ADC Payload Selector by Turbine.AI

The ADC Payload Selector by Turbine is an AI-powered virtual biology platform designed to maximize the probability of success for antibody-drug conjugate (ADC) programs. Built for oncology drug developers and researchers, it enables teams to simulate biological experiments in silico, reducing reliance on costly and time-consuming wet lab work. The platform helps scientists select the most promising payloads, identify responsive patient populations, and understand resistance mechanisms — all before committing to clinical or preclinical investment.

Turbine's ADC Payload Selector addresses three core challenges in ADC development: selecting cancer sub-indications or molecular biomarkers predictive of payload sensitivity, identifying payloads or payload combinations with the highest chance of efficacy in heterogeneous, biomarker-positive model populations, and understanding and addressing the mechanisms driving resistance in the clinic.

Core Platform Capabilities

• Select cancer sub-indications or molecular biomarkers predictive of payload and payload combination sensitivity
• Identify payloads or payload combinations with the highest likelihood of efficacy across heterogeneous, biomarker-positive model populations
• Understand and address mechanisms driving resistance observed in the clinic
• Design dual-payload ADCs rationally rather than through trial-and-error approaches
• Identify predictive biomarkers early and link payload response to mechanistic drivers of resistance

vLab Core — Simulations on Public Data

• Simulate experiments using public data harmonized and enhanced by Turbine
• Access a library of over 1,400 oncology cell lines
• Explore 112 drugs and drug-like molecules, including 18 payloads
• Perform knockout (KO) alterations across 3,500+ genes
• Utilize a robust interpretation toolkit for practical filtering and analysis of in silico results

vLab Custom — Proprietary Dataset Integration

• Generate in silico predictions leveraging your own proprietary datasets combined with Turbine's expertise
• Add your own sample library to the simulation environment
• Incorporate your own drug or drug candidate into the platform
• Access interpretation services to translate results into actionable insights
• Leverage a virtual PDX and PDO library of approximately 400 samples
• Generate wet lab perturbation data as needed
• Validate key hypotheses through wet lab experiments

Validated Performance and Proof Points

• Accurately predicts drug responses using limited data — compound responses with known mechanisms can be predicted at high accuracy across 1,400+ cancer cell lines, achieving a 0.75 Pearson correlation of unseen cell line responses
• Correctly predicted the sensitivity order of subtypes for microtubule payloads and produced closely aligned rankings for TOP1 inhibitors, potentially guiding trial strategy and avoiding dead-end subtypes
• Predicts GDSC2 Topo1i–PARPi synergy from training only on monotherapy data, enabling rational dual-payload ADC design
• In silico pharmacology and biomarker screens suggested a viable patient selection hypothesis for Topo1 ADCs in NSCLC, pending preclinical validation

Turbine's Virtual Lab is available for ADC simulations and can be accessed via a live demo. The platform is supported by Turbine's business development and scientific teams, and is backed by ongoing financing and immunology partnerships that continue to advance its virtual biology capabilities.