Hello BioStars Community,
I need help with understanding why my WGCNA data gives this output in the soft power thresholding step. I have high R2 at soft powers 1-4 and then it declines below the set threshold (0.8). Also, the mean connectivity is very high (~3500) even at soft power 22. I have attached the plots and the sft index data. Looking forward to your thoughts and very valuable help with this. Thank you!
Brief background: It is a time-course RNAseq data. 10 time points each with 3 replicates = 30 samples. I start with a counts file with 11249 genes. After the gsg step to remove outliers, I have 10752 genes. After creating the dds, I further removed genes with counts lower than 5 in 25% of the samples to reduce noise. If I try removing genes with <15 counts in 75% of the samples as suggested by the tutorial I followed, almost all the genes were filtered out. I proceeded with 8642 genes with vst data as well as log2 transformation of the normalized count (estimateSizeFactors) for the WGCNA.
This is the code that I use for the soft power step
power <- c(c(1:10), seq(from = 12, to = 25, by = 5))
sft_25_5 <- pickSoftThreshold(norm_counts_log_25_5, powerVector = power, networkType = "signed", verbose = 5)
sft.data_25_5 <- sft_25_5$fitIndices
