enh: filter tags

[y9c]

Will it be useful to add filter for tags, such as "[NH]>1" to filter reads of multiple hits? Will the same grammar as --input-fmt-option of samtools help?

Thank you.

[brentp]

I like this idea. It would be lua syntax, like read.NH > 1 or more likely tag.NH > 1.

[brentp]

is this or any other issue blocking for you? I probably can't implement everything right now, but would like to get the top priorities started.

[y9c]

Thank you @brentp, filtering by tags should be one of the prior features I wish to have.

Another two major things are:

• The performance, especially for locus with extremely high sequence depth (eg, rRNA sites). I notice that the pileup would become extremely slow. If I remember clearly, the rust htslib toke more than one hour to pileup sites on a rRNA gene (~10kb). I am not sure if this problem can be solved without replacing the htslib?
• filter mutation. It is not hard to filter the number of mutations based on the "ref_base" column using some "if-else" statements. But it would be great if pbr can support this one out of box.

[y9c]

Hi @brentp, thank you for working on this. I would like to known if filtering by tags is supported now?

[brentp]

No, for now, I have put this project on hold. I think there is too much overhead in calling a lua expression for each base. Maybe it's possible with using some kind of macro based expressions within rust, but I don't have plans to implement further. I found that using pysam with custom code is faster than pbr.