best workflow for RNA data analysis without replications

Question

[Chao.wang2]

I am wondering could anyone tell me the best work flow (software) and Statistical model should I use for RNA seq data (without replications).

Thanks a lot

[Rob]

What are you trying to do with the RNA-seq data? If you have no replicates, you should not attempt to assess differential expression.

[Chao.wang2]

Thanks for your reply, actually, I want to assess the differential expression. If I can't assess differential exprssion, what else can I do with this kind of data.

Thanks

[Devon Ryan]

Not much, you can't really do much of the way of comparisons. You could ask whether a gene is likely expressed, or what its major isoform is, but even then any result would have the "at least in this sample" caveat applied to it.

[mkulecka]

edgeR authors provide several solutions for people who work without replicates: https://bioconductor.org/packages/release/bioc/vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf (page 21). However, I agree with previous comment that you should avoid working without replicates.

[Chao.wang2]

thank you for your advice.

[silas008]

You can analyse this data by fold change between the groups. But the data should be normalized. Is not the best way to avaluate rnaseq, but as you dont have replicates you can validate some of most down or up regulated rnas in your data by Real Time PCR, if is possible.

Answer 1

[Devon Ryan]

The best tool you'll find is GFold, and you should obviously be very cognizant of the caveats that come with trying to draw any firm conclusions from unreplicated data (i.e., the most this is useful for is getting some target genes for replicated targeted tests).

[Chao.wang2]

Thanks, I will check that.