Demuxalot: Adjust script for v. 0.4.0 #46
Conversation
Demuxalot changed its API between the version currently included in the Demuxafy Singularity image (v. 0.2.0) and the latest (as of 2024-02-22) version (v. 0.4.0): Notably, the `only_singlets` parameter of the `predict_posteriors` function was dropped (see arogozhnikov/demuxalot@3939e81). However, as far as I can tell, as long as the `doublet_prior` parameter is non-zero, the it defaults to the behaviour of `only_singlets = FALSE` (cf. arogozhnikov/demuxalot@2b1535b). As Demuxafy's `Demuxalot.py` uses, `only_singlets = FALSE` and `doublet_prior` defaults to `0.35` (i.e. a non-zero value), simply dropping the (now) unsupported parameter restores compatibility of Demuxafy's `Demuxalot.py` with the current version of Demuxalot. I have verified that the adjusted version of Demuxafy's `Demuxalot.py` works with Demuxalot v. 0.4.0 by running the corresponding tutorial (with the reduced dataset) as a test case. The final (refined) assignments (as per `assignments_refined.tsv.gz`) were identical to those obtained by running the current version from the Demuxafy Singularity image.
|
@drneavin: Any chance to get this (and subsequently PR #48) merged and and into a new release? If there is anything you'd like me to improve about any of my PR, please let me know. Thank you for all the work you have put into this. PS: I just noticed that I wanted to post this on PR #47, which included (and IMHO improves upon) this PR. I am converting this one to a draft to make it harder for you to accidentally merged the 'wrong' PR and having to wait for a rebase from my side before moving forward with PR #48. |
|
Hi @mschilli87, sorry I haven't been able to address this due to other projects and limitations on time. I'll prioritize this early next week - already on my to-do list |
|
@drneavin: Thank you for the response and heads up. I did not mean to rush you. I understand how this job can be at times. I just wanted to know if there is anything I can do and if I can hope for a timely review. Early next week sounds great to me. Thank you. |
Demuxalot changed its API between the version currently included in the Demuxafy Singularity image (v. 0.2.0) and the latest (as of 2024-02-22) version (v. 0.4.0):
Notably, the
only_singletsparameter of thepredict_posteriorsfunction was dropped (seearogozhnikov/demuxalot@3939e81). However, as far as I can tell, as long as the
doublet_priorparameter is non-zero, the it defaults to the behaviour ofonly_singlets = FALSE(cf. arogozhnikov/demuxalot@2b1535b).As Demuxafy's
Demuxalot.pyuses,only_singlets = FALSEanddoublet_priordefaults to0.35(i.e. a non-zero value), simply dropping the (now) unsupported parameter restores compatibility of Demuxafy'sDemuxalot.pywith the current version of Demuxalot.I have verified that the adjusted version of Demuxafy's
Demuxalot.pyworks with Demuxalot v. 0.4.0 by running the corresponding tutorial (with the reduced dataset) as a test case.The final (refined) assignments (as per
assignments_refined.tsv.gz) were identical to those obtained by running the current version from the Demuxafy Singularity image.As usual with my PR's this one adds a final newline to the script due to my editor configuration.
@drneavin: I could not find any place in the repository that actually defines which version of Demuxalot to include in the Demuxafy Singularity image. Else I would have bumped the version accordingly.
So please ensure that upon mergin this PR you also update Demuxafy via pip.
If I just missed it, please point me to it so I can modify this PR accordingly.