Can I use t-test to analyze RNA-Seq if replicas is large?

[ftian]

Dear all,

I am new guy to RNA-Seq. I want to use a software (originally developed for microarray data) to analyze TCGA breast cancer RNA-seq data which is based on t-test. After read some materials, it seems that RNA-seq data is not appropriate to be analyzed by T-test. Is this right?

I also find that some people say that if replicas are large, t-test can still be used. Is this right? If this right, are some special processing required (such as normalization) required?

Thanks you all.

[dpryan]

Using a T-test is generally not a great idea. If you have a large number of replicates give SAMseq a try.

[ftian]

Does this mean that all the methods (originally developed for Microarray) cannot be applied to RNA-Seq data?

[dpryan]

Not directly, but see the voom() function that now comes with limma. Realistically, you could do T-tests after transformation with voom() (limma does something similar, but it allows for information sharing between genes, which isn't needed with large numbers of samples).

[barkasn]

Hi ftian,

Consider using software that has been written with RNA-seq in mind. The tuxedo suite is a good option my my view but other solutions exist too.

[Jeremy]

Some other commonly used RNA-seq tools are cufflinks and DESeq2 (in R). Check out the
Seqanswers RNA-seq Wiki

[ftian]

Thank all of you very much. I will look at those tools for RNA-Seq.

[Dario1984]

You can use a t-test because of the central limit theorem. The special tools for microarrays and RNA-seq were made for small sample analysis projects.