Hello all,

I am running WGCNA on some data that is set up in two different conditions (A and B) and two different locations within each condition (1 and 2), simplified here to save time and avoid confusion. I have run edgeR differential expression analysis to find genes that are up and downregulated in location 1 of both conditions as well as running WGCNA on the sets individually and via consensus to find modules correlated with location 1. When running WGCNA on the individual data sets (conditions A and B), I found that the upregulated genes from the edgeR analysis were indeed present in modules also positively correlated with location 1 and downregulated genes in modules negatively correlated with location 1, as one would expect I think. However, the consensus analysis did something unexpected. Some of the genes that were upregulated in location 1 according to edgeR were found in modules with opposite correlation and vice versa for downregulated genes in some cases. I ran module membership (kME) and found that all of these genes (the ones that seemed to be in the oppositely expressed modules), also had the opposite sign than the rest of the module had.

My questions are:

1. Why would this happen even though I used a signed network? My command was as follows:

   net = blockwiseConsensusModules(maxBlockSize = 15000, multiExpr, power = 16, TOMtype = "signed", minModuleSize = 30, deepSplit = 2, pamRespectsDendro = FALSE, mergeCutHeight = 0.25, numericLabels = TRUE, minKMEtoStay = 0, saveTOMs = TRUE, verbose = 5)

2. Does it indicate some sort of problem?

Thanks,

Erik

```