mzIdentML Usage Patterns about psims

[JB-MS]

We were planning to integrate the mzid writer as a node in our framework Ursgal. At the moment we have .csv as the final output format, but since mzIdentML is the official standard we would like to add a converter to convert the .csv files into mzid, to be compatible with repositories like e.g. PRIDE.

@mobiusklein, if you are interested and have time, you are welcome to join and contribute to Ursgal, and implement your writer into the framework.

I am looking forward to contribute to improvements of the design. At the moment i have no specific adjustment/improvement ideas, but maybe we will see what can be optimized, when it is integrated into Ursgal?

[mobiusklein]

I see several nodes in ursgal dealing with csvs. Which one should I be looking at for an equivalent result data structure?

[JB-MS]

That is correct. We rely on a workflow right now, which passes the csv from one node to another, which then read and change the csv accordingly.

Simplified example workflow:

mzML --> Search engine --> result file (e.g. xml, csv or mzid) --> conversion to csv --> unify csv (add e.g. missing meta info) --> map peptides --> validate PSMs with e.g. Percolator

See also e.g. this example script

In an ideal world we would deal with mzid the whole time, but we think it is also beneficial to have the easily accessible and readable csv format in your workflow.

I imagine the mzid writer at the end of an workflow , after the PSMs were validated using Percolator and then offer a (optional) node to convert these results into a mzid.

What do you think?

[mobiusklein]

That's reasonable. My use-case holds intermediary results in a sqlite3 database before translating the database into either CSV or mzIdentML as needed.

I'm looking for a definition for an output data structure in the ursgal repository, but I don't see a class or consistent dict searching the codebase. Does one exist?

[JB-MS]

This is indeed missing at the moment. We rely on mzid to csv conversion using a Java library . We used this csv format as a template and modify and add some fields.

I will set up a new parameter in uparams.py (the general parameter storage and required for correct mapping), where the fieldnames for the csv are specified and described. This should be useful to set up a correct converter.