Study of similarity (contrary to differential expression analysis)

Question

[jgarces]

Hi there,

I wanna to compare two sets of samples in order to see how similar they are. The classical approach, for example, is a correlation or distance analysis: dendrograms, correlation heatmaps, etc.

But I was wondering if it exists some approach which does something contrary to classical differential analysis... I mean, instead of looking for statistically "divergent" genes between two conditions, search for genes that are very similar (but with a statistical background). Just for curiosity.

For example, what if we take those genes not differentially expressed according DESeq2? They mean that are similar between both conditions??

Thanks.

Answer 1

[i.sudbery]

Yes. DESeq2 allows you to specify a variety of alternative hypotheses. One such alternative hypothesis is that the abs(log2FoldChange) < x.

See https://www.bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#tests-of-log2-fold-change-above-or-below-a-threshold

[jgarces]

Nice! I didn't realize this section, thanks!