Building RNAseq pipeline STAR-RSEM-tximport-DESeq2

Question

[jkim]

Hello,

I did some googling about mRNAseq pipeline and built that pipeline. So I would like to know if my pipeline is okay.

Reference genome : Human gencode GRCh37
Aligner : STAR
Expression quantification : RSEM
Gene count table convert? : tximport
Statistical test : DESeq2

it seems like people use htseq or featureCount for gene count table and put it into edgeR or DESeq2. I am not sure, if I go for RSEM to run DESeq2 instead of htseq or featureCount. I tried kallisto and slueth, but I want to stick to this traditional alignment method.

Any advice would be apprieciated.
Thank you,

[GenoMax]

Unless you have a very specific reason to use GRCh37 you should consider using GRCh38. 38 has been available for a number of years and is now mature.

[jkim]

Thank you for your comment. I have some old data based on GRCh37, So I go with GRCh37. Maybe next project I will go for GRCh38.

Answer 1

[Matteo Schiavinato]

I would like to know if my pipeline is okay.

RSEM can use STAR within its own script, so I would do that instead of running STAR yourself, unless you want to sophisticate your mapping more.

I am not sure, if I go for RSEM to run DESeq2 instead of htseq or featureCount.

The output is comparable, it's a matter of choice, perhaps RSEM is more sophisticated. But in the end it's counting mapped reads per gene/transcript.

I tried kallisto and slueth

Read this: http://www.rna-seqblog.com/limitation-of-alignment-free-tools-in-total-rna-seq-quantification/