Unlock Therapeutics that Treat Patients, Not Tumors
Therapeutic & predictive biomarker discovery
kinase cascade
Target Discovery
Identify pathways that drive molecular subtypes of disease progression by learning from longitudinal, clinicogenomic cohorts.
Therapeutic (Tx)
immune checkpoint drug
Stratify Trials
Optimize clinical trials for safety and efficacy with patient selection strategies that leverage MOA-related genomic biomarkers.
Companion Diagnostic (CDx)
Immunotherapy R&D should incorporate the genetics of the immune system. Restricting research to tumor omics overlooks the microenvironment.
scientific posters
KeyBio research presented at
AACR 2024 and ASCO 2023
Pan-cancer studies of survival biomarkers
  • Prognostic analysis antithetically highlights germline genomics as 10x more important in predicting survival than tumor multi-omics
  • Novel immunoregulatory targets discovered in multiple indications: NSCLC, colorectal, bladder, and melanoma.
OncoGerm — prognostic biomarker discovery engine
analytical workflow
Insight from in silico simulations reduces the breadth and cost of wet lab experiments.
Built on a deep learning technology platform
  • AI Quality Control (AIQC) is our open source Python framework for systematic deep learning
  • Rapidly train & evaluate neural networks on multi-modal data without sacrificing scientific rigor
  • Designed to automate reproducibility, explainablity, & data integrity for audits
track experiments
compare models
run simulations
Bult-in UI for real-time experiment tracking
More realistic modeling of systems biology
Association Studies are Univariate
Traditionally, each mutation is tested for correlation with the health outcome one-by-one. Lack of interaction produces many disparate hypotheses. No patient-specific predictions, only population-wide statistics.
gwas simplicity
v.s.
neural network topology
Neural Networks are Multivariate
Genes interact with each other within a unified algorithm. Predict patient-specific health outcomes. Prioritize more realistic population-wide biomarkers via permutation. Incorportate multi-modal data.
Rather than chasing the latest academic biostatistical tools, why not embrace the flexibility and power of deep learning?
Guidance from scientific & strategic experts
Founder (Scorpion Therapeutics, Loxo@Lilly, Strata Oncology), Director Clinical Research - MGH, Professor - Harvard Med
Keith is a pioneer of therapeutic cancer biomarkers. He is an accomplished leader in every sphere of this domain: translational medicine, storied academic research, as well as therapeutic and diagnostic industries.
Director GU Oncology & Clinical Trials - AdventHealth Cancer Institute, Director Bladder Cancer - DanaFarber, Professor of Medicine
Guru has published and practiced extensively in the field of clinical oncology at multiple cancer institutes. He has served as the principal investigator for several clinical trials, currently directs Advent's phase I program, and is extremely cognizant of the latest developments in the cancer therapeutics industry.
Physician Scientist, Medical Oncologist, Genomics Researcher:
BWH, Dana-Farber, Broad Institute
Arvind has developed comprehensive expertise across quantitative, experimental, and clinical domains at the highest level of excellence. This gives him the rare ability to not only see the whole picture but also zoom in wherever necessary. His foundational research on the role of miRNA in oncogenesis has refined the known biological mechanisms of this omic.
Pharma R&D Strategy:
Pfizer, WuXi AppTec, WuXiNextCODE
When it comes to the biopharma industry, Alex has seen it all. Having managed R&D, M&A, and commercial strategy at companies ranging from leading pharmaceuticals, to contract research organizations (CROs) and high-growth technology startups - his innate ability to navigate challenging situations with precision is unparalleled.
→ Contact