jzook / testHG008curation

This repository is primarily for the Genome in a Bottle Consortium to curate structural variants in HG008
https://www.nist.gov/programs-projects/cancer-genome-bottle
1 stars 0 forks source link

[chr3:184714600-184717315][BND][JoinedTo=chr3:184717314]] #170

Open jzook opened 1 month ago

jzook commented 1 month ago

chr3 184714600 Minda_100 N N]chr3:184717314] . PASS SVLEN=2714;SVTYPE=BND;SUPP_VEC=PB_severus_INV4929,ONT_severus_INV5695,ONT_ID_12265_2,PB_ID_53334_2,ILL_95621117:2,ILL_gridss67bb_12113h,ILL_MantaBND:73125:0:1:1:0:0:1,PB_r_263_1,ONT_r_169_1

https://v2.genomeribbon.com/?session=https://42basepairs.com/download/s3/giab-data/ribbon-json/ribbon-hg008-cov-bedpe.json&locus=chr3:184714600#ribbon

aysegokce commented 1 month ago

Reciprocal inversion (the other side) with deletions:chr3:184716952-184714600 and chr3:184716952- 184717314.
Nearby insertion with the length 164bp. Likely to be subclonal yet represented in the majority of the cells.

Screenshot 2024-09-27 at 12 23 18 PM Screenshot 2024-09-27 at 12 29 58 PM Screenshot 2024-09-27 at 12 29 49 PM
jzook commented 1 month ago

Adding the ribbon plot after selecting a read with the inversion for clarity. Because ~50% of the reads support the inversion and the deletion, this looks likely clonal to me, though there aren't any nearby germline hets so hard to know for certain without more analysis. @aysegokce why were you thinking it is subclonal?

Screenshot 2024-09-27 at 2 16 36 PM Screenshot 2024-09-27 at 2 22 55 PM
jzook commented 1 month ago

Also, do you think this would be better called as an inversion plus a deletion, rather than as 2 inversions like dragen has it?

aysegokce commented 1 month ago

Re subclonal: There are spanning reads, and the alignments look real to me. The region is phased, but there are reads in both haplotypes that support this inversion also spanning; it can be a phasing problem as well.

The definition of INVs in vcf is not clear. In the official format, INVs are defined as reciprocal inversions only (without deletion), but in the application, most tools call all inverted junctions INVs. In this example, the deletion is large, and we cannot define it with two breakpoints, so I suggest reporting both junctions using the BND notation for this example.