Hi, I am trying to do comparative RNA-seq analysis with DESeq2.

My purposes are: 1. combine transcripts into genes 2. detect gene expression difference under different conditions 3. obtain a summarized gene expression table

1. combine transcripts into genes I basically followed these commands: https://bioconductor.org/packages/release/bioc/vignettes/tximport/inst/doc/tximport.html#kallisto

library(tximport)
txi <- tximport(files, type = "salmon", tx2gene = tx2gene)
names(txi)
library(tximportData)

dir <- system.file("/Users", package = "tximportData")
samples <- read.table(file.path("/Users", "samplelist.csv"), header = TRUE)
tx2gene1 <- read.csv(file.path("/Users", "transcript_gene.csv"), header = TRUE)
all(file.exists(files))
files <- file.path("/Users", "kallisto", samples$sample, "abundance.tsv")
names(files) <- samples$sample

library(tximportData)
 txi <- tximport(files, type = "kallisto", tx2gene = tx2gene1)

Is that a valid way? I found the sum of the gene expression of each sample is largely diverse.

1. detect gene expression difference under different conditions I followed this: http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#why-un-normalized-counts

Since it describes that "As input, the DESeq2 package expects count data as obtained, e.g., from RNA-seq or another high-throughput sequencing experiment, in the form of a matrix of integer values", I used "txt" above as input without normalization.

dds <- DESeqDataSetFromMatrix(countData = txi,
                              colData = colData,
                              design = ~condition)
result <- results(dds, contrast=c("condition","conditionA","conditionB"))

Is that a valid way to do it?

1. obtain a summarized gene expression table

dds <- estimateSizeFactors(dds)
normalized <-counts(dds, normalized=TRUE)

However, I found the sum of the gene expression of each sample is largely diverse. Something like: (two samples for condition A and B)

sample A1 A2 B1 B2
gene1  10  11  1  1
gene2  20  19  2  1
gene3  30  32  3  4
gene4  40  38  4  4

It looks odd and I wonder how to obtain a gene expression table of all the samples, which are normalized by both gene length and between-sample bias. Or, is there a significant gene expression depletion in sample B1 and B2 overall?

Thank you for your help.

Yes that is how to summarize to gene with tximport.

You should use DESeqDataSetFromTximport which is specifically designed for importing from tximport, as the name suggests. See here:

https://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#transcript-abundance-files-and-tximport-input

That table looks like there is DE for those genes. I don't follow what you are trying to show me?