Mechanistic patient stratification across transcriptomics, proteomics and metabolomics simultaneously. Where MecStra scores pathways from gene expression alone, MultiO draws on all available omics layers at once — signal is strongest where all three agree.
MecStra MultiO processes all omics layers together through a single shared biological network. Signal is strongest where all three layers agree.
Upload any combination of omics datasets from the same patients. MultiO handles normalisation, sample matching and pathway scoring across all layers simultaneously.
RNA-seq gene expression matrix. Accepts GAS output directly or any normalised counts/TPM file.
LC-MS or NMR peak intensity table. HMDB or ChEBI identifiers. Missing values handled automatically.
MaxQuant, DIA-NN or Olink NPX output. UniProt accessions matched to pathway protein nodes.
Built on TWPS-MO — Topology-Weighted Pathway Scoring, Multi-Omics — an extension of the validated MecStra algorithm. The core network propagation and NMF stratification are identical; the innovation is in how pathway nodes are seeded.
Each omics layer is log-transformed, median-centred and rank-normalised independently, then aligned to a common scale. Sample IDs are matched across layers automatically.
WikiPathways annotates each pathway with typed nodes — GeneProduct, Protein, Metabolite. Each measured entity seeds its correct node type in a single unified network graph.
Signed personalised PageRank propagates signal across all node types simultaneously. Centrality-weighted aggregation, concordance amplification, and NMF stratification produce mechanistic patient subtypes.
Not a post-hoc integration of separate analyses. A single unified scoring framework where all omics layers contribute simultaneously.
When gene expression, protein abundance and metabolite levels all point in the same direction within a pathway, the score is amplified. Independent corroboration from multiple measurement modalities is treated as stronger biological evidence.
A metabolomics layer covering only a subset of pathway metabolites is not penalised — it contributes proportionally to its coverage. Unmeasured node types seed at zero and still participate in network propagation.
When transcript is up but protein is down, MultiO flags the pathway rather than cancelling the signals. Discordant pathways often reveal post-translational regulation, enzyme inhibition or feedback loops — biologically meaningful findings.
The NMF stratification, cluster assignment, results viewer and publication-ready figure export are identical to MecStra. If you already use MecStra, MultiO produces results in the same format.
MultiO is being validated on the U-BIOPRED severe asthma cohort — one of the most comprehensively phenotyped multi-omics respiratory datasets available.
The U-BIOPRED consortium assembled matched transcriptomics, metabolomics and proteomics from the same severe asthma patients across multiple European centres. MultiO v1 is being validated on matched transcriptomics and metabolomics from this cohort, with a publication in preparation.
MecStra (single-omics) achieved 82.93% vs 0% treatment response rate in independent validation on the SoMOSA omalizumab cohort (p=0.002, n=45). MultiO extends this validated foundation to multi-omics inputs.
MultiO is a separate tool at a higher licence tier. MecStra remains the validated single-omics choice.
MecStra MultiO is available to Founding Partner Labs and early-access enterprise customers ahead of general release. Contact us to discuss your dataset and use case.