Entering edit mode
7.6 years ago
weixiaokuan
▴
140
Hi, I have a project using RNA-Seq but there are no replicates for each condition. I am wondering what method I should use for such a situation. Especially, how I should normalize the data? Apparently, I cannot use limma-voom pipeline. It seems I may be able to use DESeq2. But I am wondering if it's possible to directly compare TPM for each condition using Kallisto as the reads somehow normalized to TPM scale. I haven't seen such discussions using Kallisto before but I would like to know if you have any thoughts on this. Thank you.
-Xiaokuan
This is not an answer, but in general RNA-seq without at least two replicates isn't very reliable.
I know. But I want to have some measurement anyway, especially after some normalization/scaling. Not all the experiments have full replicates.
What's your goal? Without replicates you can't do any of the most common comparisons.
to identify some trend of differentially expressed genes then we may perform further experiments to validate.
You can try GFOLD, that's about it. In the future though you'd find it's better to spend a bit more for replicates to save a LOT later.
totally understood. However, a lot of times, things are not controlled by us and sometimes the samples are just not enough. So I think a method for no replicates is very helpful. Thank you for all the help. -Xiaokuan
Helpful, maybe. Statistically sound and reliable, no.
by they way, will TPM from Kallisto help in this situation?
No, nothing can help any more in this situation, it's impossible.
Maybe I didn't make it very clear. I know it's no way to attain statistical evaluation. But I am wondering if there is any way to scale the data such as by lib sizes. I am thinking TPM from kallisto is one way to do it. -Xiaokuan
Any of the standard methods will work there, though TPM is not exactly my preferred method (for the same reason that FPKMs are problematic).