Feat: added smiles and inchi of metabolites

[JHL-452b]

Main improvements in this PR:

This PR added a document of SMILEs and inchi information of metabolites to data/modelCuration as discussed in #728 . To ensure completeness, in addition to SMILEs, inchi information is also added to TSV.

I hereby confirm that I have:

• Tested my code on my own computer for running the model
• Selected deXvelop as a target branch
• Any removed reactions and metabolites have been moved to the corresponding deprecated identifier lists

[haowang-bioinfo]

@JHL-452b looks nice

In order to properly evaluate this PR, please provide more detail information about data source(s) and manipulation, preferably with extraction code

just curious about the column of InChIkey which was not mentioned in #728, if it's redundant to InChI?

[JHL-452b]

In order to properly evaluate this PR, please provide more detail information about data source(s) and manipulation, preferably with extraction code

Data source(s) and main manipulation:

Firstly, I downloaded and collated metabolite information from various databases(including KEGG, ChEBI, PubChem(Api), HMDB, Lipidmaps, Recon3D). The collected information of metabolites contains ID and SMILEs, so I used ID mapping relationship in metabolites.tsv to extract SMILEs. Then, the functions of MolToInchiKey and MolToInchi in RDKit was used to convert SMILEs to inchiKey and inchi, respectively.

Results

	metabolites	SMILEs	inchiKey	inchi
All	8460	7141	6435	6435
Unique	4163	3399	2973	2973

just curious about the column of InChIkey which was not mentioned in #728, if it's redundant to InChI?

Inchi is better for accurately representing and comparing chemical substances, while InChIKeys could serve mainly as shortcuts that point back to the original InChI. So I think using both together provides optimal capabilities - InchIKey for fast lookup and matching, with the original InchI storing the richer chemical structure details.

Of course, it's just my opinion.

[haowang-bioinfo]

Inchi is better for accurately representing and comparing chemical substances, while InChIKeys could serve mainly as shortcuts that point back to the original InChI. So I think using both together provides optimal capabilities - InchIKey for fast lookup and matching, with the original InchI storing the richer chemical structure details.

thanks for the info, sounds good to have both Inchi and InChIKeys

Firstly, I downloaded and collated metabolite information from various databases(including KEGG, ChEBI, PubChem(Api), HMDB, Lipidmaps, Recon3D). The collected information of metabolites contains ID and SMILEs, so I used ID mapping relationship in metabolites.tsv to extract SMILEs. Then, the functions of MolToInchiKey and MolToInchi in RDKit was used to convert SMILEs to inchiKey and inchi, respectively.

would be nice to also have the relevant code uploaded to code/modelCuration as references

[feiranl]

@JHL-452b It would be great if you could upload the code into the code/modelCuration as suggested by Hao. Then we can start review this. Thanks!

[JHL-452b]

Sorry late for uploading the code. Thanks for reminding.

[haowang-bioinfo]

@JHL-452b thanks