RNA-seq, STAR count, GSEA

Question

[Rob]

Hi friends

I have about 300 differentially expressed genes for my data of STAR count. This is for two groups: treatment & control. However, for the same data consisting of all protein-coding genes, I don't get any gene set enriched significantly at FDR < 25%.

I am doing this with GSEA software for the hallmark of 50 gene sets. The parameters are in default. permutation type: phenotype.

Does anybody have any thoughts on why is this?

Is there anything that I am not doing correctly?

[jv]

Is there anything that I am not doing correctly

We have very little information about how you actually did your analysis and it is very difficult to know how to help.

Did you use GSEA-Prerank and if so, how did you rank your genes? Else, did you use normalized counts for "standard GSEA" and what form of normalization did you use?

[Rob]

Hello Thank you for responding. I didn't pre-rank my genes, I don't use GSEA-prerank.

I normalized my count data using the DESeq2 workflow. What is normalization for "standard GSEA"? Is there a different type of normalization method for GSEA?

dds <- DESeqDataSetFromMatrix(countData = rawdata,
                              colData = metadata,
                              design = ~ 1)

normalized_count <- log2( counts(dds, normalized=TRUE) + 1 )