Enrichment of disease ontology enrichment

[Orr Shomroni]

I have recently come across a R package that allows users to check for genes in the disease ontology (DO.db), and I thought it might be interesting to run an enrichment test on this ontology.

Currently, I have come across 2 R packages that support this option:
(1) GeneAnswers package
(2) DOSE package

Apperently, GeneAnswers allows the usage of DOLite (a more condensced ontology renaming categories as groups) but not the usage of DO, and DOSE enrichment command allows both options. Furthermore, it seems both DOSE and GeneAnswers are using hypergeometric tests. Nevertheless, when I run both commands for DOLite with the same p-adjustment method (FDR/Benjamini) and same p-value threshold (0.05) for the same genes, I am getting quite a difference in results.

My questions are:
(1) Is anyone here familiar with GeneAnswers and DOSE packages enough to know how they differ technically?
(2) Is any of the 2 packages considered generally better than the other?
(3) Are there any other good tools (in any programming language) that can be used to analyse the enrichment of disease ontology from gene names?

[ygc_hku]

please refer to : http://bioinformatics.oxfordjournals...bstract/btu684

DOSE supports DOLite (which was not updated) and DO (recently updated) with hypergeometric test and Gene Set Enrichment Analysis.

[ygc_hku]

For comparing GeneAnswers and DOSE (internal algorithm is same as in clusterProfiler), please refer to: http://ygc.name/2014/08/07/why-clusterprofiler-fails/

I have recently come across a R package that allows users to check for genes in the disease ontology (DO.db), and I thought it might be interesting to run an enrichment test on this ontology.

Currently, I have come across 2 R packages that support this option:
(1) GeneAnswers package
(2) DOSE package

Apperently, GeneAnswers allows the usage of DOLite (a more condensced ontology renaming categories as groups) but not the usage of DO, and DOSE enrichment command allows both options. Furthermore, it seems both DOSE and GeneAnswers are using hypergeometric tests. Nevertheless, when I run both commands for DOLite with the same p-adjustment method (FDR/Benjamini) and same p-value threshold (0.05) for the same genes, I am getting quite a difference in results.

My questions are:
(1) Is anyone here familiar with GeneAnswers and DOSE packages enough to know how they differ technically?
(2) Is any of the 2 packages considered generally better than the other?
(3) Are there any other good tools (in any programming language) that can be used to analyse the enrichment of disease ontology from gene names?