In a groundbreaking shift towards advancing medical research, the technology giant Salesforce, renowned for its cloud-based sales software and ownership of Slack, embarked on an innovative project known as ProGen. This ambitious venture aimed to harness the power of generative AI for the design of proteins, potentially transforming the landscape of medical treatment discovery by offering a more cost-effective alternative to traditional methodologies. As detailed in a January 2023 blog post by the researchers involved, ProGen heralded a new dawn in the pursuit of medical breakthroughs, albeit its journey from concept to commercial viability remained uncertain until a pivotal development occurred.
The research undertaken by ProGen culminated in a significant publication within the prestigious journal Nature Biotech. This study demonstrated the AI’s capability to successfully generate 3D structures of artificial proteins, marking a milestone in the project’s academic achievements. However, despite the research’s importance, ProGen’s application within Salesforce and the broader commercial sphere appeared limited. That is until Ali Madani, a key figure behind ProGen, initiated a bold step forward.
Madani founded Profluent with the vision of translating the conceptual prowess of ProGen into a practical tool for the pharmaceutical industry. During an interview with Digital Market Reports, he outlined Profluent’s mission as a revolutionary approach to drug development. This strategy begins with the identification of patient and therapeutic needs, subsequently crafting tailor-made treatment solutions through the innovative use of AI-designed proteins.
Madani’s insights into the similarities between natural languages, such as English, and the ‘language’ of proteins, have been instrumental in Profluent’s foundational principles. Proteins, essentially chains of amino acids performing diverse bodily functions, can be analogized to words forming coherent paragraphs. This analogy opened avenues for utilizing generative AI to predict and create new proteins with unique functionalities.
Together with Alexander Meeske, an assistant professor of microbiology at the University of Washington and co-founder of Profluent, Madani aims to extend this concept into the realm of gene editing. This approach seeks to address the limitations of existing gene editing techniques and natural proteins in treating genetic diseases by developing optimized, custom-designed gene editors.
Key Milestones in Profluent’s Journey
| Milestone | Description |
|---|---|
| Launch of ProGen | Salesforce’s project leveraging AI for protein design. |
| Nature Biotech Publication | Academic validation of AI’s capability to design protein structures. |
| Foundation of Profluent | Transition from research project to startup aiming at revolutionizing drug development. |
| Collaboration Initiatives | Partnerships with pharmaceutical companies for developing genetic medicines. |
Profluent’s endeavors are part of a larger trend where companies and research institutions leverage generative AI for protein prediction and drug discovery. Notable initiatives include Nvidia’s MegaMolBART, Meta’s ESM-2 model, and DeepMind’s AlphaFold, each contributing to the field by predicting drug targets, protein sequences, and complete protein structures with unprecedented speed and accuracy.
Profluent differentiates itself by focusing on the expansive potential of generative AI, training models on data sets comprising over 40 billion protein sequences. This approach not only aims at creating new proteins but also fine-tuning gene-editing systems in collaboration with pharmaceutical companies, potentially streamlining the path to regulatory approval for new treatments.
The pharmaceutical industry is notoriously time-consuming and costly, with drug development often spanning 10-15 years and costing up to $2.8 billion. Profluent’s strategy represents a paradigm shift towards intentional design of treatments, moving away from the serendipitous discoveries that have historically characterized medicine development.
- Reversing Drug Development Paradigm: Focusing on patient needs to design custom-fit treatments.
- Language of Proteins: Utilizing generative AI to predict new proteins for therapeutic applications.
- Gene Editing: Addressing genetic diseases with optimized, custom-designed gene editors.
- Collaborative Model: Partnering with pharmaceutical companies to develop genetic medicines.
- Cost and Time Efficiency: Aiming to significantly reduce the time and capital required for drug development.
Based in Berkeley and supported by a 20-person team, Profluent has garnered attention and financial backing from venture capital firms like Spark Capital, Insight Partners, and others, raising a substantial $35 million in a recent funding round. The involvement of notable figures such as Google’s chief scientist Jeff Dean underscores the promising potential of Profluent’s technology.
Looking forward, Profluent is set to focus on enhancing its AI models through expanded training datasets and forging new partnerships.
Related News:
Featured Image courtesy of DALL-E by ChatGPT