RNA-seq: differentially expressed genes without genome reference

Question

[guiu]

Hello,

I have RNA-seq data for an organism for which the transcriptome is available, but not the genome (it is not assembled yet). I am interested in differential expression analysis at gene level.

I have been reading about RSEM, CuffDiff, Sailfish, Kallisto, HTSeq, but I didn't get a clear idea of which is the best approach to get differentially expressed genes.

What is your recommendation? Thanks

Answer 1

[Rob]

I'd recommend using Sailfish or Salmon with tx-import to get fast accurate gene-level results (example). tx-import will allow you to import these results directly into DESeq2 (or prepare them easily for most other downstream DE packages).

Answer 2

[Antonio R. Franco]

You have two choices in the absence of a reference genome 1. Annotate you transcriptome with tools like Trinotate, do the mapping taking into account this information (Bowtie2, but, kallisto), and then the DE analysis (limma, DESeq2, edgeR, NOISeq) 2. Do the mapping and DE analysis and annotate afterwards

Answer 3

[Kamil]

If you know which transcripts belong to a gene, then you can sum their TPM values to get the gene's TPM.

Consider using Kallisto to get quick and accurate results.

Next, use your preferred method for differential expression analysis: limma, DESeq, edgeR, etc.

[Picasa]

Do you I need annotation for that ? or only transcriptome is enough ?

[Kamil]

Kallisto requires sequencing reads and a file with target sequences.

Please see the kallisto manual for more information.