Comparing the output from DESeq and DESeq2

Question

[bioinforesearchquestions]

Hello friends,

Based on this post Question: Differential gene expression based on read counts using DESeq package.

Using htseq_count, I generated the raw read count for each gene for my cancer and normal sample. Then I used DESeq and DESeq2 package to identify differential gene expression.

1) First 4 rows are from DESeq output

2) Next 4 rows are from DESeq 2 output

3) Then followed by HTseq_count output

Just to understand, I manually computed the log2(fold_change) for both DESeq and DESeq2 R package output.

• For DESeq output, I first divided (value2/value1) and then took log(value2/value1,2). I received the same value as computed by DESeq.

• But for DESeq2 output, I divided (val2/val1) and then took log(val2/val1,2). The DESeq2 computed log2 fold change is different from mine. Am I missing something?

Answer 1

[igor]

DESeq2 fold change is "shrunk" to account for sample variability. You can get both shrunk and raw fold change like so: results(dds, addMLE=TRUE).

See more detailed explanation here: https://support.bioconductor.org/p/77461/

[bioinforesearchquestions]

Thanks, Igor for the suggestion and link.