ReactomeGSA - Multi-omics, quantitative, comparative pathway analyses
ReactomeGSA is an analysis service of the Reactome Knowledgebase (https://reactome.org/gsa). Its key feature is that it provides quantitative pathway analysis methods. The most commonly used pathway analysis methods, such as GO enrichment analysis, are so-called “overrepresentation analysis”. These just take a list of gene / protein identifiers as input to test, which pathways are more frequently found. The big disadvantage of these methods is that they ignore the underlying quantitative data which reduces their statistical power. Additionally, users have to perform separate analysis for up- and down-regulated genes / proteins and the result does not contain the strength (in terms of fold-change) of the observed change.
ReactomeGSA supports quantitative pathway analyses of several types of ‘omics data, including transcriptomics, proteomics, and microarray. It additionally is able to analyze several datasets simultaneously. Thereby, users can quickly perform comparative pathway analyses between unrelated studies - which is even possible across different species and different ‘omics technologies. For these analyses, pathway-based methods have the advantage that the pathways can be used as a conserved, common ground.
Our group is in-charge of developing the backend system - a Kubernetes-based application hosted on Google Cloud - as well as the ReactomeGSA R Bioconductor package. The frontend application is developed by EMBL-EBI.
Most recently, we extended ReactomeGSA to enable users to directly load public datasets from two major resources, EMBL-EBI’s ExpressionAtlas and NCBI’s GREIN resource. Thereby, it is possible to quickly integrate or re-analyse a huge amount of public datasets.
We are continuously working on improving ReactomeGSA. This includes developing new features and tools that help users in the analysis of their data. Furthermore, we are actively researching pathway analysis algorithms to extend them to more ‘omics technologies or improve their accuracy.
Links
ReactomeGSA web interface: https://reactome.org/gsa
Reactome GSA R/Bioconductor package: