AI model claims programmable gene insertion
- On January 12, 2026, Basecamp Research announced EDEN, an AI model family it said can design programmable gene insertion tools for preclinical therapies. - Basecamp said EDEN designed active insertion proteins for 100% of tested disease-relevant human target sites in results tied to a bioRxiv preprint. - The preprint remains on bioRxiv, and Basecamp said NVentures joined its pre-Series C round after the January 12 announcement.
Basecamp Research said on January 12 that it had developed AI models that can design enzymes for programmable gene insertion, a claim aimed at one of gene therapy’s harder technical problems. The London- and Cambridge, Massachusetts-based company said the system, called EDEN, was trained with NVIDIA and described in a bioRxiv preprint posted the same day. BioScience Today recirculated the claim on May 17 in a social media post that referred to a preprint or lab project but did not identify the developer or funders. Basecamp named both in its January announcement, saying NVentures, NVIDIA’s venture arm, had invested in the company’s pre-Series C round. ### Which company is behind the claim? Basecamp Research was the named developer in the January 12 announcement cited by BioScience Today’s later post. The company said EDEN is a family of “evolutionary AI models” trained on its proprietary BaseData genomics dataset and developed in collaboration with NVIDIA. The bioRxiv preprint lists dozens of authors and says the work was “co-authored by NVIDIA, Microsoft and leading academics,” according to Basecamp’s announcement. (biosciencetoday.co.uk) The preprint itself identifies the paper as not peer reviewed. ### What does “programmable gene insertion” mean here? Basecamp defined programmable gene insertion as placing large therapeutic DNA sequences at precise locations in the human genome. (biosciencetoday.co.uk) In its materials, the company contrasted that with many CRISPR-based approaches, which it said are limited to smaller edits and often require DNA damage. (biorxiv.org) The preprint says EDEN was tested on therapeutic design tasks that included “AI-programmable Gene Insertion,” in which the model designs large serine recombinases for specified DNA target sites. That means the model’s output is not a finished therapy but a predicted enzyme design intended for laboratory validation. ### What evidence did Basecamp put forward? Basecamp said EDEN designed multiple active insertion proteins for 100% of tested disease-relevant target sites in the human genome, using only the genomic target site as a prompt. (biosciencetoday.co.uk) The company also said it had demonstrated insertion at more than 10,000 disease-related locations in the human genome. (biorxiv.org) In a second experimental result, Basecamp said therapeutically relevant integration of cancer-fighting DNA into primary human T cells at novel safe-harbor sites produced CAR-T cells with more than 90% tumor-cell clearance in laboratory assays. Those figures were reported by the company and summarized in the preprint-linked announcement, not in a peer-reviewed journal paper. (biosciencetoday.co.uk) ### Why do insertion sites matter so much in gene therapy? The U.S. Food and Drug Administration’s Center for Biologics Evaluation and Research said in a 2024 workshop notice that stable integration can offer therapeutic benefits but can also alter host gene expression and contribute to secondary malignancies through insertional mutagenesis. The agency said integration site analysis is the established method for identifying where vectors insert and described method design and interpretation as complex and product-specific. (biosciencetoday.co.uk) That regulatory backdrop helps explain why companies talk about “safe-harbor” sites and insertion-site design. A model that scores or proposes insertion targets would still need wet-lab confirmation and safety testing before any clinical use. That is an inference from the FDA’s description of insertion-site risk and from Basecamp’s own characterization of the work as preclinical. (fda.gov) ### Is this a commercial product or an early research claim? January 12 is the only verified public launch date surfaced in the reporting and source material reviewed here. Basecamp presented EDEN as a platform and research program, and the linked manuscript remains a bioRxiv preprint rather than a peer-reviewed paper. John Finn, Basecamp’s chief scientific officer, said in the company announcement that the work could accelerate development of treatments for cancer and genetic disease. (biosciencetoday.co.uk) The next concrete public reference point is the preprint on bioRxiv and any later peer-reviewed publication or company update naming additional collaborators, financing, or experimental milestones. (biorxiv.org)