Identifying Long Reads Spanning Chromosomal Breakpoints

Question

[Noah]

Hey all,

Currently I'm doing battle with some ONT Long-Read WGS data, and trying to get my feet wet with Structural Variation analysis. Right now I want to extract reads from my alignment that support a breakpoint's existence, since the usual tools have not produced satisfactory results.

My question is: if you were to fish for reads that span a chromosomal breakpoint, would you use reads that have supplementary alignments (i.e., the SA tag), or reads that have been softclipped a considerable amount, or both? I'm taking into consideration all of these subsets, but wanted to hear from the community if there is some established way to do this.

[GenoMax]

Have you tried sniffles2 to do structural variant calling for long read data: https://github.com/fritzsedlazeck/Sniffles

If you want to identify all types of variations at the same time then using Epi2ME workflow may be one step solution : https://github.com/epi2me-labs/wf-human-variation

If you have tried some of these tools then it would be best to list what you have already tried.

Answer 1

[trausch]

I would use both, supplementary alignments and soft-clipped reads. With delly's long-read mode you can dump all SV-supporting reads to a file:

delly lr -d dump.gz -g genome.fa input.bam