Breakthrough at Basecamp Research: Launch of World's First AI Models for Programmable Gene Insertion

A New Era in Genetic Medicine: Basecamp Research’s AI Models

Basecamp Research, an innovative AI laboratory dedicated to transforming drug development through evolutionary science, has recently unveiled groundbreaking AI models capable of programmable gene insertion. This breakthrough addresses longstanding challenges in gene medicine, presenting a unique solution for replacing faulty genes and reprogramming cells for therapeutic uses.

In collaboration with NVIDIA, Basecamp developed these AI models by training the largest evolutionary AI on an extensive, globally sourced dataset, significantly accelerated by NVIDIA BioNeMo. This partnership has not only enhanced the models but positioned Basecamp Research as a pioneering force in developing a new generation of treatments for cancer and inherited diseases.

John Finn, Chief Scientific Officer at Basecamp Research, emphasized the potential of these advancements: "We believe we stand at the cusp of a major expansion of options for patients suffering from cancer and genetic disorders. Harnessing AI in the development of therapeutic enzymes has the potential to expedite the creation of cures for thousands of previously untreatable conditions, potentially changing millions of lives."

Programmable Gene Insertion: A Game Changer

Programmable gene insertion has long been a pivotal goal in gene medicine, with the ability to place large therapeutic DNA sequences at specific sites in the human genome. Traditional CRISPR-based approaches have limitations, mainly being restricted to small edits and often causing DNA damage, which hinders their usability. However, Basecamp Research has become the first organization to demonstrate that AI can craft enzymes capable of inserting large gene sequences precisely into the human genome, thereby opening doors to much-desired programmable therapies.

The AI-driven platform, aiPGI™, powered by a family of evolutionary AI models called EDEN, learns the language of DNA and evolutionary patterns. This learning allows the algorithms to design new programmable therapies for cancers and genetic diseases.

Recent laboratory results, detailed in a publication co-authored by NVIDIA, Microsoft, and leading scientists, showcase that the EDEN models successfully designed insertion proteins that target 100% of tested disease-relevant sites in the human genome using only the genomic target site as input. This represents a significant leap forward in the capabilities of AI models.

Basecamp Research has already demonstrated successful insertions at over 10,000 disease-relevant sites in the human genome, including therapeutic integration of cancer-fighting DNA into primary human T-cells at novel safe-harbor locations. This breakthrough has resulted in CAR-T cells that effectively target and kill cancer cells, achieving over 90% tumor cell destruction in laboratory tests.

AI-Developed Molecules to Combat Superbugs

In a further testament to the versatility of their models, Basecamp took aim at the global crisis of antibiotic resistance, proving the AI's capability by designing a targeted library of novel antimicrobial peptides (AMPs) – small proteins with the power to eradicate harmful bacteria. Impressively, 97% of the candidates displayed validated activity in lab tests. Collaborative efforts with researchers from the University of Pennsylvania, led by Professor César de la Fuente, demonstrated that the top-performing AMPs showed high efficacy against critical multidrug-resistant pathogens, providing a potent new weapon in the fight against dangerous superbugs.

The Technology Behind the Breakthrough

The EDEN models driving aiPGI™ were trained on over 10 trillion tokens of evolutionary DNA from more than one million newly discovered species. This unique data was collected over five years from over 150 sites across 28 countries, employing an innovative data collection strategy developed by the company. The largest EDEN model was trained on a computational power equivalent to 1.95 x 10²⁴ FLOPS using a cluster of 1,008 NVIDIA Hopper GPUs, placing it among the most computationally intensive biological models ever reported.

Building a Pipeline for Therapeutic Innovations

These revolutionary capabilities support Basecamp Research’s growing pipeline of cell and gene therapies, paving the way for treatments that are more precise, predictable, and personalized than those currently available. The company aims to develop potentially curative therapies for a range of cancers and genetic diseases, leveraging ongoing improvements in BaseData™, the EDEN models, and aiPGI™.

About Basecamp Research

Basecamp Research is committed to addressing fundamental challenges in the life sciences by exploring Beyond Known Biology™. The company creates groundbreaking AI models based on BaseData, recognized as the world’s largest ethically sourced and globally representative biological dataset. By collecting and curating its own biological data through partnerships with over 152 organizations in 28 countries, Basecamp grants its AI access to a genetic diversity unavailable to models trained on public databases. This enables Basecamp Research to design new protein sequences and biological systems that can accelerate therapeutic research and development.

Collaborating with biopharmaceutical companies and academic institutions globally, Basecamp Research has received numerous accolades, including recognition as one of Fast Company’s Top 10 Most Innovative Companies in Biotech and a place on the Financial Times-backed Sifted AI100 list of leading AI startups in Europe. For more information, visit basecamp-research.com.