Open jmrussell opened 3 years ago
I have run into the same issue. I created a custom database with interactions_input file labelling gene_a, gene_b. Then I run statistical analysis using this database, and in the resulting pvalues file, gene_a and gene_b have been swapped almost half the time. I can easily fix in my results, but unfortunately since I did not expect this behavior, spent quite some time analyzing the output incorrectly. It would be great to have this fixed.
When creating a database of ~3000 interactions, roughly half of them are reversed from the order they are in my interaction_input.csv. This isn't a huge problem because I can show the cluster comparisons between cluster a | cluster b and cluster b | cluster a depending on the interaction, but it would be nice if the order was maintained throughout the entire db creation process.