Weekly Tech+Bio Highlights #8
ALSO: Causal AI and Digital Twins as a New Paradigm for Target Discovery; Toward Ending Placebo Arms in Clinical Trials; Largest Open-Source Foundation Model for Pathology; The Power of TNIK
Hi! I am Andrii Buvailo, and this is my weekly newsletter, ‘Where Tech Meets Bio,’ where I talk about technologies, breakthroughs, and great companies moving the biopharma industry 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!
News Highlights
💰 Cambridge, Massachusetts-based Flagship Pioneering raises $3.6 billion to develop 25 breakthrough companies in human health, sustainability, and AI. This expansion, supported by strategic partnerships with Pfizer, Samsung, and Thermo Fisher, increases Flagship's capital pool to $10.9 billion. The company also expands globally with new hubs in London and Singapore to foster international collaborations and talent acquisition.
🧬 Komo Biosciences launches with new integrase technology from the University of Hawaii, aiming to advance gene insertion techniques. The startup, backed by seed funding from Genesis BioCapital and others, will operate in Honolulu and Cambridge, MA.
💰 Canadian biotech Aspect Biosystems secures $72.75 million from the Canadian and British Columbia governments to advance its bioprinting technology for developing implantable tissue therapeutics, supporting a $200 million multi-year project.
🚀 Biostate AI launches from stealth with Total RNA sequencing and Copilot for RNAseq data analysis, partnering with academia and biotech firms. The startup raised over $4M to enhance multiomic data collection and AI training, aiming to reduce costs and improve biological data acquisition.
🔬 Kymera Therapeutics announces Sanofi's decision to expand Phase 2 trials of KT-474 (SAR444656) for Hidradenitis Suppurativa (HS) and Atopic Dermatitis (AD) following a positive interim review of safety and efficacy data, aiming to accelerate progress towards pivotal studies.
💰 Two European biotech VC funds, Venga Ventures and Biovance Capital, launch to support drug development startups in Austria and Portugal, aiming to strengthen local biotech ecosystems and address gaps in translational work.
💰 Illumina acquires Fluent BioSciences to enhance its single-cell analysis capabilities, aiming to make the technology more accessible and advance its multiomics growth strategy. Fluent's PIPseq™ V technology will be integrated into Illumina's product portfolio, offering a complete solution for researchers.
🧬 Iktos acquires Synsight to enhance its AI-driven drug discovery platform with Synsight's MT Bench™ technology for protein-protein and RNA-protein interaction screening.
🤖 Exscientia launches an AWS-powered platform integrating generative AI and robotic lab automation to accelerate drug discovery, aiming to reduce costs and enhance efficiency. This expanded collaboration with AWS supports Exscientia's Design-Make-Test-Learn loops, leveraging advanced AI/ML technologies to streamline the drug development process.
🔬 French startup Bioptimus, having raised $35 million in a seed round this February, has released H-optimus-0, the largest open-source AI model for pathology with 1.1 billion parameters. The model, trained on millions of histopathology images from 500,000 slides, aims to improve cancer diagnostics and detect genetic abnormalities.
🔬 Eligo Bioscience developed a base editor tool that modifies gut bacteria genes in living mice without killing the bacteria, achieving over 90% target gene alteration in E. coli colonies. Published in Nature, the tool uses bacteriophage components for delivery and successfully edited genes related to antibiotic resistance and neurodegenerative diseases.
💰 Courier Health raises $16.5M in Series A funding, led by Norwest Venture Partners, to expand its workforce and enhance its CRM platform for biopharma. The platform aids in understanding patient journeys, data analysis, and workflow management, aiming to improve patient engagement and support for biopharma teams.
🔬 MIMETAS supports argenx's IND application with human organ-on-chip data from their OrganoPlate platform. This collaboration, aligned with the 2022 FDA Modernization Act 2.0, marks a milestone in using advanced in vitro models for drug safety and effectiveness testing.
🚀 ImmuneAge emerges with $2M in funding to rejuvenate the immune system via its ‘Immune Refresh’ approach, expanding and rejuvenating bone marrow stem cells. The Swiss-US company, backed by notable investors, also identified a promising immune-enhancing drug.
🔬 Phesi's study in Bone Marrow Transplantation demonstrates AI-powered digital twins can replace standard of care control arms in clinical trials, potentially streamlining trial processes and reducing patient burden in conditions like chronic graft versus host disease (cGvHD).
🔬 Parallel Bio, led by tech veteran Ari Gesher, is building a fully automated "lights-off" lab to revolutionize drug discovery using proprietary immune data and AI, allowing biologists to focus on experiment design rather than execution. They aim to improve pharma's R&D efficiency with organoid-based models for earlier, high-fidelity human testing, potentially reducing the industry's high failure rate.
Causal AI and Digital Twins as a New Paradigm for Target Discovery
In the realm of drug discovery, Aitia (formerly, GNS Healthcare) harnesses causal artificial intelligence (AI) and Gemini digital twins to decode the hidden genetic and molecular mechanisms driving diseases.
By combining multiomic data with clinical outcomes, Aitia's approach significantly accelerates the identification of novel drug targets for neurodegenerative disorders and cancer. This method represents a technical leap towards precise and effective therapeutic development.
Key Takeaways:
Causal AI and Digital Twins: Aitia’s Gemini digital twins perform billions of virtual experiments to reverse-engineer genetic and molecular interactions, uncovering previously unknown disease mechanisms.
Comprehensive Data Integration: Utilizing extensive datasets from over 25 strategic partnerships, Aitia incorporates DNA sequence variation, gene expression, proteomics, and clinical outcomes to build detailed disease models.
Reduced Attrition Rates: By identifying drug targets with direct causal links to disease mechanisms and accurately predicting clinical outcomes, Aitia’s technology aims to reduce the high attrition rates typically seen in phase 2 clinical trials.
Why It Is Important: Aitia's causal AI and digital twin technology provide a precise, hypothesis-free method for discovering and validating drug targets. This approach increases the accuracy of predicting clinical outcomes, potentially lowering development costs and improving success rates in drug discovery.
Abandoning the Control Arm: Addressing the Ethical Dilemma of Placebos in Clinical Trials
Expanding on the topic of digital twins, Connecticut-based Phesi, a clinical development analytics company, has published a study in Bone Marrow Transplantation demonstrating the potential of digital twins to replace standard control arms in clinical trials.
The study focused on chronic graft versus host disease (cGvHD), which affects 30–50% of the 50,000 cancer patients receiving hematopoietic cell transplantation annually.
Utilizing Phesi's Trial Accelerator platform, which contains data on over 108 million patients, the company constructed a digital twin cohort of 2,042 cGvHD patients from 32 cohorts. This was compared with a standard-of-care cohort of 438 patients from 8 cohorts to model first-line prednisone treatment outcomes.
The digital twin approach eliminates the need to administer placebos to patients who require treatment, addressing the ethical concerns associated with placebo control groups. According to Gen Li, Ph.D., founder of Phesi, the predictive capabilities of digital twins allow for patient outcome forecasting without the use of traditional placebo groups.
Digital twins streamline clinical trials, reducing time and resource requirements. Traditional trial designs often require iterative adjustments based on patient recruitment and response. Digital twins optimize trial design from the outset. Phesi's platform not only creates digital twins but also identifies gaps and misalignments in trial design, enhancing study robustness and efficiency.
Ensuring data accuracy is critical. Phesi employs human verification and validation of all source documents used in constructing digital twins. Regulatory approval remains a crucial step, with the FDA's increasing engagement with AI and machine learning in medicine being a positive indicator for the adoption of digital twin technology.
Let’s Meet In Copenhagen!
We are proud to be a media partner for the upcoming 11th Aging Research & Drug Discovery Meeting (#ARDD2024), one of the worlds most prestigious conferences with a focus of longevity.
Register and mark your calendars for August 26-30; you can participate either on site in Copehnagen, Danmark, or virtually.
Register soon, the conference is reaching full-subscription soon, as the venue (University of Copenhagen) can not host more than 1,000 people on site. ARDD team especially welcomes attendance among the investors and big pharma to ensure that the 40+ startup companies on site get more partnership opportunities.
Targeting TNIK: A Multifaceted Approach to Treating Cancer and Metabolic Diseases
In a new open access article published in Trends in Pharmacological Sciences, authors explore a promising new target, TNIK (Traf2 and Nck-interacting kinase). It has emerged as an important player in various biological processes, making it a promising target for therapeutic intervention in cancer, metabolic diseases, and neurodegenerative disorders. By modulating TNIK activity, researchers aim to address the underlying mechanisms driving these conditions, paving the way for innovative treatments.
Key Takeaways:
1. TNIK in Cancer Treatment:
Mechanistic Insights: TNIK regulates Wnt signaling, promoting cancer cell proliferation, migration, and treatment resistance.
Therapeutic Potential: Inhibitors like NCB-0846 have shown efficacy in preclinical models, impairing cancer growth and enhancing immune-mediated clearance.
2. Metabolic Regulation:
Obesity Resistance: TNIK knockout (KO) mice demonstrate resistance to diet-induced obesity through improved lipid and glucose metabolism.
Insulin Sensitivity: Enhanced insulin signaling and glucose clearance observed in TNIK KO mice suggest potential for treating metabolic disorders.
3. Neurodegenerative Disease:
Alzheimer's Disease: TNIK's interaction with Tau protein highlights its role in neurodegeneration, presenting a novel target for slowing disease progression.
4. Aging and Chronic Conditions:
Fibrosis and Inflammation: TNIK inhibition reduces fibrosis and chronic inflammation, common in aging-related diseases.
Why It Is Important:
Targeting TNIK offers a multifaceted approach to treating some of the most challenging diseases. By addressing key processes like cancer cell proliferation, metabolic dysregulation, and neurodegeneration, TNIK inhibitors have the potential to improve patient outcomes significantly. This approach not only promises to enhance the efficacy of existing treatments but also opens new avenues for therapeutic development, particularly in age-related diseases and chronic conditions. The continued exploration and clinical testing of TNIK inhibitors could lead to breakthroughs in precision medicine, offering targeted and effective solutions for complex health issues.
As Dr. Alex Zhavoronkov, founder and CEO of Insilico Medicine and the co-author of the paper, pointed out in his LinkedIn post: “IMHO, the most impactful way to use AI in biopharma is to discover drugs with massive blockbuster potential and massive indication expansion potential. To do this you need to find targets and generate small molecules with the highest possible safety profile that are likely to work in many chronic diseases.
This was the main idea behind TNIK and its inhibitor 055”.
055, or ISM018-055, is a small molecule inhibitor of TNIK that Insilico Medicine designed using its generative AI platform Pharma.AI and managed to advance into clinical trials within just a couple of years since the start of the project.
The Largest Open-Source Foundation Model for Pathology
French startup Bioptimus has released H-optimus-0, the largest open-source AI foundation model for pathology.
With 1.1 billion parameters, this model is trained on a proprietary dataset comprising several hundreds of millions of images from over 500,000 histopathology slides across 4,000 clinical practices.
This model aims to advance medical diagnostics, particularly in tasks such as identifying cancerous cells and detecting genetic abnormalities in tumors.
Technical Specifications and Training Data
H-optimus-0's model consists of 1.1 billion parameters, facilitating comprehensive analysis and accuracy in pathology diagnostics. The training dataset includes images from over 500,000 pathology slides, encompassing a diverse array of cases. This diversity enables the model to generalize across different diagnostic scenarios.
Diagnostic Capabilities
H-optimus-0 has been evaluated on five tile-level tasks for identifying tissue types and characteristics, and six slide-level tasks for detecting biomarkers or metastasis in various cancer types. The model's performance reportedly aligns with or exceeds existing diagnostic tools.
Open-Source Model and Collaborative Potential
H-optimus-0 is accessible as an open-source model, enabling researchers and developers to collaborate and develop novel digital pathology models. The open-source nature of the model is intended to promote the sharing of methodologies and innovations within the scientific community.
Future Developments
Bioptimus co-founder and CEO, Professor Jean-Philippe Vert, PhD, stated that H-optimus-0 is the first in a series of models planned for release.
Future models will expand the number of pathology images from regions such as Europe, Asia, and Africa, and incorporate other modalities like genomics and proteomics.
The goal is to develop a multiscale foundation model of biology, integrating diverse biological data to facilitate scientific discoveries and accelerate biomedical innovations.
I have compiled a list of AI foundation models in biology research (together with comapnies building them) in one of the previous newsletters: 14 Foundation Models for Biology Research and Chemistry, and also we have discussed the new paradigm of foundation models and some historical trajectory in one of my most read newsletters here: AI Foundation Models in Biotech: New Paradigm.
Last Thursday I’ve sent my musings about some of the important trends and developments that manifested themselves in the first half of 2024. You can read following the link below (for paid paid content).
Andrii, any opinion as to whether Illumina's Fluent acquisition is also an attempt to figure out what Ultima is doing? Harder to get competitive intelligence when your competitor is privately held!