Structural Variants (SV) Manta output interpretation

Question

[aLex97]

Hi everyone,

I ran variant calling using Manta (the SV caller) with both my tumor and normal BAM files. I checked the results in somaticSV.vcf.gz. How should I interpret these results (particularly the PR:SR)? Did Manta produce output that only shows the differences between the normal and tumor BAM files, or is it related to the reference genome?

Here’s an example of the rows I got:

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  TUMOR _SAMPLE   NORMAL_SAMPLE
chr1    6032123 MantaDEL:535:0:1:0:0:0  T       <DEL>   .       PASS  END=6065447;SVTYPE=DEL;SVLEN=-33324;CIPOS=0,2;CIEND=0,2;HOMLEN=2;HOMSEQ=GC;SOMATIC;SOMATICSCORE=130     PR:SR   60,0:92,0       46,17:59,21*

If you have any recommendations for other tools to use for detecting large structural variants, please let me know

Thank you in advance.

Answer 1

[DGTool]

What might help is on Illumina's/Manta's GitHub, there is a section that tries to explain what some parts of the VCF output means: https://github.com/Illumina/manta/blob/master/docs/userGuide/README.md#manta-vcf-interpretation. Specifically regarding PS:SR:

PR  Number of spanning read pairs which strongly (Q30) support the REF or ALT alleles
SR  Number of split-reads which strongly (Q30) support the REF or ALT alleles

Regarding the second question, to my understanding, in somatic variant calling, if you have a variant that passed all the filters etc would mean that:

A) It varies against the reference (i.e. GRCh38)

B) It is not present in the normal sample (if it was, then it would be considered germline)

[aLex97]

Thank you :)

I've read on github I’m not sure how to make sense of this part:

FORMAT TUMOR _SAMPLE NORMAL_SAMPLE

PR:SR 60,0:92,0 46,17:59,21*

What is considered a good SOMATICSCORE in this context?

[DGTool]

Ah, to that I have sadly no idea (I haven't really used Manta much before). There seems to be some additional information on how they score the somatic variants (which I'm guessing is SOMATICSCORE) in the supplementary zip file on their publication (here). It should be in the file named manta_supplementary_text.pdf.