[Bug] Pathway analysis doesn't work

[saracastel]

Background A user reported that pathway analysis doesn't work. This issue was reproduced using two different datasets:

• User's dataset (experiment ID: 526c18ab-4530-4b88-bdeb-cbfe435adf9c)
• Workshop dataset - GSE183716 - Covid19 MISC v3 (experiment ID: 7dc3cd24-4497-4ed0-b37a-a28de3d56727).

For pantherdb, the error "We couldn't launch pathway analysis" shows up (see attached screenshot).

For enrichr, the error "504 Gateway Time-out" shows up (see attached screenshot).

Goals Investigate and fix the issue so that pathway analysis can be run successfully.

[alexvpickering]

There seems to be a couple of things going on here:

• Advanced filtering doesn't work. E.g. after filtering for down-regulated genes, the table doesn't change (no idea why):

• The UI is selecting an attribute that doesn't exist (some times) in order to get genes:

// src/utils/pathwayAnalysis/launchPathwayService.js
const { total, data } = genesList;

  switch (serviceName) {
    case pathwayServices.PANTHERDB:
      url = serviceUrls[pathwayServices.PANTHERDB];
      params = {
        correction: 'fdr',
        format: 'html',
        resource: 'PANTHER',
        ontology: 'biological_process',
        input: data.gene_id.join('\n'),
        species,
      };
...

• data.gene_id doesn't seem to always exist. I suspect this happens e.g. when features.tsv.gz or seurat_object.rds uploads only contain the gene SYMBOL, not the ENTREZID. PantherDB should be disabled in these cases.

• the PantherDB url is now https (can remove warning about HTTP from modal once we fix)

• the issues with EnrichR seems to be that we were sending too large of a list of genes (filtering wasn't working) so the request was timing out