DESEQ2 analysis - PCA plot

Question

[shome]

I am not getting a lot of genes from my DESEQ2 analysis hence I was checking the PCA plot and this is how it looks like.Any suggestions how to mitigate this ?

[Ram]

I am not getting a lot of genes from my DESEQ2 analysis

And why is that a problem?

[shome]

I think this is because there is a lot of variation in the samples so the markers I am interested in are not coming out as statistically significant.What do you think i should be doing ?

[Kevin Blighe]

Which comparisons are you doing? Why are you showing two different bi-plots?

[bkleiboeker]

looks like you might have some batch effect. idk exactly what your label means, but if 697-5_CTRL and 697-5_LPS come from one biological replicate (maybe cells derived from a single mouse/rat?), 697-2_CTRL and 697-2_LPS from another biological replicate, etc.. then it could be worth passing the info for those biological replicates into your design matrix, as 697-5 and 697-2 do seem pretty different from the rest in your first PCA

Answer 1

[swbarnes2]

The basic conclusion to draw from this is that LPS isn't doing much in these experiments. There's some other source of variation (maybe batch effect, maybe just natural variability) that is changing more genes more strongly than LPS.

I thought that LPS was basically a sledgehammer, but maybe not in this tissue, or experimental setting. There isn't much you can do computationally to magically fix it.