Search
Question: PCA plot on DE genes do not separate samples. DESeq2.
0
gravatar for g.atla
2.7 years ago by
g.atla0
g.atla0 wrote:

I am running the DE analysis using Deseq2. When I plot the PCA of differentially expressed genes ( ntop=500) using the command below, I do not see a clear difference between in treated vs untreated.

de <- rownames(resdds[ (resdds$padj<0.05) & (!is.na((resdds)$pvalue)) & (!is.na(resdds$padj)),])
data <- plotPCA(vsd, intgroup="treat", returnData=TRUE, ntop=500)
percentVar <- round(100 * attr(data, "percentVar"))
ggplot(data, aes(PC1, PC2, color=treat)) + geom_point(size=3) + geom_text_repel(aes(label=row.names(data))) + xlab(paste0("PC1: ",percentVar[1],"% variance")) + ylab(paste0("PC2: ",percentVar[2],"% variance"))

Are my results are reliable ? Here is the MA Plot and the GO looks god.

Tanks in advance.

 

ADD COMMENTlink modified 2.7 years ago by James W. MacDonald48k • written 2.7 years ago by g.atla0
4
gravatar for James W. MacDonald
2.7 years ago by
United States
James W. MacDonald48k wrote:

I disagree. You get almost complete separation between the groups on the second principal component. Anyway, principal components is looking at the differences in samples based on a linear combination of the top 500 genes, whereas each univariate comparison is based on individual genes. There will always be differences between an aggregation of data and individual observations, and the former doesn't invalidate the latter.

ADD COMMENTlink written 2.7 years ago by James W. MacDonald48k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 356 users visited in the last hour