[feature request] Spectra metadata completeness

[Adafede]

Hi , it seems like you back-calculate some metadata (chemical classes, InChIKey, InChIKey-Planar) from the user input (SMILES only?) of a given spectrum ID, which is really nice! However, it looks like the metadata values are not always filled. They mainly look to rely on the SMILES, however many spectrumIDs have only InChI as structural identifier. Some InChIs have their InChIKeys (and planar) linked to it, others do not.

Would it be possible to do something like:

1. if(is.null(SMILES)){calculate smiles(InChI)}
2. calculate InChIKey+Clasification(SMILES)
3. add 2 additional (or 1 at least) metadata starting from SMILES: SMILES-planar,InChI-planar?

Best,

[mwang87]

Do you mean in GNPS Networking or in the GNPS Library pages?

[Adafede]

Wow, that was quick!

GNPS Networking, when submitting a job and then looking at the table block=main&file=DB_result

[mwang87]

Yes, that makes sense. I think it would be pretty straightforward, do you think you'd be able to submit a quick PR for it? Then we could test it out together?

Best,

Ming

[Adafede]

Huh, will have to dive into some funny parts of GNPS!

I'll do my best and hopefully submit one soon!

Thank you again for the responsiveness!

[mwang87]

Definitley, its kind of a mess and a rabbit hole so apologies. The business end of the code I think you're looking for is here:

https://github.com/CCMS-UCSD/GNPS_Workflows/blob/master/metabolomics-snets-v2/tools/metabolomicsnetsv2/scripts/getGNPS_library_annotations.py#L143

[Adafede]

You were right, more than straightforward! Did the PR for the smiles <-> inchi conversion (5b5c044) as it was easy thanks to the API!

Do you have a similar service for generating planar smiles and inchis or think it could be easily adapted? Might need a bit more work else!