Hello all,

I'm doing single cell DE analysis using DESeq2. I'm getting high log2fc around 20. What mistake do you think could lead to this? My code is as below '' dds <- DESeqDataSetFromMatrix(cluster_counts, colData = cluster_metadata, design = ~ sex+batchlib+status+libsize+region+status:region)

dds$status <- relevel(dds$status,ref="con") dds$region <- relevel(dds$region,ref="ocu")

dds <- DESeq(dds) res1 <-results(dds, contrast=list(c("status_als_vs_con"))) res1 <- lfcShrink(dds, contrast = contrast, res =res1, type="ashr") '' Thanks, Paria

20 is from positive vs 0.

Take a look at the vignette, the workflow, or the paper on the topic of LFC shrinkage.

Thanks for the reply. I did so. But when I do the lfcShrink the log2fc shrink a lot. Here is my results doing lfcShrink and wild test: ' dds <- DESeqDataSetFromMatrix(cluster_counts, colData = cluster_metadata, design = ~ sex+batchlib+status+region+status:region)

dds <- estimateSizeFactors(dds) keep <- rowSums(counts(dds, normalized=TRUE)>=10)>=1).9*36
dds <- dds[keep,]

dds$status <- relevel(dds$status,ref="con") dds$region <- relevel(dds$region,ref="ocu")

dds <- DESeq(dds, fitType='local')

1 res <- results(dds, name ="statusals.regionsc")

2 reslfc <- lfcShrink(dds, coef = "statusals.regionsc", type="apeglm") '

The result for res

' gene baseMean log2FoldChange lfcSE stat pvalue padj

CALCA 17.7 -17.4 3.58 -4.86 0.00000120 0.00559 '

and the result for reslfc

' gene baseMean log2FoldChange lfcSE pvalue padj

CALCA 17.7 -0.000000190 0.00144 0.00000120 0.00559 '

The shrinkage of log2fc resulted in too small value which does not make sense to me. Do you have any idea what it happens? or if it's normal? Thanks, Paria