i want to identify and retrieve all DEGs from the data and set the p value<0.05 and fdr<0.01 how can i set the values for p and fdr value using R on linux suse platform? i am using the code given below but it is not showing FDR value coloumn in result and i dnt knw how to set the threshold of fdr which i want to set 0.01 .
> gse <- gse[[1]]
> fData(gse) <- fData(gse)[,c(1,2,10:12)]
> grp <- factor(sapply(strsplit(as.character(pData(gse)[,1]), " "), "[", 1))
> grp
[1] A_set A_set AS_set AS_set AS_set F_set F_set S_set S_set S_set
Levels: A_set AS_set F_set S_set
> design <- model.matrix(~grp)
> fit <- lmFit(gse, design)
> fit2 <- eBayes(fit)
**> tt<- topTable(fit2, coef = 2, adjust = "fdr",n = 50)**
> UP <- subset(tt, tt$logFC > 0)
> DOWN <- subset(tt, tt$logFC < 0)
> head(DOWN)
ID GB_ACC
210004_at 210004_at AF035776
236646_at 236646_at BE301029
1560477_a_at 1560477_a_at AK054643
Gene.Title
210004_at oxidized low density lipoprotein (lectin-like) receptor 1
236646_at transmembrane protein 52B
1560477_a_at sterile alpha motif domain containing 11
Gene.Symbol ENTREZ_GENE_ID logFC AveExpr t
210004_at OLR1 4973 -2.495755 3.516644 -14.90119
236646_at TMEM52B 120939 -1.991936 3.200324 -12.37906
1560477_a_at SAMD11 148398 -1.385055 5.790050 -11.81502
P.Value adj.P.Val B
210004_at 3.134046e-07 0.01713540 3.500307
236646_at 1.353597e-06 0.03551374 3.055998
1560477_a_at 1.948628e-06 0.03551374 2.926959
2 answers
In limma, you make use of adjusted p-values. By default it is using FDR's (Benjamini Hochberg's) method for adjustment. This is not the same as FDR itself. Please read what Gordon Smyth says about it here:
I reckon OP wants to use the adjust.p filters. If you want to use the real FDR method then b.nota is correct. You will have to intact select genes from tt object you have based on logFC and avgexp and then recalculate the FDR of those genes against the background. Then you will be able to have a new column with FDR values
This helped me http://stackoverflow.com/questions/4935479/how-to-combine-multiple-conditions-to-subset-a-data-frame-using-or
If you're only interested in the AS_set result the adjusted P-value is also known as the Q-value.
degs.fdr<- subset(tt, tt$adj.P.Val < 0.05)
This is how you set, here adjust.P.Val is what you need. Now if you want to regenerate again new FDR values then its different
Log in to answer this question.
I changed format for clarity.
You don't need FDR if you use adj.P.val (P-value adjusted for multiple testing). It's not the same as FDR, but it has the same purpose. It's quite equivalent in usage.
It's simple just use order function of subset the data from of top table with for less than 0.01 and then select degs based on direction. Your selection of degs should be first on for threshold followed by logFC.