Omics-defined pathways
Transcriptomic, proteomic, phosphoproteomic, epigenomic, metabolomic, single-cell, or spatial signals define the disease-associated module and the molecules driving it.
Bioinformatics Services
Pathway-Level Target Nomination Service
Prioritize actionable intervention points inside disease-relevant pathways using transparent, context-specific evidence.
Start Your ProjectOverview
Nominate targets by their role in a disease pathway—not by a single score
Pathway-level target nomination asks which molecular intervention points are most likely to modify a disease-relevant program with an acceptable path to experimental testing.
We begin with a defined disease process, pathway module, phenotype, or response program and generate candidate genes or proteins from measured omics data, genetics, curated reactions, regulatory networks, and prior biological knowledge. Candidates are then compared across disease association, pathway position, direction of effect, cell and tissue context, tractability, safety, selectivity, competitive precedent, and validation feasibility. The result is a short, evidence-traceable nomination set rather than a long list of network hubs.
This service supports human projects and model organisms when pathway annotation, orthology, and target evidence are adequate. Inputs may come from patient tissue, blood, single cells, spatial samples, organoids, cell lines, animal models, perturbation screens, or public cohorts. A nominated candidate remains computational and requires independent confirmation; association, network proximity, or a high composite rank does not establish that modulating the target will improve disease.
Data Inputs
Candidate nomination depends on context-rich evidence
We accept FASTQ, BAM/CRAM, VCF, BED, count or normalized matrices, differential-analysis tables, H5AD/RDS single-cell objects, protein or phosphosite intensities, metabolite tables, CRISPR screen results, and curated candidate lists. Required metadata include organism, genome and annotation build, assay platform, tissue or cell type, disease state, control definition, subject or experimental-unit ID, replicate type, batch, treatment, dose, time point, phenotype, and relevant covariates.
Human genetic evidence
GWAS, fine-mapping, QTL, rare variants, somatic alterations, allelic direction, and phenome-wide associations can support causal relevance when mapping and interpretation are defensible.
Functional evidence
Perturbation screens, dependency data, expression response, model phenotypes, ligand–target evidence, and curated experiments help distinguish pathway members from controllable intervention points.
Critical quality risks include pathway annotation bias, ambiguous variant-to-gene mapping, tissue mismatch, cell-composition effects, batch–condition confounding, incomplete interaction networks, overrepresented literature for well-studied genes, and circular reuse of evidence in both candidate generation and evaluation.
Evidence Lenses
Six independent questions shape the nomination
Disease relevance
We assess genetic, molecular, perturbational, clinical, and literature evidence for the exact disease or phenotype, retaining provenance and direction. Evidence from a neighboring disease is contextual rather than equivalent.
Pathway position
Candidates are evaluated as upstream regulators, bottlenecks, feedback controllers, parallel-route nodes, cell-type-specific effectors, or downstream readouts. Centrality alone is not assumed to equal therapeutic leverage.
Tractability and modality
Protein class, location, known ligands, structural information, antibodies, degradation strategies, RNA approaches, gene therapy, and assayability are reviewed against the desired direction and tissue.
Safety and selectivity
Essentiality, normal-tissue expression, paralogs, mouse phenotypes, human loss-of-function observations, adverse genetic associations, pathway pleiotropy, and on-target liabilities inform risk flags.
Precedence and differentiation
Approved drugs, clinical programs, failed mechanisms, patents, competitor intensity, and disease-stage precedent are organized to distinguish validation from crowding and to expose white-space assumptions.
Validation feasibility
Available reagents, measurable proximal and functional readouts, relevant models, rescue strategies, biomarker options, and expected time to falsification influence which candidates should enter experiments first.
Six-Stage Workflow
From pathway definition to nomination gate
| Stage | Key activities | Decision output |
|---|---|---|
| 1. Scope | Define disease, pathway boundary, target product concept, desired modulation, tissue, modality, and exclusion rules. | Nomination charter and acceptance criteria |
| 2. Audit | Verify data, metadata, reference versions, sample design, mapping coverage, batch, confounding, and evidence gaps. | Analysis-ready evidence inventory |
| 3. Candidate generation | Extract drivers, regulators, reactions, genetic links, screen hits, and network neighbors using prespecified rules. | Inclusive candidate universe with provenance |
| 4. Evidence scoring | Evaluate disease link, pathway leverage, direction, tractability, safety, selectivity, precedent, and feasibility. | Dimension-level evidence matrix |
| 5. Robustness review | Test alternative weights, databases, network thresholds, background definitions, missing-data assumptions, and subgroup consistency. | Stable ranks, uncertainty, and sensitivity flags |
| 6. Nomination | Expert review, orthogonal evidence check, shortlist rationale, assay plan, falsification criteria, and backup selection. | Tiered targets and validation roadmap |
Statistical & AI Rigor
Ranking must remain stable, interpretable, and independent of leakage
Omics inputs require adequate independent samples, controls, and biological replication for the intended contrast. Feature-level models should account for batch and measured confounders, report effect sizes and uncertainty, and control multiple testing. When samples are small or groups are inseparable from batch, results are labeled exploratory; an algorithm cannot repair an unidentifiable design.
Composite ranking makes trade-offs visible but can hide uncertainty. We report component evidence, missingness, provenance, and sensitivity to weighting. Knowledge-graph or machine-learning models are used only when they add a defined function. Candidate generation, feature construction, normalization, and tuning remain inside training folds. Grouped or nested cross-validation prevents leakage across related samples, and evaluation on an independent disease cohort or held-out evidence source is preferred. Applicability is bounded by species, tissue, disease stage, assay, network coverage, and evidence date.
Nomination boundary: the final tier is a decision aid, not proof of efficacy or safety. A strong candidate can fail because direction is wrong, the relevant cell type is inaccessible, the pathway adapts, the modality cannot achieve selective exposure, or the model does not represent human disease.
Deliverables
Decision-ready outputs for experimental review
Evidence inventory
Data and metadata audit, pathway definition, databases and versions, mapping rates, exclusions, and evidence gaps.
Candidate universe
All generated candidates with source rule, pathway role, identifiers, direction, and inclusion rationale.
Evidence matrix
Dimension-level scores, source links, positive and negative evidence, missingness, uncertainty, and conflict notes.
Tiered shortlist
Primary, backup, and deprioritized targets with transparent rationale and sensitivity of rank to project weights.
Pathway maps
Candidate locations, regulatory direction, feedback routes, cell context, accessible intervention points, and evidence provenance.
Validation roadmap
Orthogonal assays, perturbation and rescue experiments, proximal and functional readouts, models, milestones, and go/no-go criteria.
Applications
Target decisions across discovery settings
New disease program
Nominate first-pass targets from patient omics, genetics, and curated disease pathways.
Mechanism expansion
Find intervention points upstream, downstream, or parallel to a validated pathway signal.
Resistance strategy
Prioritize compensatory nodes and rational combination hypotheses from response or perturbation data.
Portfolio triage
Compare candidates on evidence maturity, differentiation, modality fit, safety risk, and validation speed.
Scientific Evidence
Integrated evidence supports transparent therapeutic hypothesis building
Network and pathway expansion can connect genetically supported disease genes to additional intervention candidates that occupy the same functional modules. MacNamara and colleagues evaluated how different expansion strategies recover successful drug targets, illustrating both the opportunity and the need to distinguish canonical pathway membership, high-confidence interactions, and broader network propagation.1
How we use public knowledge
Open Targets, Reactome, STRING, UniProt, Ensembl, Human Protein Atlas, ChEMBL, clinical-trial registries, genetic resources, and project-specific evidence can contribute complementary information. Each source has coverage bias, update cycles, identifier conventions, and evidence semantics. We preserve the release and source-level evidence, reconcile conflicts manually when they affect a decision, and avoid counting duplicated observations as independent support.
Network proximity and functional association can reveal plausible pathway neighbors, but edges may represent direct binding, co-expression, text mining, curated knowledge, or prediction. The edge type and confidence must match the mechanistic claim.
1 MacNamara, A.; Nakic, N.; Al Olama, A. A.; et al. Network and pathway expansion of genetic disease associations identifies successful drug targets. Scientific Reports 2020, 10, 20970. https://doi.org/10.1038/s41598-020-77847-9. Distributed under Open Access license CC BY 4.0.
2 Szklarczyk, D.; Nastou, K.; Koutrouli, M.; et al. The STRING database in 2025: protein networks with directionality of regulation. Nucleic Acids Research 2025, 53, D730–D737. https://doi.org/10.1093/nar/gkae1113.
Project Strategy
Design the shortlist around the experiment that can disprove it
We agree the nomination charter before scoring: biological objective, pathway scope, modality constraints, desired direction, tissues, evidence cut-off date, weighting priorities, unacceptable risks, and number of candidates that can be tested. The highest-ranked target is not automatically recommended if its evidence depends on one fragile assumption. We favor a primary candidate, mechanistically distinct backups, and explicit falsification experiments.
Nomination tiers are created after reviewing both supportive and contradictory evidence. Tier 1 candidates should have a coherent disease link, a plausible directional role in the selected pathway, an accessible intervention concept, and a feasible experiment capable of changing the decision. Tier 2 candidates may offer strong biology but carry tractability, safety, context, or evidence gaps. Backup candidates are selected to test a different pathway position or mechanism, rather than merely repeating close paralogs. Deprioritized targets remain in the evidence matrix with the reason for exclusion, allowing the decision to be revisited when a new dataset, modality, structure, clinical result, or safety observation becomes available.
High-value nominations should be checked in an independent cohort where available and validated by orthogonal expression or protein assays, genetic or pharmacological perturbation, dose response, pathway-proximal readouts, functional phenotype, and rescue or reversal. A useful first experiment should verify target engagement, confirm the expected direction of pathway change, and measure a disease-relevant phenotype in the same model. Where possible, a rescue experiment or second perturbation modality helps distinguish on-target biology from reagent-specific effects. For guidance on a pathway, disease program, internal dataset, or experimental shortlist, please Contact Us or submit the Online Inquiry below.
Related Services
Connected Bioinformatics Services
Online Inquiry
Submit your project details below, and our team will respond within 24 hours.
x
Need help getting the data you need?
Talk to our technical team about your project!
I Want To Talk