quantms subworkflows to be built

[daichengxin]

What subworkflow to be built? such as QC?

DecoyDatabase and OpenmsPeakpicker should be called as single module.

[ypriverol]

@daichengxin @jpfeuffer :

• DecoyDatabase should be one single module because is just one step.
• OpenmsPeakpicker I think could be added to the Conversion subworkflow (the one that you already implemented). The OpenmsPeakpicker is part of the conversion process and should be enable on request by a parameter.

[ypriverol]

The next subworkflows:

• DatabaseSearchEngines including the following processes: process search_engine_msgf, process search_engine_comet
• index_peptides: Could be a single module that can be reuse multiple times.
• PercolatorPeptideIdBoost including the following processes: extract_percolator_features, percolator
• PSMFDRControl including the following processes: fdr_idpep, idpep, idscoreswitcher_to_qval, consensusid , fdr_consensusid, idfilter
• PhosphoScoring including the following processes: idscoreswitcher_for_luciphor, luciphor
• ProteinInference including idemapper, file_merge , protein_epifany, protein_inference, epi_filter , resolve_conflict
• ProteinQuant including pro_quant, MsstatsConverter
• pmultiqc should be a single module.

What do you think about this proposal @daichengxin @jpfeuffer ?

[ypriverol]

I think index_peptides can be the end of the DatabaseSearchEngines subworkflow because all the time will be needed.

[daichengxin]

DatabaseSearchEngines is ok. And they shared parameters. PercolatorPeptideIdBoost is also ok, they have close relationship. PSMFDRControl seem to have too many modules. idfilter could be a single module that can be reuse in ProteinInference. PhosphoScoring is ok. idemapper and file_merge seem to have no direct relationship with protein inference. idemapper should be with isobaricanalyzer. file_merge can be a single module. ProteinQuant and pmultiqc are ok.

[jpfeuffer]

I would do:

• DatabaseSearchEngines including the following processes: process search_engine_msgf, process search_engine_comet and index_peptides (starting from 2.7.1, indexing is done internally per default anyway)
• PSMRescoring including the following processes: extract_percolator_features, percolator, fdr_idpep, idpep
• ConsensusID could also be a single module
• PSMFDRControl including the following processes: idscoreswitcher_to_qval (if no consensusID was used), fdr_consensusid (if consensusID was used), idfilter
• PhosphoScoring including the following processes: idscoreswitcher_for_luciphor, luciphor
• IDMapper could be a single module
• FileMergecould be a single module
• ProteinInference including protein_epifany, protein_inference, epi_filter
• ProteinQuant including resolve_conflict , pro_quant, MsstatsConverter
• pmultiqc should be a single module.

[ypriverol]

Looks like a great strategy. I will open independent issues.

[daichengxin]

Isobaricanalyzer could be with IDmapper ? In LFQ, there is no IDMapper

[jpfeuffer]

Yes could be together, in LFQ this happens in the ProteomicsLFQ step.

[ypriverol]

@daichengxin if everyone agreed with @jpfeuffer #8 (comment) we should do a PR by subworkflow. I have created an issue for each subworkflow. This can help us to review the PRs more easily.

[ypriverol]

Done.