is quality evaluation of transcriptome assembly still required if my workflow is (good quality reads (fastp) -> bowtie2 alignment to reference genome -> rsem expression calculation)?

Question

[mrj]

I am working on a RNA-seq analysis of mice. My workflow is as follows.

1. quality filtering of reads
2. using bowtie2-build to index my reference genome
3. performing rsem-prepare-reference with gtf file for mouse genome (Mus_musculus.GRCm39.dna_sm.toplevel.fa)
4. aliging NGS reads to reference genome using bowtie2.
5. using rsem-calculate-expression for obtaining expected frequencies of genes.

I have been reading about Trinity. It suggests evaluation of the quality of transcriptome assembly. I am not using trinity, I am using rsem. With rsem and the reference genome, do I still need to perform the quality of assembly at any point Do I need to evaluate the quality of assembly at any point? If the answer is yes, can you explain? if answer is no, can you explain?

Please, excuse me if I am asking you an obvious question. I have zero experience on this.

Answer 1

[GenoMax]

You are not doing transcriptome assembly in your workflow. You are using a known set of genes from mouse genome and aligning your data to that transcriptome using bowtie2.

trinity on the other hand assembles a transcriptome starting from raw sequence data. This is generally done if you have no genome sequence available or if you are interested in discovering novel transcripts.

Since you are working with a well characterized genome (which has a corresponding transcriptome) you do not necessarily need to do de novo transcript assembly. In that sense, your current workflow is ok.

You may want to also look at a tool like salmon (LINK) which does transcript abundance estimation and require much smaller amount of resources.

[mrj]

GenoMax, Thank you for your very clear answer. Now, I understand the difference very well.