salmon for gene expression quantification
Are the results of salmon reliable for gene expression quantification? Because it just gives gene expression in terms of TPM and Number of Reads by just one command taking input as fastq files.
• 2,013 views
•
link
1 answer
Yes, it's reliable. Just because a workflow is more complicated doesn't mean that that workflow is "better". The tuxedo suite was used in 2010 as the primary workflow of rnaseq analysis -- it involves many steps/tools and lots of time+memory but it has been deprecated because it has actually been shown to be less accurate than newer one-step approaches.
• 0 views
•
link
Log in to answer this question.
Isn't that how tools are supposed to work?
Read the paper and understand the algorithm. This is the first I've see someone judge a tool adversely by the number of commands it takes to get the job done.
I am not judging the tool adversely. Its just that I was a bit skeptical that just by one command i am getting the gene expression and read counts. If you look at other methods like implementing whole RNA Seq pipeline right from alignment to differential expression, it takes two or three tools to get the result which salmon is doing by using just one tool or even just one command.
STAR can get you from FASTQ to GeneCounts in one command, and so can rsem-calculate-expression. You can detour using any tool but the expectation that a multi-step pipeline is somehow better than a single-step tool is ridiculous - the number of steps is no way to measure anything except the number of steps.