The Reactome project provides a suite of tools for our users to perform pathway- and network-based data analysis via its web site and ReactomeFIViz, the functional interaction (FI) network-based Cytoscape app.

The Reactome FI network is built upon curated pathways in Reactome, supplemented with curated pathways from other pathway databases such as KEGG PATHWAY and PANTHER pathways. These curated pathway interactions are then integrated with gene-gene interactions derived from curated and high-throughput sources using machine learning to generate a reliable set of high-probability functional interactions.

Recently, we have released the 2018 version of the Reactome FI network. To construct this FI network, we improved the training of the Naive Bayesian Classifier, which resulted in an increase in the recall rate. Consequently, the SwissProt coverage was increased from 12,441 to 13,469 proteins (8.3% increase) and the total number of functional interactions among genes increased from 241,338 to 262,321 (8.7% increase). We attributed this to an increase in the number of predicted FIs due to the improved recall rate.

In addition, the ReactomeFIViz app (version 7.2.0) now supports the 2018 version of the FI network and contains updated Reactome Pathways dataset (Release 67). The new ReactomeFIViz app is available to download within the Cytoscape software, or directly from the Cytoscape app Store. The User Guide provides instructions on using the Reactome FI network and ReactomeFIViz app.