I think I can finally say we've got a handle on our data! It's a lot to manage, but it seems like after kind of a rough week of rerunning (and rerunning, and rerunning...), everything is going fairly smoothly. We're continuing to produce more results for the various confidence measures. Today BioGRID GoSemSim confidence scores were the highest priority.
Again, for GoSemSim, we found little to no change in the function prediction accuracy. Frustrated, we talked to Lenore about why this was happening. This result is due to the distribution of DSD and the expected value of any given averaged element. She suspected the result would vary a little bit more from the original than it did, but the reason remains the same. So, knowing that I was about to head back home for the 4th of July, Lenore charged Thomas and I with brainstorming new methods to incorporate confidence into function prediction.
Happy 4th of July!