TechBio Latest News
MetaphysicalCells: A newsletter about Science, Technology and AI Drug Discovery
Welcome back to another free edition of MetaphysicalCells 🪐🛸
▶️ Evo 2 Can Design Entire Genomes
A new AI model for biology, released by Arc Institute and NVIDIA, can predict which mutations within a gene are likely to be harmful and even design small, eukaryotic genomes.
AI Technology Breakthroughs, Innovation and Applications 💪
AI continues to revolutionize drug discovery, genetic analysis, and biotech research:
BostonGene Corporation was recognized for its AI innovation, underscoring the growing importance of ML in advancing precision medicine (BostonGene Named a Fierce Life Sciences Innovation Awards Winner in the AI Innovation Category). ➡️ BostonGene founded in 2015 is a biotechnology company specializing in advanced computational biology and precision medicine. Their AI-powered multiomics platform decodes cancer patients' molecular profiles, including their immune system and tumor microenvironment, to uncover key disease drivers, identify novel drug targets and recommend the most effective treatments. With advanced bioanalytics, an indication-specific cancer library and a next-generation CLIA-certified, CAP-accredited high-complexity laboratory, they deliver precise, clinically validated insights that drive precision medicine and advance oncology research. On February 4, 2025, BostonGene Contributed to Friends of Cancer Research White Paper Introducing a Novel Process for Companion Diagnostics for Rare Biomarkers and Indications.
GeneDx's Multiscore leverages AI for genetic analysis, enhancing diagnostic capabilities and personalized medicine (GeneDx Launches Multiscore: AI-Powered Tool for Genetic Analysis). Through the combination of GeneDx’s industry leading proprietary dataset, publicly available data, and the power of AI, Multiscore prioritizes genes with positive findings in exome and genome sequences by ranking them based on alignment with a patient’s clinical presentation. More specifically GeneDx’s AI-powered gene ranker accelerates analysis, reducing turnaround times and costs for patients, and ultimately increasing access to care. ➡️ GeneDx (GeneDx Holdings Corp) (Nasdaq: WGS) is a genetic testing company that was founded in 2000 by two scientists from the National Institutes of Health, Sherri Bale and John Compton. Fueled by one of the world’s largest rare disease data sets, their industry-leading exome and genome tests translate complex genomic data into clinical answers that unlock personalized health plans, accelerate drug discovery, and improve health system efficiencies.
Elsevier (Elsevier B.V. part of the RELX Group), a global leader in advanced information and decision support, launched (March 12, 2025) ScienceDirect AI, a cutting-edge generative AI tool for researchers, enabling them to instantly extract, summarize and compare trusted insights from millions of full-text articles and book chapters on ScienceDirect, the world's largest platform for trusted, peer-reviewed research (Elsevier launches ScienceDirect AI to transform research).
Rakovina Therapeutics Inc is advancing AI-designed cancer drugs, reflecting the growing role of computational tools in oncology (Rakovina Therapeutics advances AI-designed cancer drugs). In particular, the company announced the receipt of a synthesized batch of ATR inhibitor compounds, a result of its collaboration with Variational AI Inc (Rakovina and partnership with Variational AI). ➡️ Rakovina Therapeutics (TSX.V: RKV) in Canada, Vancouver is developing the Deep Docking™ AI (DD) that is a novel DL platform suitable for docking billions of molecular structures in a rapid and accurate way by utilizing quantitative structure–activity relationship (QSAR) deep models, trained on docking scores of subsets of a chemical library, to approximate the docking outcome for yet unprocessed entries and therefore to remove unfavorable molecules in an iterative manner (Deep Docking: A Deep Learning Platform for Augmentation of Structure Based Drug Discovery). ➡️ Variational AI in Vancouver (developer of the Enki™ platform) leverages a powerful new form of ML known as generative AI to free scientists from reliance on screening libraries (both experimental and virtual) and to eventually generate de novo molecules with all the optimized properties, in order to discover efficacious, safe and synthesizable small molecule therapeutics in a fraction of the time and cost (100 AI-Generated Molecules Are Worth A 1,000,000 Molecule High-Throughput Screen).
Quercus Biosolutions Inc is disrupting crop protection with designer mini proteins, showcasing AI’s versatility beyond human health applications (Exclusive: Quercus Biosolutions emerges from stealth to disrupt crop protection with designer mini proteins). ➡️ Quercus Biosolutions is bringing the power of genAI to create a new generation of biology-based crop protection solutions. Agriculture 🌾👩🌾🚜 needs crop protectants that are safe, specific, and environmentally responsible. Traditional chemical approaches are failing. They face growing regulatory and consumer acceptance challenges, and cannot meet the innovation needs of an agricultural system facing the challenges of growing global population and changing climate. Accordingly Quercus is focused on leveraging the power of genAI to create effective and affordable biology-based solutions for crop protection.
Syntec Optics is aiding novel drug discovery technologies, emphasizing the role of advanced manufacturing in biotech (Syntec Optics aids novel drug discovery tech). ➡️ Syntec Optics (Nasdaq: OPTX) is an American optics and photonics manufacturing company that develops, manufactures, and assembles custom optical solutions for defense & security, medical, and virtual reality applications.
Mindoo AI (Mindoo AI BV) is a start-up that develops AI assistants 👨⚕️ for healthcare. ➡️ Mindoo AI - Healthcare Assistants integrates AI as an active participant in the healthcare journey, to automate a range of tasks, from collecting and analyzing medical history before a doctor's visit to taking notes during consultations, managing administration and bridging language barriers. Their platform creates an AI-powered interface that engages with patients and collaborates with healthcare professionals to deliver personalized, high-quality care.
Sightera Biosciences is a pre-clinical stage biotech spin-off from the University of Antwerp in collaboration with the Antwerp University Hospital. ➡️ Sightera focuses on developing innovative small molecules for oncology and regenerative medicine by combining patient-relevant models with AI. Sightera aims to enhance the efficiency of traditional drug development by designing highly patient-relevant small molecules from the start of the drug discovery process. Through their unique technology, they leverage chemical patterns recognition together with a drug synergy approach using patient-derived models to create the next-generation potentiators.
Federated Learning & Knowledge Distillation are emerging as critical tools for collaborative, data-driven drug discovery while preserving data privacy (Data-driven federated learning in drug discovery with knowledge distillation). ▶️ FLuID (federated learning using information distillation), is a data-centric application of federated distillation tailored to drug discovery aiming to preserve data privacy. FLuID was validated in two experiments, the first involving public data simulating a virtual consortium and the second in a real-world research collaboration between eight pharmaceutical companies. Although the alignment of the models with the partner specific domain remains challenging, the data-driven nature of FLuID offers several avenues to mitigate domain shift. FLuID fosters knowledge sharing among pharmaceutical organizations, paving the way for a new generation of models with enhanced performance and an expanded applicability domain in biological activity predictions.
π‐PhenoDrug, a DL‐based tool, is developed for cell phenotype‐driven drug screening. It integrates cell segmentation, morphological profile construction, and phenotype analysis modules of high‐content screening (HCS) processes. ▶️ π‐PhenoDrug was applied to evaluate drug response in various human melanoma cell lines, demonstrating that π‐PhenoDrug can evaluate drug killing effects across different cell lines via both supervised and unsupervised modes. Furthermore, π‐PhenoDrug identified drugs with potential killing effects on melanoma cells from a library of diverse compounds. When compared with traditional single‐readout assays, this method was more sensitive to compounds that induce weaker phenotypes, reducing the risk of overlooking effective drugs. These results confirmed that π‐PhenoDrug can achieve high‐throughput and accuracy analysis of cell phenotypic data using an unbiased and automated workflow, improving drug discovery efficiency (π‐PhenoDrug: A Comprehensive Deep Learning‐Based Pipeline for Phenotypic Drug Screening in High‐Content Analysis).
Newcastle University scientists have created a new 3D bio-printer which closely mimics human tissues. For creating this printer, the researchers got funding from Versus Arthritis and earned patents in the US and Europe. The team was led by Professor Kenny Dalgarno and it overcame various limitations of existing printers. The process discovered by the team is called ▶️ Reactive Jet Impingement (ReJI) bioprinting technology. The groundbreaking process jets two different liquids at each other, one has a polymer solution and another one has a cross-linking solution (Newcastle University Scientists Unveil New Technology To Speed Up Drug Discovery).
Simulations Plus Inc announced that Shawn O’Connor, Chief Executive Officer, will be participating in a fireside chat at the KeyBanc Capital Markets Virtual Healthcare Investor Forum on Tuesday, March 18, 2025, at 2:15 p.m. Eastern Time. In addition, Mr. O’Connor will host one-on-one meetings with institutional investors throughout the day. (Simulations Plus to Present at the KeyBanc Capital Markets Virtual Healthcare Investor Forum). ➡️ Simulations Plus (Nasdaq: SLP) in US—a premier developer of drug discovery and development software as well as a leading provider of both preclinical and clinical pharmacometric consulting services for regulatory submissions and quantitative systems pharmacology models for drug-induced liver injury and nonalcoholic fatty liver disease—has ADMET Predictor 12 (AP12) that is their latest version released (July 30, 2024) of the flagship ML platform for ADMET modeling and its extended capabilities for data analysis, metabolism prediction and AI-driven drug design (Simulations Plus Releases ADMET Predictor® Version 12). Apart AP12 and GastroPlus—a software for modeling and the conducting of simulations of properties of drugs or chemicals including absorption, release rate, bioavailability, pharmacokinetics and pharmacodynamics—Simulation Plus has also the following ADMET software: DILIsym, a Quantitative Systems Toxicology (QST) software capable of predicting and explaining Drug-Induced Liver Injury (DILI).
Harbour BioMed (HKEX: 02142) (Harbour BioMed Therapeutics Limited), a global biopharmaceutical company committed to the discovery, development and commercialization of novel antibody therapeutics in immunology and oncology, just launched Élancé Therapeutics to improve current obesity treatment, specifically to increase overall body weight loss and fat mass loss, but preserve and even increase muscle and lean mass (Harbour BioMed Launches Élancé Therapeutics to Advance Next-Generation Obesity Therapies). ➡️ Élancé Therapeutics will leverage Harbour BioMed’s HCAb-based bispecific antibody technology to advance these next-generation therapies. Moreover, Élancé will refine and expand Nona Biosciences' Hu-mAtrIxTM AI platform (Nona Biosciences Integrates Cutting-Edge AI Technology to Enhance Its Fully Human Antibody Platform, Accelerating Antibody Discovery Across Key Therapeutic Areas) to support bispecific antibody discovery, with AI applications guiding antibody sequence discovery, enrichment, optimization, bispecific geometry design, and PK assessments, as well as patient biomarker studies. ➡️ Nona Biosciences (a wholly-owned subsidiary of HBM Holdings Limited, HKEX: 02142) is a global biotechnology company committed to providing a total solution for partners worldwide, from academies, biotech startups to biopharma giants. Their integrated antibody discovery services range from antigen preparation, animal immunization, single B cell screening, to antibody lead generation and engineering, developability assessment and pharmacological evaluation, leveraging the advantages of Harbour Mice® platforms and the experienced therapeutic antibody discovery team.
Key Partnerships, Collaborations & Acquisitions 🤝
Collaborations between tech giants, biotech firms, and academic institutions are driving innovation:
ImmunoPrecise Antibodies Ltd (IPA) (NASDAQ: IPA) just announced a strategic partnership 🔗 with a leading biotechnology company to advance the discovery and development of Antibody-Drug Conjugates (ADCs) and bispecific antibodies for the treatment of cancer, integrating IPA’s proprietary B-cell Select™ platform and AI-driven discovery capabilities to enhance the efficiency and precision of therapeutic development (ImmunoPrecise Antibodies and a Global Biotechnology Leader Enter into a USD $8M-$10M Partnership for the Development of Novel Cancer Therapeutics). The agreement, valued at an initial $8M with the potential to expand to $10M based on program progression, will span 18 to 24 months, integrating AI across the workflow to enhance discovery and optimization. ➡️ IPA is a Contract Research Organization (CRO) providing highly specialized full-continuum therapeutic antibody discovery, development and out-licensing services (with ➡️ Talem Therapeutics LLC)—combined with advanced omics and complex intelligence technology (LENSᵃⁱ) developed by ➡️ BioStrand (BioStrand: The Ultimate Convenience in Multi-Omics Data Search and Analysis 🔝).
Amazon Web Services (AWS) and ElevateBio (ElevateBio, LLC) are partnering to enhance CRISPR gene editing therapies using generative AI, combining cloud computing power with cutting-edge biotech (AWS and ElevateBio join forces to advance gene editing). More specifically, by integrating AWS’s powerful cloud computing and AI solutions with ElevateBio’s Life Edit expansive CRISPR dataset, the collaboration is set to accelerate drug development for a vast range of genetic diseases. ➡️ ElevateBio (Cell and Gene Therapy Innovation Right From The Start) is a technology-driven company built to power transformative genetic medicines. The ElevateBio ecosystem combines multiple R&D technology platforms including: Life Edit, a next-generation, full-spectrum gene editing platform; comprehensive cell engineering technologies; and an expanded viral and non-viral therapeutic delivery platform – with BaseCamp®, its end-to-end genetic medicine cGMP manufacturing and process development business, to accelerate the discovery and development of advanced therapeutics.
Nosis Biosciences, an AI-driven biotechnology company specializing in RNA therapeutics for challenging cell types, announced it entered into a Research Collaboration and Option Agreement with Daiichi Sankyo (Daiichi Sankyo Chemical Pharma Co., Ltd) (a global pharmaceutical company and the second-largest pharmaceutical company in Japan), to leverage Nosis’ Connexa™ platform, an advanced AI-powered drug design and delivery system (pipeline), to enable therapeutic access to extrahepatic cell types across multiple vital organs affected by chronic diseases (Nosis Biosciences Announces Research Collaboration and Option Agreement with Daiichi Sankyo to Develop Cell-Targeted RNA Medicines).
Pheiron and Ribocure are excited to announce that the first two strategic collaboration milestones have been successfully achieved, marking a significant leap forward in our AI-assisted drug development for RNA therapeutics (Advancing siRNA Drug Discovery and Development with AI). Leveraging Pheiron’s advanced AI platform, PheironGPS, multiple candidate targets have systematically been profiled by integrating large-scale human multi-omic data with cutting-edge AI analytics. ➡️ Pheiron GmbH in Berlin founded in 2023 is the GPS for drug development. Their platform uses AI and large-scale human genetics (thousands of proprietary disease signatures capturing disease onset and progression in high-resolution) to infer causal insights on targets and populations. Pheiron has entered a strategic collaboration with Ribocure Pharmaceuticals AB and Suzhou Ribo Life Science Co., Ltd to accelerate the discovery and development of RNA-based therapeutics with evidence from large human populations generated through PheironGPS.
Tempus AI, Inc (NASDAQ: TEM) just announced it has acquired Deep 6 AI, a leading AI-powered precision research platform for healthcare organizations and life sciences companies (Tempus Announces Acquisition of Deep 6 AI). The Chicago-based ➡️ Tempus—that specializes in AI and precision medicine and has one of the world’s largest libraries of clinical and molecular data—provides integrated solutions such as sequencing, companion diagnostics, clinical trial solutions, data collaborations and biological modeling etc by applying AI (link). ➡️ Deep 6 AI, utilizes AI to mine medical records to accelerate finding and recruiting patients for clinical trials within minutes. Its AI-powered software matches patients to clinical trials by mining real-time structured and unstructured electronic medical record (EMR) data across a broad ecosystem, which includes academic medical centers, National Cancer Institute (NCI)-Designated Cancer Centers, and NCI Community Oncology Research Programs. Deep 6 is integrated with over 750 provider site locations spanning more than 30 million patients, which will materially add to Tempus’ existing network.
MeiraGTx Holdings plc (Nasdaq: MGTX), a vertically integrated, clinical stage genetic medicines company, entered in collaboration with Hologen Limited, a world-leading developer of multi-modal generative AI foundation models of real-world clinical data for clinical medicine and pharmaceutical drug development (MeiraGTx Enters into a Strategic Collaboration with Hologen AI to Expedite Phase 3 Development of AAV-GAD for Parkinson’s Disease and Industrialize MeiraGTx’s Proprietary Manufacturing Process). ➡️ MeiraGTx and ➡️ Hologen are forming a joint venture, called ➡️ Hologen Neuro AI Ltd and Hologen will contribute its proprietary multi-modal generative foundation models (LMMs) to the joint venture. In forming this joint venture, MeiraGTx and Hologen have created the first neuro-AI clinical drug development company in which pioneering technologies from both companies will be deployed to transform the discovery and development of therapies targeting CNS circuitry in neurodegenerative and neuropsychiatric disorders.
AsedaSciences and Excelra’s co-marketing agreement aims to deliver a unified data platform for small molecules, streamlining drug discovery workflows (AsedaSciences and Excelra Announce a Co-Marketing Agreement to Deliver a Unified Data Platform for Small Molecules). This strategic agreement with Excelra aims to co-market and expand the reach of AsedaSciences' groundbreaking 3RnD® platform, which will be strengthened with Excelra's 10M+ chemical structure and growing GOSTAR® database. This sets a new benchmark for information consolidation and providing content accessibility through a single, easy-to-use interface for the scientific community. ➡️ Excelra Knowledge Solutions Private Limited is a leader in biopharma data and analytics, uniquely blends scientific understanding with technological expertise. This dual mastery empowers innovative drug discovery and development, transforming research data into actionable insights for groundbreaking discovery. ➡️ Aseda Sciences Inc was established with the vision of becoming a leader in optimized drug discovery by providing multi-parametric data driven, high content, high throughput phenotypic screening services to the pharmaceutical and bio-pharmaceutical industry for the assessment of compound toxicity or mechanism of action.
Salt AI announced a partnership with the Ellison Medical Institute, leveraging expertise in life sciences and AI (Salt AI announces new life sciences offering and ongoing partnership with Ellison Medical Institute). ➡️ Salt AI (Development Engine for Life Sciences) offers life sciences teams access to best-of-breed AI workflows to solve the hardest biology problems. Model-agnostic, Salt's platform prioritizes data integrity and reliability while providing a robust interface for cross-functional collaboration and rapid interchangeability of best of breed AI models.
Funding & Grants 💰
Significant investments are fueling innovation in AI and biotech:
Proscia just secured $50M in funding, bringing its total raised to $130M. The round was led by Insight Partners. This investment follows Proscia’s record-breaking growth in 2024 (Proscia Raises $50M To Advance AI-Driven Pathology And Deliver The Future Of Precision Medicine). ➡️ Proscia, a global leader 🎩 in AI-enabled digiatal pathology solutions for precision medicine, now counts 16 of the top 20 pharmaceutical companies among its users and is on track for 22,000+ patients to be diagnosed on its Concentriq® software platform each day. Proscia’s Concentriq® digital pathology platform works with every leading whole slide imaging (WSI) scanner on the market today, and its open API centralizes all your pathology data in one place making it more accessible and driving insight into your routine practice. Moreover, users can launch any AI application from the Concentriq platform and power computational pathology at every organization. The Concentriq Embeddings program relies on AI models to deliver numerical representations of whole-slide images, simplifying the process of feeding them into algorithms for computing cancer risk scores and image classifiers. As a result, Proscia is rewiring pathology, shifting it from microscope to data-rich images, to overcome current challenges, unlock new insights, and shape the future of care.
Insilico Medicine Inc just raised $110M, achieving unicorn status, and is exploring a potential IPO in Hong Kong. The company is using AI to accelerate drug discovery and development pipelines, alongside new steps in its new push into laboratory robotics (Insilico raises $110m to develop drug pipeline and AI platform). Moreover, Insilico Medicine is now considering an initial public offering after securing funding at a valuation of more than $1 billion, reflecting growing interest in the potential use of AI to discover ground-breaking new drugs (AI Drug Startup Insilico Weighs Hong Kong IPO After Funding). ➡️ Insilico Medicine is a global leader in the field of AI-driven drug discovery and aging research. Founded in 2014, Insilico Medicine has pioneered the use of generative adversarial networks (GANs), reinforcement learning, and other advanced AI technologies to accelerate the development of novel therapeutics and unlock the mysteries of human aging. Insilico Medicine uses AI algorithms (Pandomics, Chemistry42, InSilicoBio, Aging.AI) to predict molecular properties, design new compounds, and identify potential drug candidates for various diseases. Its platforms can screen billions of molecules in a fraction of the time required by conventional methods. By using generative models like GANs, Insilico creates entirely new chemical structures with desired therapeutic properties. This approach allows the company to explore vast chemical spaces that were previously inaccessible. A publicly available tool (Aging.AI) estimates biological age by using blood test results and demonstrates Insilico Medicine's commitment to democratizing access to aging research (biomarkers of aging, and developing interventions to delay or reverse age-related decline, "aging clocks" based on omics data) and empowering individuals to track their health over time.
Lila, a "scientific super-intelligence" firm, was launched with $200M in funding, signaling investor confidence in AI-driven scientific advancements (Scientific 'superintelligence' firm Lila launches with $200m). The latest biotech to emerge from the Flagship Pioneering incubator–➡️ Lila Sciences–has emerged to advance its vision of AI-powered, fully autonomous labs for the life sciences and other industries. The goal 🥅 is to combine AI, automation, and robotics to create 'self-driving' labs that run experiments, collect and analyze data, and make recommendations on the next stage in the experimental process, effectively pursuing the scientific method at scale.
The UK 💂♂️ government awarded £82.6M to AI drug discovery companies, reinforcing national support for AI in healthcare (UK awards £82.6m to AI drug discovery companies).
GenPharmaChain, a pioneering blockchain-based AI pharmaceutical research platform, has successfully secured $500,000 in its pre-seed funding round. The investment was led by M2M Capital (@M2MCapital), FNODE Venture (@FNodeVenture), Pharma Lab (@Pharmalab700), and Avatia Pharmaceutical Company, with strategic support from Google, AWS and several angel investors from top-tier Web3 exchanges and VC (GenPharmaChain Secures $500,000 in Pre-Seed Funding to Revolutionize AI-Driven Drug Development). GenPharmaChain’s goal is to build a global 🌍 decentralized platform where everyone can contribute their computational resources to help with complex protein structure prediction and functional studies.
Market & Strategic Moves 💹
Several trends indicate sustained growth in AI-driven healthcare:
The virtual nutrition 🥑 coach app market is poised for growth by 2025, driven by consumer demand for personalized health solutions (Virtual Nutrition Coach App Market Strategic Drivers for 2025).
NVIDIA’s AI technologies are accelerating the digital transformation of MedTech, with three key stocks gaining attention from investors (NVIDIA's AI Speeds Up MedTech's Digital Boom: 3 Stocks in Focus).
The latest study released on the Global Drug Discovery Informatics Market by USD Analytics Market evaluates market size, trend, and forecast to 2034 (Drug Discovery Informatics Market Is Booming So Rapidly | Schrödinger, Certara, Evotec).
The global computational 💻 biology 🧬market size is calculated at USD 7.18 billion in 2025 and is forecasted to reach around USD 21.95 billion by 2034, accelerating at a CAGR of 13.22% from 2025 to 2034 (Computational Biology Market Size to Hit USD 21.95 Billion by 2034).
The India 👳 AI in healthcare market is projected to reach $6.5 billion, driven by increasing adoption of AI tools in diagnostics, drug discovery, and patient care (India AI in healthcare market to hit USD 6.5 billion).
F. Hoffmann-La Roche AG, one of the world’s largest biotechnology companies, will establish an Innovation Center on Harvard’s Enterprise Research Campus in Allston. The Swiss multinational company is leasing 30,000 square feet of space from the ERC’s 1.9 million square feet, and plans to eventually employ up to 500 employees (Biotech Giant Roche Launches Innovation Center on Harvard’s Allston Research Campus).
Investor Insights 🕵️
🌡️ Absci (ABSI) Expected to Announce Quarterly Earnings on Tuesday.
How Artificial Intelligence Benefits Health Care: 🌡️ 10 AI Healthcare Stocks to Watch.
🌡️ Tempus Stock May Rise on the Acquisition of AI-Driven Deep 6 AI.
🌡️ Recursion Pharmaceuticals: Stretched Thin And Chasing Unicorns
Leadership 🎩
Industry veterans are joining startups and biotech firms to guide innovation:
Mikael Dolsten, former Pfizer R&D chief, joined Immunai’s board, bringing decades of experience to AI-driven immunology research (Former Pfizer R&D chief Mikael Dolsten joins Immunai’s board of directors). ➡️ Immunai (Immunai Inc) is building an AI model of the immune system to facilitate the development of next-generation immunomodulatory therapeutics by combining immunomics and ML. Immunai’s rich clinico-genomic data assets are harmonized in AMICA™, the Annotated Multi-Omic Immune Cell Atlas. Fed by massive scale curated public -omics data and proprietary cohorts and experiments, AMICA™ covers 10s of millions of cells across hundreds of disease settings and is growing rapidly.
Challenges and Risks 🕹️
Despite the optimism, challenges remain:
Ensuring the reliability of AI models (e.g., addressing hallucination) remains a critical hurdle. Folding-based end-to-end chemical drug design tackles hallucination issues in AI models, improving reliability in molecular predictions (Folding-Based End-To-End Chemical Drug Design with Uncertainty Estimation: Tackling Hallucination in the Post-GPT Era). ▶️ Llama-Gram represents a promising advancement in AI-driven chemical drug discovery, grounded in the chemical principle that molecular structure determines properties, and seeks to address the hallucination issues 😵 of traditional large language models by integrating protein folding embeddings, graph-based molecular representations, and uncertainty estimation to better capture the structural complexities of protein–ligand interactions. By leveraging the frozen-gradient ESMFold model and a Graph Transformer variant, Llama-Gram aims to enhance predictive accuracy and reliability through grouped-query attention and a Gram layer inspired by support points theory. By incorporating protein folding information, the model demonstrates competitive performance against state-of-the-art approaches such as Transformer CPI 2.0 and Graph-DTA, offering improvements in compound–target interaction. Llama-Gram provides a scalable and innovative chemical theory that could contribute to accelerating the chemical drug discovery process.
Until next time 🪴