Today, a user reported an inconsistency between correlations reported by the volcano plot and the scatter plot in DE2.
Upon investigating we discovered a bug stemming from the way that data was exchanged between two components of the portal. This bug was introduced in the 24Q2 release and manifests when running custom analysis on specific datasets.
This means for the volcano plot may have been incorrect starting with the 24Q2 release when custom analysis run on one of the following datasets:
Proteomics
Sanger Combinations (Library AUC)
Sanger Combinations (Anchor viability)
Batch corrected Expression Public 24Q2
Sanger Combinations (viability)
Protein Array
Sanger Combinations (AUC)
Sanger Combinations (Library viability)
Methylation (1kb upstream TSS)
Copy Number Public 24Q2
While we do have automated testing in place for custom analysis, our synthetic data did not cover this case. We’ll be investigating putting in additional automation to verify behavior on real datasets to avoid issues like this arising in the future.
