Hello!

I am relatively new to using DESeq2 and was wondering if anyone could help me explain why there is a discrepancy between my normalized counts and the log2FoldChange in the results. Basically, my normalized counts show that a specific gene (dptA and dptB in the example) should be downregulated in my treatment, however, the DESeq results shows a Log2FoldChange which is greater than 0. Below you can find the normalized counts as well as the results from the code below.

coldata <- data.frame(row.names=colnames(countdata), condition)
coldata
dds <- DESeqDataSetFromMatrix(countData=countdata, colData=coldata, design=~condition)
dds

# Run the DESeq2 pipeline
# Get differential expression results

dds <- DESeq(dds)
keep <- rowSums(counts(dds)) >= 10
dds <- dds[keep,]
res<-results(dds,alpha=0.05)
summary(res)
## Write results, Table containing the significantly differential expressed genes
write.table(res,"DESeq2_resuls.txt",sep="\t",col.names=NA)
DESeqResults<-read.table("DESeq2_resuls.txt",sep="\t",header=T)

These kinds of questions come up regularily and usually it is because either the counts one compares with are not (properly) normalized or there is something wrong with the factor levels of our condition. Check the column of the colData that is relevant and ensure that the correct entry is set as reference level. Also be sure to use counts(dds, normalized=TRUE) to get the norm counts and be sure to use the correct coef or contrast during the results() step. Do you only have two groups here?

Just make your life easier... Specify the contrast you want in the results call.