search

bigbio / quantms

Quantitative mass spectrometry workflow. Currently supports proteomics experiments with complex experimental designs for DDA-LFQ, DDA-Isobaric and DIA-LFQ quantification.
https://quantms.org
MIT License
34 stars 35 forks source link

Remove EXTRACTPSMFEATURES/IDScoreswitcher from workflow #448

Open ypriverol opened 8 hours ago

ypriverol commented 8 hours ago

Description of feature

In the latest OpenMS (dev) version, the FeatureExtraction process happens within each Search engine adapter MSGFAdapter, SAGEAdapter, etc. We don't have to rerun this step for every Search engine, then this step can be removed it.

Thoughts @timosachsenberg @jpfeuffer before starting.

jpfeuffer commented 8 hours ago

Also remove IDScoreswitcher. I think @timosachsenberg merged my simplification changes

timosachsenberg commented 8 hours ago

I defnitely merged also some IDScoreSwitcher changes but I did not test it in the context of quantMS. I think it is a valuable change but should be done with two PRs that are each tested on some small data.

ypriverol commented 7 hours ago

What about changing all the tests in singularity and docker to point to OpenMS Dev. Which we sync the test and new deployments and changes in OpenMS with quantms more clearly-tidy. ?

ypriverol commented 6 hours ago

Related to this topic, we have one flag called posterior_probabilities = 'percolator' this flag defines the posterior_probabilites used in the workflow. However, I have never run this flag without that value. Can we remove it @daichengxin @jpfeuffer? We should always use Percolator probabilities.

daichengxin commented 6 hours ago

We could remove this parameter and the corresponding modules if we make sure always use the Percolator probability. But I don't see any harm in leaving it there. I don't know if any other users use it, at least we

ypriverol commented 6 hours ago

No harm, but confusing if is always used with same value.

jpfeuffer commented 6 hours ago

Fine with me. Percolator will not work for very small data though.

ypriverol commented 6 hours ago

How small are we talking here?

jpfeuffer commented 5 hours ago

Hard to say. @timosachsenberg knows better. 1000 PSMs after FDR filter?

ypriverol commented 5 hours ago

Ok, quantms is not designed for that anyway. I haven't seen any msruns with less that 1k psms.

timosachsenberg commented 4 hours ago

Hard to say. @timosachsenberg knows better. 1000 PSMs after FDR filter?

Yeah maybe 400 would still work. I think we added it as fallback to be able to analyze small studies or blanks (!) if we do per file percolator

ypriverol commented 4 hours ago

If people don't use it really, it should be here unless someone needs it, BTW, the pipeline becomes more complex for something that hardly happens.

BTW, @timosachsenberg @jpfeuffer @daichengxin The PR removing posterior_probabilities = 'percolator' is ready here https://github.com/bigbio/quantms/pull/450