Well, you haven't done research until you've wasted a full day messing around with data issues.
:P
Really, it's bound to happen at some time, and that day was definitely today. Off-by-one errors are especially traumatizing when your code takes awhile to run (this was completely my fault), and file corruption problems are equally frustrating. After attempting to rerun the pipeline starting from the off-by-one problem, we made it to the 97th file out of 100 and the whole thing crashed and burned due to a corrupted file. Ah, well.
The stars (and bits) were not aligned for us today, so we kicked off the program a third time and caught up on reading academic papers.
Our code [finally] ran to [correct] completion, so we ran majority vote on the averaged results. We got waaay different results than expected -- alarmingly close function prediction values to the original. It's very interesting, and we'll get to do some digging (soul searching?) on why that is.
Another big piece of today was keeping up with documentation as we rerun DSD for Biogrid weighted literature confidence values.
We ended up running a lot of the 100 Biogrid weighted literature networks again, due to being kicked off the server. Instead, we were able to dedicate the extra time to making more optimizations, which is super fun. I'm finding that I learn a LOT more about python going through this process. I love it.
Optimizations are becoming really fun and tinkery. We had a slow day in terms of churning out results for DSD, but I'm ecstatic because of all the optimizations we've been able to make. It's really rewarding. The Biogrid 100 weighted lit DSDs are averaging themselves off in serverland with the shiny new code I wrote yesterday. I'm SURE it could be improved more, since a) almost any optimizations will make a big difference when you have 100 - 6000x6000 matrices, and b) I'm still learning the ins and outs of python. Fun times!
We finally got the 100 biogrid weighted literature networks to finish averaging, and packaged them up to run in majority vote. Again, we got alarmingly similar results to the original DSD. My guess is that next week we'll be spending a good deal of time figuring out why this is, and discussing these results with Lenore.
So, I ended up buying a road bike...
I've been meaning to get a nice one for about 3 years now, and now that I'll be living 1-1/2 miles away from campus next year I'll get even more use out of it. I ordered one a few days ago and it finally got here! Yay mobility!
Spent the day biking around again, getting groceries and running errands, and so on. It's sooo much easier to get to Davis square (only takes about 5 minutes), which means that I can get virtually anywhere in the Boston area in about 45 minutes (mostly less). Off to play tourist now!