Fix: Strip None values from orthology mapping API response

[Swargambharath987]

Fixes
#1150

The list comprehensions in both the GET and POST handlers of the orthologs API could include None values when normalize_gene() fails to find a match, which gets serialized
as null in the JSON response. Updated both comprehensions to filter out None results using a walrus operator guard.

[adkinsrs]

We need to ensure that if the resulting JSON is empty after all the None/nulls are removed, the UI handles that situation appropriately if it isn't already doing so. Still need to figure out the root cause as well as per ticket #1150

[Swargambharath987]

Looked at the root cause: The None values come from comma-separated gs2 values in the orthology files — calm1a maps to "CALM1,CALM3,CALM2" stored as a single string. map_single_gene() was treating this as one gene name, which never matches anything in the dataset's gene map, causing the mapping to fail and return None. Fixed by splitting gs2 values by comma so each ortholog is returned as a separate gene symbol. Tested against a real local dataset : calm1a now correctly returns CALM1, CALM3, CALM2 and all three are found in the dataset. Single-mapping genes like "Actb" are unaffected.

[adkinsrs]

Good catch there!

Can you also apply this to the map_multiple_genes function below the one you edited? Also I don't think the "strip" method is necessary... did you find any genes with whitespace or newline characters?

A good way to test this ortholog functionality on the UI is to perform a gene expression search for "calm1a" on the front page using the Hearing (default) profile, and see if some of the displays show a dropdown box in the upper-left to let you choose an ortholog. I'm attaching the inverse of this (searching "calm1" in zebrafish datasets, as "calm1a" is a zebrafish gene)

[Swargambharath987]

Removed .strip() from map_single_gene() and applied the same comma-split fix to map_multiple_genes().