Asymmetric/biased log2FC values for low-expressed genes in DESeq2
Hello everyone,
I apologize if this question seems a bit naive. When we ran DESeq2 for our scRNAseq data (where we first pseudobulked the data by samples), we observed that genes with low expression (those with low baseMean values) exhibited an asymmetry in the log2FC values, as shown in the graph. However, when we used edgeR, we did not observe this asymmetry or bias. Currently, we have not applied any shrinkage methods such as apeglm or ashr, even though we are aware that using shrinkage methods can potentially resolve this issue.
Do you have any ideas or thoughts on why this might be happening? Thanks!
0 answers
No answers yet.
Log in to answer this question.
More posts like this
-
Best choices for DGE and pathway enrichment analysis in single cell data using pseudobulk?
written by txema.heredia •
-
Best Practices for ATACseq for Performing Differential Accessibility Analysis
written by newbioinfograd2323 •
-
DESeq2: adjusted p-values become NA when sub-setting samples
written by txema.heredia •
-
Tutorial: Bulk RNA-seq: Different Expression Analysis
written by Julia Ma •
-
DESeq2 lfcshrink, shrinkage apeglm vs normal
written by joonhong kwon •
-
General question about clustering in scRNAseq
written by Rox
-
Massive L2FC in DESeq2
written by Alex •
-
High fold change values
written by Lluís R.
-
Interpretation of RNA seq pipeline in a paper
written by n.anuragsharma •
-
Intersection of multiple methods for RNA-Seq differential expression: conservative or crazy?
written by Adamc
Why is this an issue? It's not significant. Prefiltering (recommended) will likely solve it. FCs of these genes have large standard error, so it's not reliable.