant caller you use (Strelka, MuTect, etc.) should accept a tumor.bam and normal.bam. As a result, each locus should be characterized by

• tRD/tAD (tumor read count for a position) and
• nRD/nAD (normal read count for the same position)

Hope this helps! :)