Closed dwthomas closed 8 months ago
Thanks @thermokarst
Thanks @dwthomas. So to make sure I'm following, would an alternative solution be to instead filter all empty samples from the final feature table? If so, why didn't you take that approach here?
I think the two behaviors that make sense are:
The current behavior retains some but not all zero frequency samples, which I think is a potential source of confusion. I implemented the first because that's the behavior I want. But my argument for that behavior as the default is that you can go from that to retaining only non-zero samples with a qiime feature-table filter-samples command, but I do not know a way to add back removed samples without using the Artifact API.
Awesome, thanks for explaining, that makes sense to me! Okay yeah, I'm leaning towards going with the functionality proposed here. In the future I think it would be great to add in a new flag that lets us toggle the empty-sample retention, wholesale, but I agree, that is a separate issue. I'll think about tests and get back to you. Would you mind if I push the tests up to your feature branch?
That works great for me.
@thermokarst I'm doing some work on the unit tests, I figured out where the data files are.
Hey @thermokarst I added unit tests and a toggle as discussed. All the functionality is in the _denoise_helper function used by all single, paired and pyro so I modified the paired tests to include an empty sample and test both modes of the retain-all-samples toggle.
@dwthomas - there is one minor test failure - can you please resolve that when you get a chance? Thanks!
Looks like there is just a new dada2 flavor that needed the toggle added to it as well.
@dwthomas are you still working on this PR? No pressure, just taking stock of the things we have in review for the coming release.
Hi @dwthomas! We are currently cleaning out our forum backlog! I fixed conflicts but it looks like there are still some errors! Do you have the bandwidth to take this pull request across the finish line? No pressure if not!
Hey @cherman2, I unfortunately don't have the bandwidth at least for the time being.
No problem! All these remaining issues look super superficial (and I may have caused them when I fixed conflicts). I will get this passing and then pull down to test locally!
Fixes issue #136