Drug Development with AI-Designed Proteins

In a groundbreaking fusion of artificial intelligence and molecular biology, Profluent, a startup backed by significant industry players like Jeff Dean and venture capital firms like Spark Capital and Insight Partners, is paving the way for a new era in drug development. Leveraging the power of generative AI, Profluent aims to expedite the discovery and design of medicines by creating custom-fit protein solutions tailored to individual patient needs.

The Genesis: ProGen and the Potential of AI in Protein Design

The foundation for Profluent's innovative approach was laid by Salesforce's ProGen project. ProGen was Salesforce's ambitious venture into the realm of protein design using generative AI. The project aimed to utilize AI to design proteins, potentially revolutionizing the way medical treatments are discovered and developed. A key achievement of ProGen was the successful creation of 3D structures of artificial proteins, as demonstrated in research published in the journal Nature Biotech.

However, despite the promising results and scientific breakthroughs, ProGen did not transition into commercial applications. This gap between academic research and practical implementation was the catalyst for Profluent's inception.

Profluent: Bridging the Gap Between AI and Pharmaceutical Innovation

Ali Madani, one of the key researchers behind ProGen, recognized the untapped potential of AI-designed proteins and founded Profluent with a mission to "reverse the drug development paradigm." The startup aims to start with patient and therapeutic needs and work backward to create custom-fit treatment solutions.

"Many drugs, such as enzymes and antibodies, consist of proteins," Madani explained. "Ultimately, this is for patients who would receive an AI-designed protein as medicine."

Drawing parallels between natural language and the "language" of proteins, Madani discovered that proteins, which are chains of bonded-together amino acids used by the body for various purposes, can be treated like words in a paragraph. By feeding data about proteins into a generative AI model, it is possible to predict entirely new proteins with novel functions.

Advancing Gene Editing with AI

Going beyond protein design, Profluent aims to apply this technology to gene editing. Many genetic diseases cannot be addressed using proteins or enzymes directly sourced from nature. Moreover, existing gene editing systems often suffer from functional trade-offs that limit their effectiveness. Profluent seeks to overcome these limitations by optimizing multiple attributes simultaneously to create custom-designed gene editors tailored for each patient.

The AI Revolution in Protein Prediction

Profluent is not alone in harnessing the power of AI for protein prediction and drug discovery. Nvidia introduced MegaMolBART, a generative AI model trained on a dataset of millions of molecules to search for potential drug targets and forecast chemical reactions. Meta developed ESM-2, a model trained on protein sequences, which enabled the prediction of sequences for over 600 million proteins in just two weeks. Additionally, DeepMind's AlphaFold system has achieved remarkable speed and accuracy in predicting complete protein structures, surpassing older, less sophisticated algorithmic methods.

Profluent is training its AI models on massive datasets containing over 40 billion protein sequences to create new and fine-tune existing gene-editing and protein-producing systems. Rather than developing treatments itself, the startup plans to collaborate with external partners to produce "genetic medicines" with the most promising paths to regulatory approval.

Accelerating Drug Development and Reducing Costs

One of the most significant advantages of Profluent's approach is the potential to dramatically reduce the time and capital typically required for drug development. According to industry group PhRMA, it takes an average of 10-15 years and costs between several hundred million to $2.8 billion to develop a new drug from initial discovery through regulatory approval.

"Many impactful medicines were accidentally discovered, rather than intentionally designed," Madani noted. "Profluent's capability offers humanity a chance to transition from accidental discovery to intentional design of our most needed solutions in biology."

Strategic Backing and Future Plans

Based in Berkeley and currently employing 20 people, Profluent has secured substantial backing from venture capital heavyweights, including Spark Capital, which led the company's recent $35 million funding round, Insight Partners, Air Street Capital, AIX Ventures, and Convergent Ventures. Google chief scientist Jeff Dean has also contributed to the platform, lending additional credibility and support.

Looking ahead, Profluent's focus for the next few months will be on upgrading its AI models by expanding the training datasets and acquiring customers and partners. With competitors like EvolutionaryScale and Basecamp Research also investing heavily in protein-generating models, Profluent aims to move aggressively to scale its platform and enable solutions with partners that match its ambitious vision for the future.

"We've developed our initial platform and demonstrated scientific breakthroughs in gene editing," Madani said. "Now is the time to scale and start enabling solutions with partners that match our ambitions for the future."

Conclusion

Profluent represents a paradigm shift in drug development, harnessing the power of AI to design custom-fit protein solutions and accelerate the discovery and development of new medicines. With the backing of industry leaders and a visionary approach to leveraging AI in molecular biology, Profluent is poised to revolutionize the way we approach medical treatments and offer new hope for patients in need.