In today’s rapidly evolving healthcare landscape, the intersection of cloud computing and the de novo design of therapeutic antibodies with generative AI stands as a beacon of transformative potential. Traditionally, the journey from drug discovery to approval has been a marathon fraught with complexity, costing upwards of $3 billion and spanning an average of 15 years, with an approval rate hovering around a mere 1%. However, the dawn of de novo antibody design, powered by generative AI, heralds a new era in drug development.
By utilizing the immense computational capabilities of cloud technologies, coupled with the innovative prowess of generative AI, we’re witnessing the birth of a new paradigm in drug discovery and development. This approach not only increases the precision of therapeutic antibodies, raising their success rates significantly, but also may help save billions of dollars and years of labor that traditionally have been sunk into the development pipeline. In essence, the integration of cloud-based generative AI into the de novo design of antibodies is set to revolutionize how we develop life-saving treatments, turning the tide toward a future where efficacious drugs are brought to market faster, with greater accuracy, and at a fraction of the current costs.
In a trailblazing collaboration, Silica Corpora and Oracle Cloud Infrastructure (OCI) are revolutionizing de novo drug design. Silica Corpora trained cutting-edge generative AI models to redesign from scratch the most crucial region of an approved therapeutic antibody, while OCI provided essential technical support with its robust cloud infrastructure featuring NVIDIA GPUs. This partnership highlights the transformative potential of combining AI innovation with advanced cloud capabilities, setting a new benchmark for pharmaceutical development.
Silica Corpora is at the forefront of revolutionizing therapeutic antibody design, employing cutting-edge generative AI for de novo creation of novel antibodies. Its mission is to enhance the overall precision of therapeutic antibodies through innovative design techniques, aiming to deliver safer and more effective therapies to patients worldwide.
Silica Corpora chose OCI for its OCI BM.GPU4.8 shape within its bare metal servers powered by 8 x NVIDIA A100 Tensor Core GPUs, offering the type of high performance critical for the intensive computational demands of generative AI models in drug design. This decision was driven by the need for robust, scalable computing power to efficiently handle complex algorithms and massive datasets. OCI’s cutting-edge infrastructure provided the optimal environment for Silica Corpora’s AI-driven projects, delivering fast, reliable, and effective processing capabilities.
In the results phase of its groundbreaking endeavor, Silica Corpora successfully trained several protein large language models (pLLMs), each boasting around 5 billion parameters, to enhance performance specifically for the therapeutic antibody in question. This meticulous training resulted in the identification of several dozen promising variants of the therapeutic antibody, marking a significant milestone in the quest for more effective treatments.
The next critical step involves in-vitro testing of these selected antibody variants. This phase is essential for validating the efficacy and safety of the antibodies, ensuring that only the most effective candidates move closer to clinical application. The success of this selection process underscores the precision and effectiveness of Silica Corpora’s AI-driven approach in identifying viable therapeutic candidates from a vast pool of possibilities.
Six novel pLLMs were trained, each with about 1 billion parameters, for generation of antibody binding regions. In addition, 15 novel pLLMs were also trained, each with about 0.5 billion parameters, for selection of better candidates among the generated ones. Silica Corpora utilized all eight NVIDIA A100 GPUs on OCI Compute BM.GPU4.8 for this LLM training. The computational resources of that scale were crucial for this project.
Using those models, the customer generated 40 variants of trastuzumab with redesigned regions and sent them to their wet lab partner for the wet lab validation. These results show that OCI can help reduce the time needed to train pLLMs with a simple ethernet network architecture. OCI Compute bare metal instances powered by NVIDIA GPUs offer ultra-low-latency and bare line-rate network performance.
Throughout this process, OCI’s performance has been a cornerstone of success. The high-performant BM.GPU4.8 instances powered by NVIDIA A100 GPUs in OCI’s bare metal servers provided the necessary computational power and speed, handling extensive data and complex algorithms with ease. This technological support has not only facilitated the efficient training of pLLMs, but also accelerated the overall drug design process, highlighting OCI’s instrumental role in pushing the boundaries of medical research and pharmaceutical development.
For more information, see the following resources:
I bring to the table more than 19+ years’ experience, with 2 years in end-to-end deployment of generative AI, LLM’s AI/ML in OCI, and the rest of the year with high-performance computing infrastructure and solution development across a variety of cloud platforms. Experienced in data centre and automotive, financial, and healthcare domains, my area of expertise includes HPC infrastructure architecture, design and solution development with application profiling and benchmarking, scientific application deployments, and business systems.
With 12+ years of experience in applying artificial intelligence, machine learning to solve complex problems in biopharma, Tim Ermak is a co-founder and CTO of Silica Corpora, a startup that leverages generative AI to design antibodies de novo for more effective therapeutics. He gained a PhD-equivalent experience in bioinformatics and holds an MBA in Life Science Management degree, which equip him with a unique combination of technical and business skills for leading a cutting-edge venture in the life science sector.
Tim's mission is to bring innovation and excellence to the field of antibody engineering and design, and to create value and impact for patients and society with Silica Corpora.
Carmen is a Spanish account manager for cloud technology based in The Netherlands, working for the German SMB market. She has been working at Oracle for 2 years and loves connecting ideas with state-of-the-art technology. She believes in the power of technology in the service of people as a means to improve our community.
Previous Post
Next Post