Best statistical test for 1 vs 2 replicates

Question

[from the mountains]

Hello all,

I have to do a comparison of two groups. The first group has 1 sample and the second group has 2 samples. What is the most appropriate statistical test/software for this comparison? The resources that are most readily available to me are:

1. DeSeq2
2. Baggerly's test (test on proportions
3. Edge (Edge R basically)
4. I have a Gaussian model but I have never tried it out before.

I know 1 vs 2 is not ideal, but it's the challenge I've been given. I reason that DeSeq2 needs ideally 3 replicates per group and 2 replicates per group is pushing it. Not sure which of the rest is most appropriate.

Help would be very much appreciated. Thanks!

[Michael]

There is no good solution for such a low number of/no replicates. Which method is appropriate depends on the source of your data. I am guessing RNA-seq according to the tag. You can try both edgeR and Deseq but you risk having no significant genes.

[from the mountains]

thanks! yes, it is RNA-seq.

Answer 1

[h.mon]

You may try GFOLD or NoiSeq, both were designed for few or no replicates.

[from the mountains]

thank you! i will have to put that in my repertoire.