So last time I ran WGCNA, I was running it solely on individual groups of crabs (i.e. all elevated-temp or ambient-temp crabs over time), examining solely the effects of day rather than temperature. However, that isn’t really the best way to use WGCNA - it can look at correlation to multiple variables simultaneously. So I decided to go back and look at two more comparisons of libraries over time!

All of these comparisons were performed on libraries aligned to our three main transcriptomes - cbai_transcriptomev2.0 (unfiltered by taxa), cbai_transcriptomev4.0 (filtered to only include presumed Chionoecetes sequences), and hemat_transcriptomev1.6 (filtered to only include presumed Alveolata sequences).

Furthermore, each comparison included variables for day (0, 2, 17), temperature (elevated, ambient, lowered), infection status if relevant (infected, uninfected), and crab ID.

First comparison: Ambient and Elevated-temperature crab (Ambient vs. Elevated). For this, all available time points for crabs A-C (ambient-temp treatment group) and G-I (elevated-temp treatment group) were used. Reminder: due to a mortality event in the elevated-temp group, we only have samples from timepoints 0 and 2 for this group. Furthermore, since they were kept at ambient temperature when Day 0 samples were taken, Day 0 samples for crabs G-I are designated as being ambient-temperature.

Second comparison: All crabs. This added crab E to the mix (and for cbai_transcriptomev4.0, crabs D and F). All three of these crab were part of the lower-temperature treatment group.Crabs D and F were only used when examining cbai_transcriptomev4.0, as these two crab were uninfected, and therefore their libraries weren’t aligned to transcriptomes that included Hematodinium sequences.

When I ran WGCNA on all six of these comparisons (3 transcriptomes x the above 2 comparisons), it found a variety of significant modules (pval <= 0.05)! Here’s a table of them. Also I found no significant modules when examining hemat_transcriptomev1.6. That said, since the transcriptome is quite small, I should consider re-running with a smaller minimum module size - it’s currently at 30.

Transcriptome Variable Comparison Module color p-value Number of genes
cbai_transcriptomev2.0 day Ambient vs. Elevated steelblue 0.04 656
cbai_transcriptomev4.0 day Ambient vs. Elevated darkgreen 0.02 90
cbai_transcriptomev4.0 day Ambient vs. Elevated darkgrey 0.01 399
cbai_transcriptomev4.0 day Ambient vs. Elevated white 0.02 43

Modules, heatmaps, and other graphs can be found at the following link. Click on the transcriptome of choice, then either amb_vs_elev_no_filter (for the first comparison) or all_crabs_no_filter (for the second). Then select day0_elev_as_amb (as background, I also ran WGCNA on each comparison counting Day 0 Elevated and Day 0 Lowered with temperature as either Elevated or Lowered, rather than ambient. However, since I already included crab IDs, the temperature at the time of sample seems more relevant and is therefore used. These comparisons are put in folders day0_elev_as_elev, meaning that Day 0 Elevated-treatment group crab had their temperature noted as Elevated).

Next Steps:

  • Determine whether to reduce minimum module size for hemat_transcriptomev1.6 WGCNAs
  • Separate genes by taxonomy for cbai_transcriptomev2.0 modules
  • Use GO terms to link genes to biological processes.

NOTE: A previous version of this post noted an error that I realized I made for the All Crabs comparison, but gave the results with errors included (and the caveat that the error was present). I rectified the error, and edited the post and table to show the error-less results