miRNA and DESeq

[unique379]

Hello all,

Just a quick answer by experts
I am dealing with miRNA-seq data and having some doubt.
1) Filtering by resSig <- res[ res$padj < .1, ]
I am not able to get any significant miRNAs on my sample while by using
resSig <- res[ res$pval < 0.05, ]
i am getting about 41 DE miRNAs (up and down)…

what should i do ??? may i rigid with only pval makes sense ??? I mean "i just keep going only filtering by pval < 0.05, is that enough to consider my genes significantly differentially expressed? "

2) after this analysis may go further for filtering by overall counts, plotting Heat map and PCA plots would be good results ??? This makes any significant sense ?? I am asking because, i am not very expert in statistics.

Thanks, any help would be appreciated

[dpryan]

Cross-posted on biostars. As I mentioned there:

I wouldn't use unadjusted p-values unless really desperate (they're less likely to validate). If you haven't done so already, try doing some independent filtering to increase statistical power.

[unique379]

Cross-posted on biostars. As I mentioned there:

Thanks for your suggestion but could you please explore your advice. As i am not so expert of R

[dpryan]

I've replied over on biostars since the thread there is getting a bit more traffic.