It depends on what you mean by "best." Primarily all packages out there rely on some form of enrichment analysis to score pathways, but that method relies on the assumption that all genes in pathways are independent - but we know that's a false assumption because the pathway is there specifically to describe how all the genes interact with one another.
If you are dead set on using R for your own analysis, the package we use is called ROntoTools, developed by our lead bioinformatician. In addition to enrichment, it also takes the interactions between genes into account and calculates the perturbation of the pathway.
However, if you just want to analyze your data, we have this implemented in our application called iPathwayGuide. Not only does it do the above, but it also highlights sections of the pathway where the data and the pathway relationships are coherent which serves as a putative mechanism for that pathway. We also allow you to model drugs, miRNAs, and SNPs directly on top of the expression data on the pathways. iPathwayGuide will take you gene-expression data and provide the following analyses:
- DE Genes
- Predicted miRNAs
- GO Analysis (with advanced correction factors)
- Pathway Analysis using Impact Analysis (Perturbation & Enrichment)
- Diseases
Plus all analyses come with a printable summary with full methods and reference sections. It's completely free to try. Each analysis is available for 3-days to preview and see if you want to keep the results. If you like it, you can purchase that specific report. We also offer unlimited subscriptions.
Here is a sample of a printable summary.
Do you now have the answer to your own question?
Not a complete answer.