Areca's NEMo Achieves 74% Neoantigen Prediction Accuracy — Nearly Double the Next Best Competitor
NEMo uses evolutionary machine learning trained on 25,000+ immune checkpoint blockade mutations, treating immune-driven neoantigen elimination as a training signal. Validated across the TESLA benchmark and two cancer vaccine clinical trials.
ML and Genome-Wide Genetics Combine to Sharpen Type 1 Diabetes Risk Prediction
Integrating expanded genetic association data with machine learning improves polygenic risk stratification for type 1 diabetes well beyond current clinical approaches.
Foundation Model Trained on 30,000 Tumor Genomes Predicts Drug Resistance Across Cancer Types
MutationProjector projects tumor mutation profiles into unified biological coordinates, achieving best-in-class accuracy predicting immunotherapy and chemotherapy resistance — and uncovering unexpected biomarkers like KMT2A mutation.
NeoPrecis Beats Tumor Mutation Burden as a Predictor of Immunotherapy Response
By combining MHC-I/II neoantigen immunogenicity with tumor clonal architecture, NeoPrecis outperforms TMB across five melanoma and three NSCLC cohorts.
Germline and Somatic Integration Uncovers Two Distinct Immune Pathways to Checkpoint Blockade Response
ML analysis of integrated germline and somatic features reveals divergent immune mechanisms — including T-follicular helper infiltration in MHC class-I deficient tumors — explaining why some patients respond to ICB through unexpected routes.
New ML Tool Identifies Four Genetic Subtypes of Type 1 Diabetes — Enabling Earlier, Personalized Treatment
UC San Diego researchers developed T1GRS, a machine learning tool that predicts genetic risk for Type 1 diabetes across broader populations by analyzing complex gene interactions, uncovering four distinct disease subtypes with unique onset patterns and complication profiles.