I have a metagenomic dataset crossing three time points from which I have mined CAZymes and am using DESeq2 to identify differentially abundant CAZymes from using trimmed mean depth generated my CoverM (very similar to Q2Q3 contig coverage). From this I am see results I do not understand in the form of enzymes that look differentially abundant but DESeq2 says the are not. below is an example. I would appriciate anyone who could enlighten me as to why this is the case.
When looking at a relative abundance* heatmap of a subset of the CAZyme data with those marked as differentially abundant across any time point I notice on the far right several enzymes which appear to the naked eye to be differentially abundant. Endo−1,4−beta−mannanase for example on the far right
When exploring the DESeq2 output for that given enzyme I see the following:
#Between time point 1 and 2 PreVsP_df[rownames(PreVsP_df) == "220.127.116.11",] baseMean log2FoldChange lfcSE stat pvalue padj 18.104.22.168 0.5565868 0.2389833 0.8940182 0.2673137 0.7892277 0.8563372 #Between time point 1 and 3 PreVsF_df[rownames(PreVsF_df) == "22.214.171.124",] baseMean log2FoldChange lfcSE stat pvalue padj 126.96.36.199 0.5565868 -1.132044 0.9063071 -1.249074 0.2116381 0.3301718
As we see here there is no significant difference. However when I explore relative abundance values of this enzyme individually I do not understand why this is the case. Pre = time point 1, Post = time point 2 and Field = time point 3: