filter genes based on statistical test

Question

[elhamdallalbashi]

hi all, I have datasets derived from RNAseq experiments (each experiment is the effect of one drug on a cancer) ,I did differential expression.for identifying differential expression,I had 3 statistical test: t-test,DGE and Beggerly test.in next step I should filter some genes by high fold change and low FDR p-value (I used Fold change ≥ 1.5 FDR P-VALUE ≤ .05 )

But the problem starts,for one experiment t-test gives the appropriate number of genes, in another one, Beggerly test gives a suitable number of genes and t-test doesn't answer ,also in some experiments there are lots of genes even with these filters or in each 3 test I have a lot of genes.

what should I do?should I use different type of test for each experiment and On what basis should I choose statistical test؟

,

Answer 1

[Devon Ryan]

Since you're using a commercial product you should call the manufacturer for help.

Anyway, the appropriate answer is to use "DGE" for everything, since the other tests should be removed from the software.

Answer 2

[Farbod]

Dear Elham, Hi

Why you did not use your expression results (e.g TMM) and use it's matrix as input in some R package for DEG analysis (e.g DESeq2 or edgeR) and just select your threshold according to FDR and then log FC ?

Then you will report your DEG according to the method that the selected package hes described. ;-)

~ Best

[elhamdallalbashi]

dear Farbod,thank you for a replay.because of my supervisor!I should use CLC genomics for my thesis

[Devon Ryan]

Sounds like a good time to start ignoring your advisor.