Thanks a lot, Actually I was going to detect the relationships of two experiments. Two developmental courses one done by bulk and another done by single cell seq. PCA says the samples are pretty different with each other (2h in bulk vs 2h in single cell seq). Then I am going to detect what is the reason, for instance read depth, difference in kits, biological basis, etc. But that would be worth to detect genes explaining this difference.
Defining genes explaining a principal component
Hi,
I have 2 matrices of two separate but close experiments, I merged matrices and ran PCA on merged data by DESeq2 that looks so
dds=DESeqDataSetFromMatrix(countData = merged_matrix,colData = mycols, design = ~condition)
vsd <- varianceStabilizingTransformation(dds)
plotPCA(vsd, intgroup = "condition")
How I can extract the genes or the cause of 24% PC2?
1 answer
What you are looking for are the rotated component loadings, i.e., each gene's contribution to each PC. I'm not sure that the plotPCA function of DESeq2 returns these. If you do PCA using base R functions, then you can access the rotated component loadings via the rotation variable of your PCA object.
If you follow my quick tutorial ( PCA plot from read count matrix from RNA-Seq ), you could access these via project.pca$rotation. These will be the eigenvalue representation of each gene's importance to each PC.
See also Michael Love's answer on Bioconductor: https://support.bioconductor.org/p/82805/
Kevin
• 1 views
•
link
written
by
Kevin Blighe
Okay, so you've merged single cell and bulk data together? Have you dealt with the obvious batch effects that exist or is it the batch effects that you want to explore? There are too many colours in your plot, so, I will leave it to you to interpret the different groupings :)
• 1 views
•
link
written
by
Kevin Blighe
Thank you, Actually I don't know how to explore any batch effect or reason of this much of difference between bulk and sc seq. I have two experiments : R and T in 2, 4, 6, 8, 10, 12, 14 and 16h time points for each of experiments. If you consider, the track of development is appear from both R and T assays but they are not match and a task would be the exploitation of this difference. I thought about extracting highly expressed genes for PCA2 that seems a gap between assays.
Log in to answer this question.
More posts like this
-
Batch correction of two RNA-Seq data sets
written by gogeni5529 •
-
DESeq2: DESeqDataSetFromMatrix Error
written by Athena •
-
In DESeq2, using plotPCA how do I go about extracting genes making up PC1 and PC2?
written by mgranada3 •
-
Constrained rda for a PCA made with DESeq2?
written by luzglongoria •
-
PCA plot, less points than samples
written by eridanus •
-
DESeq2 unpaired sample RNAseq analysis
written by Adeler001 •
-
Correcting for batch effect in RNA-seq data
written by Rimma •
-
Individual genes plots after batch correction with Deseq2
written by Eugene A
-
DESeq2 multiple group comparison
written by dazhudou1122
-
Different PCA plots using rlog and vsd on the same data set
written by lirongrossmann •
