ResA3: A Web Tool for Resampling Analysis of Arbitrary Annotations

Aaron Ruhs,Franz Cemic,Marcus Krüger,Thomas Braun,Maureen J Donlin

doi:10.1371/journal.pone.0053743

Abstract

Resampling algorithms provide an empirical, non-parametric approach to determine the statistical significance of annotations in different experimental settings. ResA3 (Resampling Analysis of Arbitrary Annotations, short: ResA) is a novel tool to facilitate the analysis of enrichment and regulation of annotations deposited in various online resources such as KEGG, Gene Ontology and Pfam or any kind of classification. Results are presented in readily accessible navigable table views together with relevant information for statistical inference. The tool is able to analyze multiple types of annotations in a single run and includes a Gene Ontology annotation feature. We successfully tested ResA using a dataset obtained by measuring incorporation rates of stable isotopes into proteins in intact animals. ResA complements existing tools and will help to evaluate the increasing number of large-scale transcriptomics and proteomics datasets (resa.mpi-bn.mpg.de).

Highlights

Gene and protein annotations like Gene Ontology, KEGG or Pfam provide a systematic approach to classify protein function and localization
The estimator statistic is evaluated for each sample and stored in the empirical resampling distribution (RD). 4
Data upload, and optional Gene Ontology annotation the resampling analysis on the chosen estimator is applied to each sample defined by the annotations provided

Summary

Introduction

Gene and protein annotations like Gene Ontology, KEGG or Pfam provide a systematic approach to classify protein function and localization. Tools like GOrilla, GoMiner and Catmap [1,2,3] use separate target and background sets to calculate enriched GO terms, or use a ranked gene list without experimental values. We present ResA, a universal web tool designed to determine the statistical significance of sample distributions defined by annotation in genomic and proteomic data sets.

Results

Conclusion