Alchemab Therapeutics, headquartered in the United Kingdom, takes a distinctive approach to drug discovery by studying people who stay healthy despite being at high risk of disease. The company's core idea is that resilient patients often carry naturally protective antibodies, and that by decoding these immune responses with machine learning, it can identify novel disease-modifying therapeutics that conventional target-led discovery would miss.
The platform operates like a search engine for the immune system. Alchemab has assembled what it calls the DataCube, a proprietary database of over six thousand highly curated patient-derived antibody samples collected from more than thirty global collaborators across metabolic, immune, and neurological conditions. Advanced machine-learning models interrogate this dataset to find antibodies associated with protection and resilience, then map them back to the targets they engage. This unbiased, patient-first approach lets the company discover both novel antibodies and the previously unknown biological targets they act on.
Alchemab pairs its computational platform with in vitro and in vivo drug-discovery capabilities, advancing candidates from data signal to validated therapeutic. Its lead program, ATLX-1282, entered Phase 1 clinical trials, and the company is progressing additional wholly owned programs including ATLX-2847 for muscle atrophy, as well as earlier-stage assets for immune and neurological conditions.
In September 2025 Alchemab announced a $32 million Series A extension, bringing its total Series A investment to $114 million. The round included Ono Venture Investment alongside a syndicate of specialist investors comprising RA Capital, SV Health Investors, DCVC Bio, and Lightstone Ventures, with participation from Eli Lilly and Company. The capital supports advancing its clinical and preclinical pipeline. By grounding discovery in real human resilience and applying machine learning at scale, Alchemab aims to uncover therapeutics that address the root biology of disease.