Back in mid-July, I manually created modules to cluster genes by expression patterns. Again, “manual” oversells it a bit - I really just provided a cut height, looked at the resulting clusters, and then named them by overall expression patterns. A recap of that naming system can be found in my lab notebook entry on 2021-07-12 (named Manual Modules 3)
Anyway, clusters were created for each individual crab. However, after some discussion, we concluded that the previous bar for genes was a bit low. Previously, our bar was taking the libraries for the individual crab (say, the 3 libraries for Crab A), and only including genes with 5+ counts among all libraries. So for instance, a gene with 4 counts on Day 0, 0 on Day 2, and 1 on Day 17 would be included. Of course, the downside of this is that a) that’s a really low bar, so we were looking at expression that probably wasn’t meaningful, and b) we were ignoring genes that might be highly-expressed in other crab
As a result, we changed the bar - the new one is only including genes with 30+ counts in at least one library for ANY crab, and at least 1 count in that particular crab. So a Crab A gene with 1 count on Day 0 and 0 on Day 2 and 17 (but 35 counts on Day 2 for Crab B) would be included for Crab A.
Here are our results! I’ll post the tables for percentages, along with a link to the actual files for both percentages and raw counts.
For cbai_v2.0, Crabs D-F (lowered-temp treatment group) seem to have more HTL and LHL genes than Crabs A-C (ambient-temperature treatment group). This also seems to be true for cbai_v4.0, but does NOT seem to be true for hemat_v1.6
Overall, it doesn’t look like the variation is quite as extreme as it was with our previous method.
Crabs G-I really don’t have much useful data - too many MIX to indicate much of anything
Generally, not really seeing too much of an established pattern here