CNLOH from ~start of chr1 to chr1:23272628

[jzook]

Start with copy number plot from wakhan for chr1: wakhan_cna_loh_analysis_HG008_hifi/variation_plots/chr1_cn.html, and see that it generally supports a CN=2 for one hap and CN=0 for the other hap in the region in the wakhan CN bed (chr1 50001 23300000) and the DRAGEN CN vcf (chr1 818022 with END=16526823 plus the 3rd line: chr1 16949283 END=23272950).
However, there are some nuances here:

1. the wakhan plot has a significant decrease in coverage for the first ~4Mbp down to CN ~ 1.5. This seems likely to be a mosaic deletion, but it is not called by dragen or wakhan or bionano. I was hoping the severus, sniffles, dragen, or bionano SV vcfs might have an SV breakpoint ~4Mbp, but none did except for a questionable sniffles call in a VNTR that isn't related to a drop in copy number. Maybe the first Mbp of chr1 was lost in some cells without causing an SV breakpoint, but this'll take single cell data or more sophisticated analyses to detect. For now, we should just make note of this, I think
2. wakhan starts and 50kb and dragen starts at 800kb. these are in segdup regions where it's hard to know but I don't see any clear evidence of breakpoints in these regions
3. dragen calls a lower quality DUP instead of DEL:DUP/CNLOH at chr1 16605768 with END=16949283. This is a segdup region with >2 haplotypes in the normal, so I suspect this is an error, but this is likely where mapping to the HG008-N assembly rather than GRCh38 would help to know for sure if something more complicated might be happening. For now, I'd lean towards treating this call as an error, though we could note this as an uncertain region
4. To determine the precise end position, it's often useful to look for SV breakpoints, in this case ~23.2-23.3M. Severus, sniffles, dragen, and bionano all have BND calls with chr5 in this region. Generally sequencing based SV calls will be more precise, and severus and dragen both have the call at 23272628, and sniffles has it one bp offset at 23272629. Looking in IGV, the last base before the breakpoint is 23272628, and since vcf prefer to pad 1bp, I think this follows vcf convention, though should check the vcf spec fo how to define BND calls.
5. It's probably worth noting that this CNLOH or DEL:DUP event is likely caused by the deletion of the p arm of chr1, followed by the duplication of the first 23272628bp, which joined with chr5. Except for the first 4Mbp, this seems to be fixed in all or practically all cells based on the CN~2.0 and that it seems to be the same in practically all bioskryb cells