UK’s ARIA Backs Deep Origin & Arctoris Autonomous Scientist Initiative for Endometriosis
Deep Origin, in partnership with Arctoris, has been selected for the Advanced Research + Invention Agency’s AI Scientist program (ARIA), one of the UK’s most selective innovation initiatives. The joint program will focus on building an AI-enabled autonomous scientist designed to accelerate research in endometriosis.
Despite affecting millions of women globally, endometriosis remains underdiagnosed and comparatively underfunded, with significant gaps in mechanistic understanding and therapeutic innovation. The initiative aims to address both the scientific and translational bottlenecks that have slowed progress in this field.
This work complements Deep Origin’s predictive ADME-Tox modeling initiatives under ARPA-H’s CATALYST program. By integrating early-stage discovery intelligence with downstream developability prediction, the broader strategy targets a persistent industry challenge: high attrition rates driven by late-stage safety and pharmacokinetic failures.
The partnership integrates two complementary capabilities:
- Deep Origin’s computational drug discovery platform, engineered to generate, prioritize, and optimize hypotheses across complex biological systems
- Arctoris’ automated laboratory infrastructure, capable of generating high-quality wet lab data at scale with tight experimental control
Arctoris is a Partnership Research Organisation (PRO) that combines scientific expertise with a fully robotic, cloud-controlled drug discovery platform to run assay workflows across biochemistry, cell biology, and structural biology. Built around its Ulysses system, the company generates highly reproducible, FAIR-compliant experimental datasets by capturing every step and parameter in real time, supporting data-driven R&D.
See also: Using End-to-End Wet Lab Automation Combined with Machine Learning to Industrialize Drug Discovery
The collaboration aims to combine machine learning with automated experimentation to speed up hypothesis generation and testing, connect experimental data directly back into computational models for faster iteration, and progressively move toward a system that can plan, run, and analyze experiments with limited manual input.
Elements of DeepOrigin’s AI architecture are already accessible:
- Balto is a no-code molecular simulation platform that allows chemists to retrieve chemical data from databases such as ChEMBL and BindingDB, summarize literature, identify protein pockets, and dock small molecules through an AI-guided interface. It integrates large-scale data retrieval with a docking engine to produce binding scores and AI-predicted chemical properties in a single workflow.
- DO Patent is an AI agent that extracts full molecular structures from patents and scientific PDFs and converts them into verified, editable SMILES strings within minutes. It reports greater than 98 percent full-molecule image extraction accuracy and assigns confidence scores linked to the original source pages, enabling direct validation and correction within the interface.
Topic: AI in Bio
