Applications

Data and models
for predictive biology.

Three connected AIVC partnership routes for drug research, followed by our clinical diagnostics and laboratory products.

AIVC for drug R&D

From target discovery
to clinical research.

Our AIVC strategy is designed to support the full drug-development workflow, from target discovery and POC/PCC to preclinical animal studies and clinical trials. We work with partners through three connected routes.

Inside the AIVC platform

01 / DATA

Customized
datasets

Dataset co-development

Purpose-designed multi-omics datasets led by controlled perturbation proteomics. Match the biological system, intervention, dose, time and endpoint to your research question.

Standardized protein profiles, controls, QC and experimental metadata.
Relevant RNA, metabolic, lipid or imaging measurements, with matched functional readouts.
Dry–wet closed-loop learning to guide the next experiment and refine the model.

Project output

A documented, quality-controlled dataset for a defined biological question.

02 / PREDICTION

Predictive
vertical models

Scenario-specific co-development

Fine-tune and evaluate a drug-prediction model for a specific biological context and research decision.

Drug-efficacy and combination predictions, grounded in protein-response biology.
Toxicology co-development using dedicated safety datasets and independent validation.
Held-out evaluation and experimental confirmation of selected predictions.

Project output

A task-specific model with an evaluation report, defined scope and integration plan.

03 / FOUNDATION

Foundation
model

Oncology-first · In development

Develop reusable cellular representations grounded in protein-response biology, beginning with oncology.

A proteomics-rich foundation for adaptation to new drug-research tasks.
Expansion to immune and inflammatory systems, other diseases and chassis organisms such as yeast, supported by new data and evaluation.
Future partner access and qualified local deployment, with scope and data governance agreed per project.

Development direction

A reusable oncology foundation, progressively extended through dedicated datasets and validation.

Dry–wet closed-loop learning

Every cycle adds evidence.

Select informative experiments
Measure controlled perturbations
Check quality and functional outcomes
Refine and evaluate the model

Current published model evidence focuses on oncology response and combination tasks. Safety, new disease contexts and prospective clinical uses require their own datasets and evaluation. The virtual-yeast framework informs longer-term chassis-cell research.

Across the drug-development workflow

One platform strategy.
Decisions at every stage.

Match data generation and model adaptation to the decision at hand. Open each stage to explore its research scope.

01Target discovery and mechanism of action

Connect perturbation-response signatures to target hypotheses, pathways and resistance mechanisms. Test causal hypotheses with functional experiments.

Task-specific co-development

02POC/PCC and candidate prioritization

Prioritize drug candidates, efficacy and combination hypotheses in defined cellular systems. Confirm predictions with independent dose–response experiments.

Published oncology response and combination tasks provide a starting point

03Preclinical animal studies

Develop response, efficacy and safety hypotheses for the relevant preclinical system. Evaluate transfer with dedicated animal-study data and appropriate experiments.

Extension requiring model-specific data and validation

04Clinical trials and translational research

Investigate biomarkers, patient stratification and response hypotheses with clinical partners. Evaluate each prospective clinical task in its own study.

Published retrospective oncology evidence; prospective clinical evaluation required

Published oncology evidence · Nature 2026

ProteinTalks

A published demonstration of protein-response modeling for drug efficacy, combinations and patient-relevant research. This oncology case study underpins our AIVC development strategy.

Read the publication

16,311

Proteomic profiles across 18 breast-cancer cell lines.

38M+

Temporal protein measurements capturing cellular responses.

63 + 59

Drugs and drug combinations in the published response dataset.

3,000

Candidates prioritized

Repurposing candidates ranked using baseline organoid proteomes and drug features. Selected candidates were tested in three TNBC patient-derived organoids.

501

Retrospective patients

A TNBC cohort supports clinically relevant prognostic research. The reported survival-analysis subgroup includes 62 low-risk and 59 high-risk patients.

ProteinTalks, Nature (2026), Figure 5 · Independent in-vitro and retrospective research evidence.

Clinical diagnostics

Protein measurements.
Clinical laboratory applications.

Explore our diagnostic products and translational proteomics capabilities.

Clinical products and enquiries

Clinical laboratory test kit

LMDx IGF-I

Insulin-like growth factor I, measured by LC–MS/MS.

A mass-spectrometry-based test kit for IGF-I quantification, supporting laboratory assessment of growth and endocrine biology.

Explore IGF-I

Thyroid-nodule proteomics

ThyroProt

Protein evidence for thyroid-nodule assessment.

Quantitative proteomics and AI for investigating benign and malignant thyroid nodules, including tissue and fine-needle-aspiration research.

Explore ThyroProt

Product-specific intended use and availability are separate from AIVC research predictions. Contact our clinical team for current product documentation.

Research partnerships

Find the right application.

Contact our team for AIVC datasets, predictive models, research services or diagnostic-product enquiries.

Partner with us

Visit and contact

Westlake Omics, Hangzhou

B3-1001, No. 1 Yunmeng Road
Cloud Town, Hangzhou 310024
Zhejiang, China

For AIVC projects, research services and clinical products, contact the appropriate team.

Contact our teams

+86 571 86780630

Data and modelsfor predictive biology.

From target discoveryto clinical research.

Customized datasets

Predictive vertical models

Foundation model