Hi, I am analyzing RNA-seq data and I used Stringtie to assemble transcripts and quantify them. Now to perfom DE analysis on them (using DESeq2), I used tximport which is recommended by the authors of DESeq2 to convert the count values generated from Stringtie to integer values that DESeq2 can use.

This is my R code -

quantdata <- c('path/to/18T3L/t_data.ctab', '/path to/18T6L/t_data.ctab', ....)

tmp <-readr::read_tsv(quantdata[2])

tx2gene <- tmp[, c("t_name", "gene_id")]

txi_output <- tximport(quantdata, type = "stringtie", tx2gene = tx2gene)

sampleTable <- data.frame(condition = factor(c("P", "P", "NP", "P", "P", "P", "P", "NP", "NP", "P", "P", "P", "P", "P", "P", "P", "P", "P", "P", "P", "NP", "P", "P", "P", "NP", "P", "P", "P", "P", "P", "NP", "NP", "NP", "NP", "P", "P", "P", "P", "P")))

coldata <- sampleTable

coldata$condition <- relevel(coldata$condition, ref = "NP")

library(DESeq2)

dds <- DESeqDataSetFromMatrix(
     countData = txi_output$counts,
     colData = coldata,
     design= ~ condition)

However, running this gives the error -

Error in DESeqDataSet(se, design = design, ignoreRank) :   some values in assay are not integers

And when I check the values using head(txi_output$counts), the corresponding values are not integers, but decimal.

So basically, tximport is not converting the count values to integer, which is one of its functions. Any pointers to what I may be doing wrong here?

EDIT : The problem was, as Ram points out below, that I was using DESeqDataSetFromMatrix(), whereas I should use DESeqDataSetFromTximport(). Changing the last line to the following solved the issue -

dds <- DESeqDataSetFromTximport(
     txi = txi_output,
     colData = coldata,
     design= ~ condition)
dds <- DESeq(dds)