Dear everyone,

I'm using EGAD to evaluate a gene coexpression network based on RNA-Seq data. My aim is to use this GCN as a tool for gene function prediction. Run_GBA function from EGAD has three different outputs: prediction scores for each of the annotation sets (AUCs matrix) , the average performance across all annotation sets (AUC value) and gene function predictions based on the whole set of interactions (Prediction matrix). Thus, prediction matrix correlates genes and annotations and it gives a value of how good that prediction is.

The problem is I didn't find any information on EGAD's manual about how interpret that matrix. I don't know what threshold use to select "good" predictions and "bad" predictions. I hope someone could help me.

Thank you!