how to choose an efficient filtering threshold to remove shorter and low quality reads before doing alignment of Nanopore long reads

Question

[Shaghayegh]

I just started processing ONT long reads which obtained from soft tissue sarcomas. I have a very general question, what would be an ideal filtering threshold for read length and read quality before performing alignment (I am will be using minimap2) on Nanopore long reads? I have seen SV caller, sniffles, looks at alignment quality when identifying structural variation, it is important we remove reads with low-quality patches. But I am not sure how to choose the threshold. Any idea?

Answer 1

[colindaven]

You haven't included enough information to really help you, and you haven't included any read length statistics. I would definitely exclude very short reads though (some are under 100bp in my experience).

I would use multiple SV callers after alignment, eg cuteSV, and not just Sniffles2.

One flexible read length filtering tool is https://github.com/rrwick/Filtlong