lfcShrink types for Volcano plots
Entering edit mode
alanh • 0
Last seen 24 days ago
United States

Hi! I got a few questions about shrinking log2foldchanges using lfcShrink apeglm vs normal. I'm pretty inexperienced in the statistics side of this so any further reading is appreciated too.

I'm trying to create volcano plots (using EnhancedVolcano) to compare differentially expressed genes between two groups. My code is below.

txi <- tximport(files, type = "kallisto", tx2gene = t2g, ignoreTxVersion=TRUE)
dds <- DESeqDataSetFromTximport(txi, samples, ~ type)
dds <- DESeq(dds)
res <- lfcShrink(dds, coef="type_nonGEM_vs_GEM", type="apeglm")
res2<- lfcShrink(dds = dds,type = "normal",contrast=c("type","nonGEM","GEM"))

enter image description here

I know apeglm seems to be recommended over normal, but it doesn't look like volcano plots I've seen in the past, whereas "normal" creates a different looking plot. Is this a normal output or did I go wrong somewhere? How should I decide which type to use?

DESeq2 • 149 views
Entering edit mode
Last seen 1 hour ago
United States

Typical volcano plots (using MLE estimates of LFC) have a lot of noise on the LFCs from genes with small counts. That's why they look that way, but it's not desirable. It distracts the eye away from estimates of LFC that have high precision. You should use apeglm or ashr.


Login before adding your answer.

Traffic: 319 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6