edgeR p-value and logFC

[alessandra85]

Dear all,

I am kind of new at RNAseq analysis. I hope someone can help me? I got like 880 genes with a p-value <0.01, using edgeR. I got the table with those 880 genes, and I separate the genes by up (+) and down (-) regulated by the LogFC option. But then I found that some of the 880 genes had a p value > 0.01 like 2,01E+05
Is there any explanation for this? What is the best way to get the DE genes, by p-value or logFC?

Thanks,

[macrowave]

You may filter the edge$table object within R using subset functions, I'm sure you'll get the right results. Usually, as a rule of thumb, I use P<=0.05 and logFC>=1 or logFC<=-1, in other words, fold change larger than two, and still statistically significant.

[alessandra85]

Thanks for your reply!! I did it again within R and I got the rigth results. But now I got another question. I have a gene that is kind of important for me:

contig03563 -11.329612 1.0838975 1.168637e-03 4.800752e-02

contig03563 41 47

Based on the logFC = 1.08 is up regulated, but if I saw the raw data is down? and this happens with some others too. is ok to look ath the raw data? Could it be because of the small logFC?
is this still statistically significant? Can I believe in this results? I am really confused with the logFC meaning.

I'll really appreciate any help!!

[macrowave]

Looks like you don't have replicates. The difference between raw counts and edgeR might be from the normalization you did in the edgeR process. So if no replicates, I wouldn't trust the results. For me, I can't say anything if there is no replicates for 41 and 47.