A while back, we decided that it’d be optimal for me to analyze the individual libraries with WGCNA. Given the complexity of WGCNA, I decided to first follow the WGCNA tutorials, and then attempt to follow along with Yaamini’s WGCNA script with the data for a single crab over time, and then scale it up to blocks of crab. The plan is to use cbai_transcriptomev2.0 (AKA cbai_hemat_transcriptomev2.0), as we’ll group both hematodinium and bairdi genes together based on expression. That way, we might see some interesting interactions between parasite and host!

Well, I did…most of that! After finishing the tutorials, I tried to run WGCNA on a single crab. I got most of the way through, but hit a wall when trying to relate modules to external traits - since I was tracking a single crab, I had no replication, and therefore couldn’t progress further.

The good news is that hey, getting that far meant I was ready to scale up to a full analysis! So I started another run through WGCNA, this time looking at all individual libraries from the ambient-temperature treatment group. This tracked crabs A, B, and C over time.

[NOTE: As of 2021-12-13, I realized I made an error - you need 15+ samples for an adequate WGCNA run. Since this only had 9, I deleted the script. The one linked in the next paragraph is a later version examining all samples available, just in case you need a reference]

Everything ran fairly smoothly at first, but then I hit another wall. My computer just doesn’t have enough memory to run the adjacency() function with this quantity of data. Therefore, I’m going to have to re-run this on a computer with more memory. But good news - looks like everything is running reasonably well!

Next step is just to figure out what the best way to re-run this is - Yaamini’s been doing some stuff to get R updated on Mox, so that might be worth doing…