I have carried out a DESeq2 analysis on a relatively small data set (8v8) and have been going through the results. A large number of genes have come out as statistically different, however a very small number of these genes seem to have a very large lfcSE (between 0.6 and 1.5). Even though they are coming out as statistically significant, should I trust these results? Or should I exclude those genes from the reports (eg have a lfcSE threshold of <0.5) and later downstream pathway analysis?
This step in the analysis is up to you, but I will note that you can use lfcThreshold to perform testing at a higher biological significance than LFC=0. If you have a gene with a LFC of 2 and a lfcSE of 0.5, this may be significant for testing a threshold of LFC=0, but would likely not appear in a list when you raise lfcThreshold to 1. The choice of threshold should be chosen based on the biology of the system, and what is a practically meaningful fold change.