8 AI Companies to Watch in Clinical Trials
ALSO: A Map of Antibiotic Resistance; Deep Mind Introduces Tx-LLM; Antibody Discovery Beyond AlphaFold; LLC Business Model in Biotech
Hi! This is our weekly newsletter, ‘Where Tech Meets Bio,’ where we talk about technologies, breakthroughs, and great companies moving the biopharma and medtech industries forward.
If you've received it, then you either subscribed or someone forwarded it to you. If the latter is the case, subscribe by pressing this button:
Now, let’s get to this week’s topics!
Brief Insights
🔬 Google Research and DeepMind officially introduced Tx-LLM, a new language model designed to predict biological properties for drug discovery. It excels in tasks like toxicity prediction and binding affinity estimation, outperforming existing models on 22 of 66 tasks, enhancing therapeutic research.
💰 XtalPi’s Ailux Biologics signed a non-exclusive licensing deal with Janssen Biotech to utilize XtalFold, its AI-driven platform that predicts antigen-antibody structures. This agreement aims to accelerate biologics discovery, highlighting AI’s growing influence in therapeutic research.
🔬 Lantern Pharma's investigational drug LP-184 has received FDA Fast Track Designation for treating glioblastoma, aiming to address the urgent need for new therapies in this aggressive brain cancer. The company plans to start a Phase 1b/2a trial for recurrent glioblastoma in late 2024/early 2025, with LP-184 showing potential to be the first new GBM treatment in over 20 years.
🔬 The LLC business model in biotech might offer flexibility in structuring deals, targeted investments, and clear IP ownership, as highlighted in the article by Abbas Kazimi. However, risks include difficulties in managing complex or interconnected IP portfolios and challenges in maintaining workforce stability, making it a model that requires careful alignment with a company's long-term objectives.
🔬 Pfizer partners with the Ignition AI Accelerator, a joint initiative by NVIDIA, Tribe, and Digital Industry Singapore, to enhance AI-driven drug discovery and precision medicine in Southeast Asia, streamlining R&D and manufacturing processes.
🔬 Evaxion Biotech’s AI-Immunology platform demonstrated good clinical trial results, identifying effective neoantigens for personalized cancer vaccines, including EVX-01, which showed a 67% patient benefit rate in metastatic melanoma, advancing precision medicine for cancer treatment.
💰 clock.bio raised $5.3M in seed funding to research anti-aging therapies. The company has identified over 100 rejuvenation genes in its "Atlas of Rejuvenation Factors," aiming to reverse aging and extend human healthspan through targeted therapies.
🔬 The Wellcome Sanger Institute and Newcastle University unveiled a single-cell atlas of prenatal human skin, providing key insights into regenerative medicine and scar prevention. This work, part of the Human Cell Atlas initiative, offers a framework for skin and hair transplant advancements.
💰 British Patient Capital invested £8M as part of Nuclera's $75M Series C round, supporting the expansion of its eProtein Discovery system, which accelerates protein production and prototyping for drug discovery across the U.S. and Europe.
🤝 BioSkryb Genomics partnered with the Human Cell Atlas to provide access to its next-gen single-cell multiomics technologies like ResolveDNA and ResolveOME. This collaboration will support advanced research into human disease and precision medicine.
🤝 Arctoris, SpiroChem, and Orion Corporation launched the "Spark" project, combining AI-driven drug discovery with chemistry expertise to speed up breakthroughs in drug development. The collaboration aims to tackle difficult therapeutic targets with enhanced efficiency.
🤝 Imperial College London, Parkinson’s UK, and pharma partners (GSK, Novartis, Roche, UCB) launched the Landmark project, a three-year initiative mapping gene expression in Parkinson’s disease. Using snRNAseq, the project aims to discover new drug targets and identify predictive biomarkers for the disease.
🤝 Tempus and Takeda expanded their collaboration to enhance oncology R&D. Tempus’ multimodal real-world datasets and patient-derived tumor organoids will help accelerate cancer drug development, focusing on antibody-drug conjugates (ADCs) and T-cell therapies.
🤝 Eli Lilly partnered with AI-driven biotech Insitro to co-develop metabolic disease treatments. Insitro will license delivery technology from Lilly while retaining full global rights to its research, with Lilly eligible for milestone payments and royalties.
🤝 Google DeepMind and BioNTech announced a partnership to develop AI-powered lab assistants that improve scientific experiment planning. BioNTech also launched an AI assistant, Laila, designed to automate biology tasks and enhance research productivity.
🚀 Basecamp Research raised $60M in Series B funding to develop an AI platform for biology, known as "GPT for biology." The platform aims to surpass AlphaFold 2 in protein structure prediction and is already supporting drug discovery collaborations with organizations like the Broad Institute.
💰 British startup Shift Bioscience raised $16M in seed funding to develop AI-powered therapies aimed at reversing the epigenetic clock and rejuvenating aging cells, marking a major step forward from its humble beginnings when CEO Daniel Ives self-funded experiments.
🔬 GE HealthCare integrated Blackford’s AI-enabled application orchestration platform into its True PACS and Centricity PACS systems, helping radiologists manage workloads more effectively while improving diagnostic speed and accuracy.
🔬 Exscientia reached two major milestones in its collaboration with Sanofi, receiving $15M for advancing drug discovery programs into lead optimization. The company is eligible for over $600M in additional payments and royalties for future development success.
A Map of Antibiotic Resistance
A global "superbug map" has been developed to track the spread of Acinetobacter baumannii, a dangerous antibiotic-resistant pathogen common in hospitals. Scientists at the HUN-REN Biological Research Centre in Hungary created this map, analyzing over 15,000 genomes worldwide. The goal is to improve the use of bacteriophage therapy, an alternative to antibiotics, for infections that no longer respond to conventional treatments.
Key findings showed that a few strains dominate local infections and persist for about six years. This gives hospitals a window to deploy pre-emptive phage therapies, potentially treating up to 80% of local infections. By preparing phage cocktails based on these dominant strains, treatment time can be reduced in acute cases. The study also found that phages isolated from wastewater were effective against 95% of infections, including the common strain ST2-KL3, showing potential for future clinical treatments.
Key points:
The map helps tailor phage therapies to local bacterial strains, tackling hospital-acquired infections that resist antibiotics.
The strain stability over six years provides a timeline to act, enabling pre-preparation of region-specific phage cocktails.
Phage therapy, validated in lab and animal models, offers a promising solution for infections where antibiotics fail.
Antibody Discovery Beyond AlphaFold
XtalPi's Ailux Biologics made a deal with Janssen Biotech to use XtalFold, Ailux’s AI platform, which predicts how antibodies interact with targets. Janssen aims to speed up drug discovery using this advanced AI tool, which is better at solving certain antibody problems than AlphaFold.
XtalFold is an AI platform designed to make biologics discovery faster and more precise.
Key points:
AI-Powered Predictions: XtalFold predicts how antibodies interact with their targets, using just sequence data—no need for complex lab experiments.
Better Than AlphaFold?: In certain tricky areas, like the antibody-antigen interface, XtalFold reportedly outperforms AlphaFold-Multimer, making it more accurate for these specific tasks.
Proven in Real Research: Validated in over 30 research programs, this platform has been key in designing antigens and engineering complex antibodies, like bispecifics.
Fast & Accurate: With a 90% success rate for high-confidence data, XtalFold helps researchers model difficult structures much faster than traditional methods.
Why It Matters: By cutting down on development time and providing clearer insights, it speeds up the process of creating new biologic drugs.
LLC Business Model in Biotech
Biotech companies are increasingly adopting the LLC (Limited Liability Company) model to gain flexibility in structuring deals and managing assets. Abbas Kazimi of Nimbus Therapeutics explains how this model has been instrumental in their success, enabling deals like the $6 billion sale of their TYK2 inhibitor program to Takeda.
The LLC model allows firms to sell individual assets without giving up the entire company, opening new doors for targeted investments and partnerships. Kazimi points out that this structure helps biotech companies clearly manage intellectual property (IP) and recycle capital, making it easier to keep operations running smoothly during and after deals. However, there are challenges, particularly around managing complex IP portfolios and ensuring teams can adapt quickly to shifting priorities.
Advantages of the LLC model:
Flexibility in selling or licensing individual programs without losing control of the business.
Clearer, separate ownership of IP, which simplifies transactions.
Ability to recycle capital through program sales, keeping operations funded and investors satisfied.
Minimal disruption to the company’s other operations when selling off specific assets.
Deep Mind Introduces Tx-LLM, Again, Refined
The July introduction of Tx-LLM was an early showcase of the model's potential. It focused on the basics—explaining how it could support drug discovery by predicting properties of small molecules, proteins, and cell lines, all built on Google's PaLM-2 architecture.
Tx-LLM was designed as a general-purpose model for therapeutic research, trained on 709 datasets covering 66 tasks. It could handle key tasks in drug development, including toxicity prediction, binding affinity, and chemical reactions, and showed competitive performance across most of these. The goal was to make the drug discovery process more efficient by using machine learning to predict properties that normally require extensive experimentation.
What changed?
The recent introduction brings a more refined Tx-LLM. While still based on PaLM-2, it's now better integrated into Google's AI ecosystem with improved performance, particularly in tasks combining molecular data and text. The model's training process has been fine-tuned, especially its use of the Therapeutics Instruction Tuning (TxT) dataset to boost accuracy in predicting therapeutic properties.
Additionally, this update signals a shift toward external collaboration, inviting researchers to explore and contribute. Future improvements include integrating the Gemini family of models, which will further enhance its instruction-following abilities and interpretability.
More Insights
🚀 AI-driven healthcare startups in New York raised over $1B in 2024, up from $670M in 2023. Investors are pouring money into AI technologies that automate both drug discovery and administrative tasks, transforming operations across the healthcare sector.
🔬 Researchers at the University of Cologne developed an AI-powered pathology platform to diagnose lung cancer. The tool automates the analysis of tissue samples, improving diagnostic accuracy and enabling more personalized treatment plans for patients.
🚀 Venture capital investors have committed over $1.7B to companies developing AI-driven autoimmune disease therapies in the first half of 2024. This surge follows groundbreaking research on CAR-T therapy’s ability to reset the immune system, particularly in lupus and other autoimmune diseases.
🔬 Univercells Technologies launched the scale-X nexo bioreactor, the smallest in its scale-X series, designed for efficient cell culture process development. It supports applications like cell and gene therapies with improved scalability, cost-efficiency, and real-time monitoring via Skaia vision software.
Companies to Watch
Drug development is in practice tackling a colossal, convoluted puzzle that costs billions and often falls short. AI enters—not necessarily a magic fix, but rather a toolkit that might ease the process and help find new angles to work from. Machine learning might help find patients, predictive models may refine trial designs, and natural language processing could manage the heaps of paperwork.
But let's be real: these tools aren't flawless—they rely on the quality of the data we provide and can introduce their own complexities. So where does that leave us? Plenty of companies are jumping in, eager to address the arduous practicalities—or rather, impracticalities—of clinical trials.
Let’s see what some of them are up to to get a glimpse of how these technologies are playing out in the real world.