We teach AI machines protein physics
So that they can reason with protein data just like human researchers do, and design experimentally testable hypotheses
Mission—Critical solutions for
Big Pharma and Biotech
We combine computational biophysics, non-linear modeling, statistical approaches to structural protein biochemistry and progressive AI to explore combinatorial mutagenesis landscapes of any protein, irrespective of its class and application.
Protein Developability Assessment
Even the most successful protein engineering campaigns achieve only 15% success rate. Effectively, laboratories may spend >85% of their time and budget designing protein mutants that fail. Our developability graphs provide unique insight into protein mutational landscapes and help with planning of large-scale engineering campaigns.
See An Example
Automated Thermostability Engineering
Our protein family-specific AI deployments explore trillions of possible mutations to unravel the smallest population of stabilizing, rare amino acid substitutions that won't compromise enzymatic activity nor decrease binding affinity.
AI — Assisted Survey of "Dark Proteome"
Recent breakthrough studies have established that approximately one-third of the proteins in the human proteome are entirely disordered (IDPs) or contain long intrinsically disordered regions (IDRs). Our AI tools give us a unique possibility to model proteins in their non-canonical states and perform genome-wide surveys in a search for novel, functional disordered proteins of possible medical relevance.
Search for an IDP
Combinatorial Biochemistry of Peptides
We have developed hybrid Generative Adversarial Neural Networks (GANs), which work in unison with real-time biophysical discriminators to enable human-driven design of small peptide sequences based on predefined biophysical properties, e.g. a propensity to form canonical and disordered conformations.
See Live Demo
Biophysics meets Deep Learning
We have developed scalable and convergent proprietary evolutionary approaches to Deep Learning architecture design, which seamlessly blend with existing protein biophysics. Through hybrid biophysics — AI deployments, we mitigate the biggest limitation of Machine Learning in structural biochemistry — insufficient experimental data.
Specific
Our hybrid AI deployments capture intricate relationships between mission — critical biophysical properties for customer designated protein families.
Non — Obvious
Our stochastic optimization algorithms search and flag the most promising, non-obvious mutations, which escape the attention of human researchers and benchmark methods.
Exclusive
Our AI deployments are designed and tailored to very specific project needs providing a sought — after exclusivity and "unfair advantage" over competitors in Big Pharma and Biotech space.
AI Designs Antibody
Our stochastic solvers were tasked with finding the most soluble (% PEG) variants of an mAb antibody line developed by MedImmune, an AstraZeneca company. We have identified mutants with highly reduced self-aggregation propensity and detected rare amino acid occurrences, which lead to specific protein-protein interactions.
*The remainder of mAb sequence has been obscured for data confidentiality reasons.
10,240,000,000,000 Mutants per Hour
Scaled Across 64 CPUs and 4 GPUs
Fully Customized User Experience and Graph Overlays
Evolution Through 3 Distinct Biophysical Features
Continuous Learning Through User Supplied Solubility Data
Solution Found in 25 min. vs 3+ Months of Experiments
Less is More
Say goodbye to public cloud deployments, software updates, awkward User Interfaces, endless list of Docker images, unread manuals, costly personal training and unstructured data. Our role is to help you navigate protein design complexity with an in-silico solution that augments your research and automates the most laborious aspects of your protein engineering exploits while extracting maximum value from your data.
Every AI deployment includes
Mission — Critical Stochastic Solver For Your Protein Family
Realtime Solution Preview and Mutant Selection
Fully Customizable User Interface and Graph Overlays
Horizontally Scalable and Expandable AI Engine
Continuous Learning AI Through Automated Data Uploads
Software and Hardware Support
AI in the Cloud
Your protein — specific AI is deployed in an encrypted container, receives CPU and GPU allocation, and we provide your team with an instant access and customized User Experience.
Single — Protein Projects
Excellent for Small Size Businesses
Basic Deployment Cost
AI in your Lab
Your protein — specific AI is deployed in an encrypted container and is installed in your lab on a barebone HPC solution for maximum performance and security.
Multi — Protein Projects
The Ultimate Choice for Large Enterprises
Built to Meet the Most Stringent Security Standards
Careers
Should you be interested in joining our team, please use our application form.
*At this stage your CV is not necessary.
*Please don't send us job requests via email.
Partnering & Info
Peptone Ltd. is a privately owned, next generation AI-first synthetic protein biology company registered and incorporated in United Kingdom.
Northcliffe House, Young St,
London W8 5EH, United Kingdomhello@peptone.io
* Please direct all investment and partnering enquiries to our global email address at hello@peptone.io.