Accelerating antigen-targeting therapy discovery with a scalable pan-cancer bioinformatics platform

Cancer drug discovery is lengthy and nearly 90% of clinical trials fail due to efficacy or safety concerns. Identifying safe and effective drug targets using data-driven approaches in the earlier stages of research is critical to improve success rates. Over 40 antigen-targeting therapies have been FDA-approved in the past three decades but have yet to reach their full potential, especially in solid tumors where selecting the right antigen targets is a major challenge.

To address this, we developed a comprehensive and scalable translational bioinformatics platform leveraging underutilized public omic data to discover, and prioritize novel antigen targets across various cancers.

We integrated 168 microarray datasets using our AI-powered, human-supervised clinical data curation and transcriptomic data normalization pipeline. These were aggregated into 18 indication-specific cohorts encompassing 7,885 tumor and 1,569 healthy samples, and profiling 20,347 genes.

We leveraged our proprietary target discovery pipeline, starting with a differential gene expression analysis, followed by proteomic filters selecting cell surface proteins with transmembrane or glycosylphosphatidylinositol-anchored domains.

We further prioritized targets using a proprietary framework incorporating bulk and single cell transcriptomic data, proteomic, biological and pharmacological data metrics into a weighted scoring system designed to anticipate clinical relevance. Metrics were classified into 3 categories (efficacy, safety and novelty) and weights were optimized using FDA-approved targets.

We identified an average of 216 potential antigen targets per indication, ranging from 41 to 556. Among these, we successfully rediscovered FDA-approved targets across several indications, including CD19 and CD22 in acute lymphoblastic leukemia, ERBB2 and TACSTD2 in breast invasive carcinoma, PDL1 in skin melanoma, PSMA in prostate cancer and FOLR1 in ovarian serous carcinoma, alongside targets currently in clinical trials for antigen-based therapies.

Our prioritization framework was then applied to our candidate targets and revealed novel antigen targets exhibiting characteristics similar to those of FDA-approved targets. For example in breast invasive carcinoma, we identified 16 highly promising targets with efficacy scores matching or exceeding FDA approval benchmarks, including 9 with better predicted safety and 2 with high novelty scores in this indication.

Additionally, 14 of these promising targets were identified in two or more indications, highlighting their tissue-agnostic potential.

Our platform integrates unbiased data-driven tools with cancer biology insights to streamline antigen target discovery, from data integration to target prioritization. Scalable to any cancer type or antigen-targeting modality, it offers a robust framework for accelerating oncology drug discovery.