AFAICT, the API doesn't currently provide a way to access a transcript's ag_id. It'd be nice if there was an endpoint like getAgIds() that accepts a vector of transcript IDs and returns a data frame (basically the MySQL transcript table for those transcript IDs).