search

cytoscape / idmapper

Identifier mapping
2 stars 1 forks source link

Support "Mixed IDs" or "Unspecified" in Map from list #19

Closed AlexanderPico closed 6 years ago

AlexanderPico commented 8 years ago

The next feature I'd like to see (and would use on a weekly basis) would be support for "mixed" or "unspecified" Map from datasource. That way I could map from a column with more than one ID type to a new column with a single type (i.e., ID unification). I could also map from a columns on unknown type to a new type.

In terms of implementation, can we add "unspecified" and then send in a batch query without specifying the source type? Nuno does this in Goliath, so there must be a way...

AdamStuart commented 8 years ago

There isn't documentation on xrefsBatch command yet, so its not clear. As I read it, you can mix source databases, but need a second column to identify the source of each id (which I don't think we have).

AdamStuart commented 8 years ago

I guess we could guess the source on a line by line basis. For many databases, the pattern is specific enough but HGNC matches everything. It would have to be the lowest priority guess.

AlexanderPico commented 6 years ago

There is a batch method for mixed queries: http://www.bridgedb.org/swagger/#/ (see post /{organism}/xrefsBatch )

It requires knowing the datasource per ID, which for this use case we do!

Maybe if the user specifies "Mixed", then they could be prompted to specify another column containing the datasources? For example, open any WikiPathways into Cytoscape and you'll see the XrefDatasource providing datasource per row of XrefId. Then it's just a matter of constructing the POST body with the info from these two columns and then a single column of requested ID types should be returned.

AlexanderPico commented 6 years ago

Decided this was a use case specific to WikiPathways. Not going to impl here.