Generative Biology
Profluent
by Profluent Bio, Inc.
Generative AI designing novel proteins and gene editors including the first AI-created CRISPR system
Category
Generative Biology
Founded
2022
Headquarters
Berkeley, CA, USA
Overview
Profluent builds large language models for protein design, treating biological sequences as a language to generate novel functional proteins with desired properties. The company's platform was trained on hundreds of millions of protein sequences and structures and is capable of generating entirely new proteins from scratch — not variants of existing proteins — in a manner analogous to how GPT-4 generates novel text. Profluent's technology spans a wide range of protein classes including enzymes, antibodies, and precision gene editors. In April 2024, Profluent published OpenCRISPR-1, the world's first AI-designed, human-ready gene editing system — a novel Cas9-like protein designed entirely by AI with no natural evolutionary ancestor. The protein functions as a programmable gene editor in human cells, validating the capability of generative AI to design clinically relevant biological tools from first principles. The system was released as open source, making it freely available to researchers worldwide. Profluent's core differentiator is the breadth and novelty of its generative capability — designing genuinely new protein families rather than optimizing existing scaffolds. The company is backed by Salesforce Ventures, Spark Capital, and others, and its team includes alumni from DeepMind, Salesforce Research, and leading academic protein engineering groups. Its open-source commitment (OpenCRISPR-1) combined with a commercial platform positions it uniquely in the generative biology space.
Key Features
Cross-Domain Generation
Unified generative capabilities spanning small molecules, peptides, proteins, and nucleic acids.
Inverse Design
Specify desired biological functions and automatically generate candidate sequences and structures.
Protein Sequence Generation
Generate novel protein sequences with specified structures and functions using deep learning.
Multi-Objective Optimization
Balance efficacy, selectivity, toxicity, ADMET properties, and synthesizability simultaneously.
Novel Molecule Generation
Generative models design molecules with desired properties including efficacy, selectivity, and synthesizability.
Pros & Cons
Pros
- +Multi-objective optimization balances efficacy, selectivity, toxicity, and synthesizability simultaneously
- +Generative models design novel molecules, proteins, and genetic sequences with desired properties
- +Cross-domain generative capabilities span small molecules, peptides, proteins, and nucleic acids
- +Interpretable models reveal structure-function relationships driving design decisions
- +Rapid iteration cycles generate thousands of candidates for experimental validation in hours
- +Inverse design capabilities specify desired functions and generate candidate sequences automatically
Cons
- −Training data biases can limit diversity and novelty of generated biological sequences
- −Generated designs require experimental validation — computational predictions don't guarantee function
- −Synthesizability of generated molecules is not always guaranteed by the model
- −Computational costs for training and inference of large generative models can be substantial
- −Interpretability of generative model decisions remains limited for regulatory submissions
Use Cases
Research Workflow Optimization
AI-powered optimization of research workflows to accelerate discovery timelines and improve reproducibility.
Data Analysis & Insights
Machine learning analysis of complex biological datasets to extract actionable insights and identify patterns.
Collaboration & Knowledge Management
Platform-enabled collaboration across distributed research teams with integrated data sharing and knowledge capture.