Methods and apparatuses for generating peptides by synthesizing a portion of a design space to identify peptides having non-canonical amino acids
Inventors
Lee, Francis • STECKBECK, Jonathan D. • Holste, Hannes
Assignees
Publication Number
US-12006541-B2
Publication Date
2024-06-11
Expiration Date
2041-08-13
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
In one aspect, a computer-implemented automated flow synthesis platform configured to use an artificial intelligence (AI) engine is disclosed and includes a reaction chamber configured to synthesize a sequence, detectors configured to monitor the synthesis of the sequence in the reaction chamber, wherein the synthesis uses an automated flow process, and a computing device communicatively coupled to the detectors. The computing device receives measurements from the one or more detectors, wherein the measurements comprise a spectral profile at each coupling of each amino acid in the sequence, trains, using training data comprising the measurements, machine learning models to determine a synthesizing recipe that enables the sequence to be synthesized, wherein the synthesizing recipe comprises parameters used during the automated flow process to synthesize the sequence, and controls, using the synthesizing recipe, the synthesis of the sequence in the reaction chamber.
Core Innovation
The invention relates to a computer-implemented automated flow synthesis platform that leverages an artificial intelligence (AI) engine and one or more machine learning models to optimize the synthesis of peptide sequences, including those with non-canonical amino acids. The system comprises a reaction chamber for synthesis, detectors to monitor the synthesis in real time by capturing spectral profiles at each amino acid coupling, and a computing device connected to these detectors. The computing device receives measurement data, utilizes these inputs to train machine learning models, and determines synthesizing recipes—comprising parameters such as temperature, solvents, protection groups, resin anchors, and more—that are used to control the automated flow process.
This approach addresses the inefficiencies and limitations present in conventional drug discovery and peptide synthesis techniques, particularly the difficulty in efficiently searching and optimizing within an enlarged chemical design space and handling the unpredictable chemical behavior of non-canonical amino acids. Traditionally, such processes are labor-intensive and require costly trial-and-error experiments that consume significant resources. The invention automates both the generation of candidate drug sequences and their synthesis via a fully integrated flow system, using AI-driven real-time feedback and optimization to improve yield and functionality, minimize undesirable side reactions, and reduce waste.
The AI-enabled platform not only learns associations between input sequences, synthesis conditions, and resulting chemical reactions—including fidelity and side reactions—but can continuously retrain its models based on spectral and outcome data. This allows the system to adaptively optimize recipes for synthesizing both canonical and non-canonical amino acid-containing peptides, enabling scalable, efficient, and robust peptide drug development and manufacturing.
Claims Coverage
The patent contains multiple independent claims covering the primary inventive features of an AI-enabled flow synthesis platform for synthesizing peptide sequences. The main inventive features described in the independent claims are as follows:
AI-driven automated flow synthesis platform with real-time spectral monitoring
A computer-implemented automated flow synthesis platform that integrates: - A reaction chamber configured for sequence synthesis. - One or more detectors capturing a spectral profile at each amino acid coupling during synthesis in the reaction chamber. - A computing device receiving real-time measurements from these detectors. - Use of one or more trained machine learning models to determine a synthesizing recipe, with parameters for controlling the flow synthesis process, based on the received measurements. - Control of the synthesis process in real time using the determined recipe.
Use of machine learning models to optimize synthesizing recipes based on spectral data
Machine learning models are trained on spectral profile data acquired at each amino acid coupling, allowing: - Determination of recipes that optimize synthesis for different sequences, including those containing non-canonical amino acids. - Generation of recipes that specify parameters such as temperature, solvents, protection groups, resin anchors, and more, used to control at least the reaction chamber.
Layered machine learning model architecture for recipe generation
A layered machine learning approach where: - The first layer receives amide coupling data, coupling fidelity, and spectral profiles, and generates a subset of recipe parameters. - The second layer receives the output of the first layer, along with data about multiple amide couplings, to generate a further subset of recipe parameters, collectively forming an optimized synthesizing recipe.
Control and feedback loop using recipe and outcome association
The computing device implements a feedback loop by: - Associating the synthesizing recipe with the observed chemical reaction based on spectral data. - Adjusting, retraining, or selecting recipes to avoid undesired reactions and improve synthesis efficiency and outcomes.
The inventive features cover an integrated, AI-powered flow synthesis system that continuously learns from process data to optimize synthesizing conditions and improve the synthesis of peptides, including those with non-canonical amino acids. The claims encompass the hardware setup, machine learning-based recipe generation, real-time feedback, and adaptive control based on spectral analytics.
Stated Advantages
Enables efficient and scalable synthesis of peptides and peptidomimetics, including those with non-canonical amino acids, by combining AI and automated flow chemistry.
Reduces computational complexity and resource consumption in searching large design spaces for optimal synthesis recipes, saving time, processing power, and materials.
Minimizes undesirable side reactions and waste by enabling real-time recipe optimization and feedback using spectral data collected during synthesis.
Automates and speeds up the development of therapeutic peptides and compounds, providing superior candidate generation and optimized synthesis compared to conventional batch and manual methods.
Improves yield and product quality through adaptive quality control and optimization based on structural and functional screening data.
Provides economic benefits by reducing the consumption of reagents and synthesis time, and improving efficiency and outcomes in peptide manufacturing.
Documented Applications
Peptide therapeutics and drug discovery, including antimicrobial, anti-infective, anti-cancer, immunomodulatory, cytotoxic, neuromodulatory, and other medical applications.
Development and synthesis of peptides containing non-canonical amino acids for research, clinical, or therapeutic purposes.
Animal health and veterinary medicine, including treatment of animal diseases such as bovine mastitis.
Industrial applications such as anti-biofouling and generation of optimized control sequences for machinery.
Applications in new therapeutic indications including eczema, inflammatory bowel disease, Crohn’s disease, rheumatoid arthritis, asthma, autoimmune diseases, oncology treatments, and inflammatory disease processes.
Use in high-throughput peptide synthesis and development pipelines requiring rapid optimization and quality control.
Interested in licensing this patent?