pnrobinson
commented
3 years ago @karleg So the simplest version of the app would be to show a summary of results for all genes with (1) a differential transcript AND (2) at least one PROSITE annotation. A more sophisticated version would be to apply a filter and ask if the differential splicing event affects at least one prosite motif -- if yes, then we show the gene in the output page. I think this would be a little nicer. We can now also easily do a GO analysis and put those results on the webpage for both DE and DAS
Figure out what the inputs are for isopret.
Data -- Suggestion would be that we expect all isoform level data to go through HBA-DEALS. We thus expect the output of HBA-DEALS to be the input for isopret. -- Event-level data: Probably we should support RMATS to start with but we need to abstract the data formats to an interface that would work with output data from other RNA-seq tools
Input data model -- We need to decide whether to transform all event-level data into an isoform model (and thus everything that comes downstream just has to worry about the same model) or if we need to retain the event-level information.
Knowledge -- We need to decide how to deal with knowledge about protein domains and motifs -- For starters, we should ingest a CSV file from biomart, but we may want a more general solution that would also work with novel isoforms.