Feature request: Set no/low coverage regions of short reads to N

[gwl2]

Feature request An option that when polishing an assembly with short reads, all bases of an assembly that are not covered by at least X (user supplied parameter) reads are set to N, indicating that this part was not polished with short reads.

Why? When users polish long read assemblies with short reads, they tend to believe that the new consensus is always better than the original assembly across the whole genome. However, it may well be the case that there are parts of the assembly that were not covered by the short reads. In these regions, it's not known whether there are (short read) resolvable errors or not. So if you need single nucleotide accuracy in this region, you should be aware of this fact and be extra careful.

[rrwick]

That's a good point, so thanks for that suggestion.

I don't think I'll build this feature into Polypolish itself, but it could be implemented in a separate script using Polypolish's debug output table. That table shows the depth (using Polypolish's fractional depth for multi-mapped reads) for each position in the assembly. Also, the status column will contain low_depth if that depth is below Polypolish's --min_depth threshold.

So a script could be made to:

• Read in the assembly
• Read the debug table and note which positions are below a user-defined depth threshold
• Output the assembly with the low-depth positions masked

You're of course welcome to code this up yourself! If you'd rather I do it, I can add it to my to-do-at-some-point list, but I can't make any promises about when it will be ready :smile: