Entering edit mode
Arun
•
0
@Arun-24476
Last seen 3.8 years ago
Hello,
I am running time-series analysis and was able to create design and analysis successfully. However, I have few doubts in using transformed matrix
ddsObj = DESeqDataSetFromTximport(txi = count_data, colData = sample_df, design = ~TIME)
# remove too low counts <1
keep <- rowSums(counts(ddsObj)) > 1
ddsObj <- ddsObj[keep,]
# run DESeq analysis with reduced model with time point 1 as reference
ddsTC <- DESeqDataSet(ddsObj, ~TIME)
ddsTC <- DESeq(ddsTC, test = "LRT", reduced = ~1)
resTC <- results(ddsTC)
resTC$symbol <- mcols(ddsTC)$symbol
head(resTC[order(resTC$padj),], 4)
I wish to know, how to use a transformed matrix? For instance I performed VST and unable to use it as deseq object.
vsd = varianceStabilizingTransformation(ddsObj)
boxplot(assay(vsd))
plotPCA(vst, intgroup=c('Temperature','TIME'))
ddsTC <- DESeqDataSet(vsd, ~TIME)
#renaming the first element in assays to 'counts'
#Error in DESeqDataSet(vst, ~TIME) : some values in assay are not integers
sessionInfo( )
R version 4.0.1 (2020-06-06)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Catalina 10.15.5
Yes. vignette says "vst and rlog are provided for applications other than differential testing". If I understand it right, size factor estimation & normalization factors incorportated in DESeq(obj) removes batch effects ?
No, this is unrelated. Normalization is correction for sequencing depth and library composition, and batch correction is adjustment for unwanted technical variation. There are many threads addressing this via google and the search function, biostars.org has many as well.