ByteDance's Anew Labs unveils its inaugural AI-designed therapy as the parent company of TikTok joins the competition in drug discovery.

      **TL;DR**: ByteDance's drug discovery division, Anew Labs, unveiled its first AI-generated therapy at a significant immunology conference in Boston. This therapy is a small molecule targeting IL-17, a protein-protein interaction that has been deemed undruggable for years. The unit also introduced AnewOmni, a generative AI framework trained on five million biomolecular complexes, claiming to be the first to design functional molecules at all scales. ByteDance enters the AI drug discovery field alongside Isomorphic Labs, Anthropic, and Insilico Medicine.

      The same company that developed TikTok's recommendation algorithm now uses a similar type of AI to forecast molecular behaviors within the human body. Anew Labs demonstrated its first AI-designed therapy at the American Association of Immunologists’ annual meeting in mid-April, detailing a small molecule that generative AI crafted to inhibit IL-17, a cytokine associated with autoimmune diseases like psoriasis and rheumatoid arthritis.

      The molecule interacts with a protein-protein target that the pharmaceutical sector has long regarded as undruggable due to challenging binding surfaces. Anew Labs asserts that its AI has found a way to target it. This presentation marked a significant debut for ByteDance's drug unit, with future exhibitions planned, including one at the BIO International Convention in June and another at the Free Energy Workshop in Barcelona next week.

      **The Unit**: Anew Labs operates from Shanghai, Singapore, and San Jose, California. It has a core team of 36 listed on its website and an advisory board featuring notable experts from companies dominating the biologics and immunology spheres. Anew Labs aims to replace costly injectable antibody therapies with oral pills by utilizing generative AI to design small molecules that mimic the function of these antibodies.

      Chris Li, head of biology, showcased one of Anew Labs' drug candidates in Boston, a pan-spectrum IL-17 inhibitor designed to block multiple variants of IL-17. Current IL-17 treatments are injectable antibodies with revenues in the billions. An oral small molecule with similar efficacy would be groundbreaking, reducing production costs and improving patient compliance.

      However, the challenge lies in IL-17's broad, shallow protein-protein interaction, which makes it difficult for small molecules to bind effectively. The gap between laboratory AI capabilities and real-world patient applications is a persistent issue in health technology, and IL-17 is an example where this gap is vast.

      **The Model**: In March, Anew Labs shared a preprint on bioRxiv detailing AnewOmni, a generative AI system trained on over five million biomolecular complexes. This model aims to function across various molecular scales, from small compounds to peptides, by assembling chemically meaningful components at atomic detail.

      The researchers demonstrated AnewOmni's ability to design functional molecules targeting KRAS G12D and PCSK9, achieving success rates of 23 to 75 percent with limited laboratory testing. The model employs programmable graph prompts for guiding the molecular generation process based on specified constraints.

      This approach is notable as it attempts to tackle a major limitation in AI drug discovery: many generative models excel at one molecular scale but falter across others. AnewOmni claims to be the first to successfully design functional molecules at multiple scales, which, if validated, would enhance Anew Labs' capabilities beyond single programs. Meanwhile, competitors such as Isomorphic Labs have developed their own tools focusing on improving precision in drug design.

      **The Context**: ByteDance is not alone in the venture into drug discovery. Other tech companies, like Anthropic and Google's DeepMind, have also made strides in this area. Nvidia has developed BioNeMo, a platform for biomolecular AI modeling. The trend shows that firms with robust AI infrastructures are channeling some of their expertise into biology since drug discovery aligns well with AI's strengths in navigating complex searches for solutions.

      ByteDance's unique entry stems from its experience with AI in predicting user behaviors on TikTok, using vast data to anticipate content engagement. Anew Labs leverages this foundational capability, processing extensive molecular data to predict atomic combinations that yield desired biological effects.

      **The Test**: Currently, over 173 AI-discovered drug programs are in clinical development worldwide, with 15 to 20 set for large trials this year. The efficacy of AI in revolutionizing drug development remains uncertain, particularly with the industry's high clinical failure rate.

      While Insilico Medicine’s rentosertib, developed using AI, has shown promising results in trials, challenges persist, as seen with the Recursion-Exscientia merger's discontinuation of its lead candidate due to inadequate long-term efficacy data. The common pattern of initial promise followed by the harsh realities of drug development—where molecules that perform well in labs often fall short in human bodies—persists.

      Anew Labs currently has four drug candidates and a generative platform that, if successful in preclinical results, could design functional molecules broadly. Backed by