Skip to content

MOSClip

Multi Omics Survival Clip — a topological pathway analysis tool that integrates multi-omics data to identify survival-associated modules and significant modules for two-class analysis, providing both pathway tests and module tests.

Talk at EuroBioC2025

  • Speaker: Anna C E De Lima Tanada (presented; Paolo Martini is maintainer / senior author)
  • Day / session: Day 1 (Wed Sep 17, 2025) — afternoon short-talk session, after the Holmes keynote
  • Talk title: "MOSClip for multi-omics pathway analysis"
  • Slides (Zenodo DOI): to verify after publish
  • Video: not recorded — short talks at EuroBioC2025 were not part of the plenary YouTube playlist
  • Abstract / schedule: eurobioc2025 schedule

What it does

MOSClip overlays multi-omics measurements onto pathway topology and tests pathway modules for association with survival or two-class outcomes. It returns both a pathway-level test and a finer-grained module-level test, so users can localize which part of a pathway carries the signal rather than only learning that the pathway is hit.

How it works

Core idea. Pathways are represented as graphs via graphite/graphNEL; within each pathway the package extracts connected-component "modules," reduces the multi-omics measurements over those modules to a small number of covariates, and tests those covariates jointly against outcome — Cox proportional hazards for survival, a generalized linear model for two-class comparisons.

Inputs / outputs. Input is an Omics object (a MultiAssayExperiment subclass) holding any combination of expression, methylation, mutations, and CNV, plus a pathway list (e.g. Reactome). Output is a per-module table of p-values with the contributing omic-covariates labelled (e.g. expPC1-3, met2k2, cnvPOS, mut).

Key innovation. Topology-aware multi-omics survival testing at module resolution: rather than collapsing a whole pathway to one score, MOSClip localizes the signal to a connected sub-graph and attributes it to specific omics layers.

Parameters worth knowing. - Reduction method per omic — summarizeWithPca (expression), summarizeInCluster (methylation CpG clusters), summarizeToBinaryEvents / summarizeToNumberOfEvents (mutations, CNV). - useTheseGenes — pre-filter genes considered for module construction. - specificArgs — per-omic arguments (e.g. methylation gene dictionary). - seed — reproducibility for the dimensionality-reduction steps.

Canonical example. The vignette runs multiOmicsSurvivalModuleTest() over reactSmall (three Reactome pathways) on multiOmics, a 50-sample ovarian-cancer TCGA cut, then aggregates results with multiPathwayModuleReport(). The output is a top-modules table ranked by Cox p-value, where the contributing covariates (expPC1–3, met2k2, cnvPOS, mut) make explicit which omics drove the association.

Where it fits in the corpus

  • AACR 2026: no current dossier; relevant axis is clinical-trials / survival modeling — MOSClip's explicit survival framing maps onto outcome-prediction talks
  • MIRit (entry) — sibling network/pathway method on the same Day 1 short-talk session
  • GBCC 2025: queued
  • Nextflow Summit 2026: not mentioned

Notes

Multi-omics with explicit survival modeling — distinct from purely descriptive multi-omics methods. The module-level test is the differentiator: it pushes interpretation below the pathway aggregate.