Topological data analysis for antibody-drug conjugate payload discovery: a computational framework for mechanistic...

Area of Science:

• Computational Chemistry
• Cheminformatics
• Oncology Drug Discovery

Background:

• Antibody-drug conjugate (ADC) payload discovery is limited by traditional methods that fail to capture crucial 3D geometric features.
• Existing molecular descriptors do not adequately represent shape-dependent target recognition and mechanism of action.

Purpose of the Study:

• To develop and validate a Topological Data Analysis (TDA) framework for ADC payload discovery.
• To leverage persistent homology for characterizing molecular shape and uncovering hidden mechanistic relationships.

Main Methods:

• A comprehensive TDA framework was applied to 22 FDA-approved ADC payloads using 1,471 clinical trial records.
• Computed 31 topological descriptors, including Betti numbers and persistence statistics.
• Utilized hierarchical clustering, PCA, and molecular docking for analysis and validation.

Main Results:

• TDA-based clustering identified eight distinct payload classes with significant separation.
• Principal components explained 79.8% of topological variance, highlighting Betti numbers and persistence lifetime.
• Three major mechanistic clusters were identified: vinca alkaloids, camptothecins, and DNA alkylators.

Conclusions:

• Established the first validated TDA framework for ADC payload discovery.
• Demonstrated that persistent homology captures biologically relevant mechanistic classifications.
• The TDA framework supports rational payload design and mechanism-of-action prediction in precision oncology.