add covariates to graphs themselves

[gkiar]

Such as dataset, subj-id, sex, age, scan-id, etc...

Hurdle: in order to be done at graph generation it requires that we only accept a specific file format (.csv) with certain organization of covariates/data within (i.e. if no value was measured should a column be blank or contain #? Should 1/0 mean male/female? What should similar columns be named so you can compare across datasets? etc... I don't know of a standard for this at the current time).

[raamana]

Hi Greg, I am thinking of functionality to handle covariates for neuropredict, and I am wondering if you have come across any other packages to handle. The concern reg input format is not a concern to me.

[gkiar]

Hey Pradeep - I'm not sure what you mean about other packages? I think any graph library that supports attributed graphs will enable you to do this, such as NetworkX or igraph.

[raamana]

I am referring to the covariates part. Actually I missed the graphs part entirely :).. I am looking to add them to neuropredict, to regress the covariates before doing the predictive modelling. Never mind then.

[jovo]

@loftusa store json/toml file associated with each graph. coordinate with @j1c and @falkben to keep track of things.