[ADD] FileConverter FeatureXML to OMS

[oliveralka]

No description provided.

[oliveralka]

Should it be possible to convert legacy featureXML? Currently it is failing.

Error:
score type must have a name (as part of its CV term)
https://github.com/hendrikweisser/OpenMS/blob/idfileintegration/src/openms/source/METADATA/ID/IdentificationData.cpp#L200

I am using the OpenMS/src/tests/topp/AssayGeneratorMetabo_ams_input.featureXML as test case.

[oliveralka]

Ok, I think the error mentioned above is specific for metabolomics .feautreXML.
This is due to a non-filled ProteinIdentification, which is needed to allow for the storage of PeptideIdentification/PeptideHits to the features.

	<IdentificationRun id="PI_0" date="2021-10-04T18:07:27" search_engine="AccurateMassSearch" search_engine_version="2.6.0-pre-idf-ams-2021-10-04">
		<SearchParameters db="CustomDB" db_version="0.0" taxonomy="" mass_type="monoisotopic" charges="" enzyme="unknown_enzyme" missed_cleavages="0" precursor_peak_tolerance="5" precursor_peak_tolerance_ppm="true" peak_mass_tolerance="0" peak_mass_tolerance_ppm="false" >
		</SearchParameters>
		<ProteinIdentification score_type="" higher_score_better="true" significance_threshold="0">
			<UserParam type="stringList" name="spectra_data" value="[file://I:\OpenSWATH_Metabolomics_data\20181121_full_data\04_PestMixes_individually_Solvent_DDA_20-50/PestMix1_1ngSolventDDA20-50.wiff]"/>
		</ProteinIdentification>
	</IdentificationRun>

This is needed in the importIDs function:
https://github.com/hendrikweisser/OpenMS/blob/idfile-integration/src/openms/source/METADATA/ID/IdentificationDataConverter.cpp#L67

Setting a dummy score in the exportIDs function:
https://github.com/hendrikweisser/OpenMS/blob/idfile-integration/src/openms/source/METADATA/ID/IdentificationDataConverter.cpp#L543

Leads to further isses in the importIDs function, such as:

Progress of 'converting peptide identifications':
Warning: Trying to import PeptideHit without a sequence. This should not happen!
-- done [took 0.00 s (CPU), 0.00 s (Wall)] --
Error: Unexpected internal error (error inserting data:  Parameter count mismatch)
<Warning: Trying to import PeptideHit without a sequence. This should not happen!> occurred 12 times

IMHO it would be best to allow empty score types (at least in this case) - what do you think @hendrikweisser?

Edit:
Maybe this can also be fixed with the "correct" settings in the AccurateMassSearchEnginge. I am setting the ProcessingSteps, so I am not sure why there is no score_type attached to the ProteinIdentification.
https://github.com/OpenMS/OpenMS/blob/idf_ams/src/openms/source/ANALYSIS/ID/AccurateMassSearchEngine.cpp#L656

[hendrikweisser]

Thanks for the investigation, @oliveralka!

IMHO it would be best to allow empty score types (at least in this case) - what do you think @hendrikweisser?

I don't think that would really solve the problem. As you've seen, with a dummy score you just run into the next issue further on ("PeptideHit without a sequence").

I think we should improve how the empty ProteinIdentification (which is only needed for legacy technical reasons) is handled. If there are no ProteinHits and score_type is empty, no ScoreType should be created. (Caveat: I haven't looked at the code again to see how this is handled currently.)