Since I use paired-end sequencing, insert size is the right way to get the fragment length. But for single-read sequencing it might not be? What's the best approach to handling this?

Hi @LudvigOlsen , nice job figuring this out. Indeed this looks like you've handled it cleanly and correctly.

isize is actually fragment length, so I think this should work as you have it for single end.

as to the PR, and the general reason I'm hesitant to add features. This feature makes sense, and now you've implemented it, but now I need to maintain, test, and document this. And it adds extra load to the user to have 2 additional parameters that will very rarely be used. That load translates into additional questions.
So, that's my concern with this.

Let me think about what to do here.

Hi @brentp

Good to hear! I think I've seen template lengths of 0 before, so I was worried you might get an insert size of 0 in single-read data.

I understand it's more complex than just the implementation I've suggested. And I really appreciate your work on mosdepth. It's such a great tool!

My colleagues and I, across multiple research groups, tend to need this functionality. And the current approach is to subset large bam files for every experiment, which is a bit of a hassle. mosdepth is amazing and this is the main functionality that I'm missing. For context, I work on cancer detection via whole genome sequenced cell-free DNA. Here, it's common to look at ranges of fragment lengths, as some tumor fragments have been found to be smaller. And in my work specifically, I wish to only look at the common range of fragment sizes across multiple datasets, in order to increase generalization across the datasets.

If I can help with the testing and documentation of these features, do let me know. I'm very new to nim, but do have extensive experience with unit/regression testing.

@LudvigOlsen, it could be a very useful feature. Thank you!

@LudvigOlsen would you write a test for this? you can add to functional-tests.sh there should be an error when max-frag-len <= min-frag-len and a test for that. Also a test that runs successfully with those options.
And I think --min-frag-len and --max-frag-len, what do you think?

Hi @brentp

Sure! I added the error and updated the argument names.

If you want fragments with a specific fragment length, I guess min and max should be allowed to be equal?

I will look at the testing soon

If you want fragments with a specific fragment length, I guess min and max should be allowed to be equal?

Ah, yes, that's right. For testing, you can just use one of the current test bams and manually find, for example an exact fragment length, then add that as the test.

thank you!

@brentp I have added some tests. First time with shell-based tests but I think, I figured it out. Perhaps try running them on your side as well though. Any more tests you want added? :-)

Thanks very much @LudvigOlsen! I'll get out a new release next week.