Skip to content

RCX (RCX2 update)

Cytoscape Exchange (CX) format for R — creates, validates, visualizes, and converts biological networks in the Cytoscape CX format to standard R graph objects (igraph, graphNEL). The EuroBioC2025 talk covered the package's v2 (RCX2) updates.

Talk at EuroBioC2025

  • Speaker: Florian Auer (University of Augsburg)
  • Day / session: Day 1 (Wed Sep 17, 2025) — morning short-talk session, after the Crowell keynote
  • Talk title: "RCX2 package for biological network exchange formats"
  • Slides (Zenodo DOI): to verify after publish
  • Video: not recorded
  • Abstract / schedule: eurobioc2025 schedule

What it does

RCX is the canonical R reader/writer for the Cytoscape Exchange (CX) network format used by NDEx and Cytoscape itself. It validates network structure, round-trips to igraph/graphNEL for downstream analysis, and provides a visualization layer. The "RCX2" talk title points to a v2 upgrade — the released Bioconductor package is named RCX (currently v1.16.0); the v2 changes likely shipped under the same package name.

How it works

Core idea. The CX format is a JSON document split into aspect-modules (nodes, edges, nodeAttributes, edgeAttributes, networkAttributes, cartesianLayout, cyVisualProperties, metaData), each with its own schema and cross-references by integer IDs. RCX mirrors that as an R RCX object whose slots are aspects, and validates referential integrity (every edge's source/target must resolve to a node ID, attribute aspects reference valid node/edge IDs, etc.) on import.

Inputs / outputs. Input: a CX JSON file (from NDEx, Cytoscape, or another producer). Output: an RCX R object; round-trippable to igraph via toIgraph() and to Bioconductor's graphNEL for downstream graph algorithms; serializable back to CX with writeCX().

Key innovation. Native R support for the full aspect model — including Cytoscape visual properties and metaData — so a CX file edited in R is still a valid CX file when opened in Cytoscape or pushed to NDEx. Earlier R network packages strip presentation when converting to igraph; RCX preserves it.

Parameters worth knowing. readCX(file) for full-file read; the lower-level readJSON / parseJSON / processCX triple for incremental parsing when a CX file has deprecated or unknown aspects. toIgraph(rcx) for graph-library export. $metaData accessor to inventory which aspects an object actually carries.

Canonical example. Vignette walks through the bundled "Imatinib-Inhibition-of-BCR-ABL" pathway (an NDEx network): load with readCX(), inspect aspects with summary() and $metaData, modify node or edge attributes, then write back with writeCX() for re-upload to NDEx.

Where it fits in the corpus

  • AACR 2026: no current dossier; relevant axis is bioinfo / AI methods (network-based analyses, gene-regulatory inference)
  • GBCC 2025: queued
  • Nextflow Summit 2026: not mentioned

Notes

Naming oddity: talk announced as "RCX2" but the released Bioconductor package is RCX. Verify after slides land whether RCX2 was a project nickname for the v2 update or a separate fork.