Molecular analysis and design using multimodal generative artificial intelligence via multi-agent modeling

Опубликована: Апрель 16, 2024

We report the use of a multimodal generative artificial intelligence framework, X-LoRA-Gemma large language model (LLM), to analyze, design and test molecular design. The model, inspired by biological principles featuring ~7 billion parameters, dynamically reconfigures its structure through dual-pass inference strategy enhance problem-solving abilities across diverse scientific domains. is used first identify engineering targets systematic human-AI AI-AI self-driving multi-agent approach elucidate key for optimization improve interactions between molecules. Next, process that includes rational steps, reasoning autonomous knowledge extraction. Target properties molecule are identified either using Principal Component Analysis (PCA) or sampling from distribution known properties. then generate set candidate molecules, which analyzed via their structure, charge distribution, other features. validate as predicted, increased dipole moment polarizability indeed achieved in designed anticipate an increasing integration these techniques into workflow, ultimately enabling development innovative solutions address wide range societal challenges. conclude with critical discussion challenges opportunities AI engineering, analysis

Язык: Английский

---

Recent Advances in the Integration of Protein Mechanics and Machine Learning

Extreme Mechanics Letters, Год журнала: 2024, Номер unknown, С. 102236 - 102236

Опубликована: Сен. 1, 2024

Язык: Английский

---

A language model assistant for biocatalysis

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Ноя. 18, 2024

Abstract Language model assistants have transformed how researchers interact with computational tools, offering unprecedented capabilities in understanding and generating complex scientific queries. We introduce a language assistant for biocatalysis (LM-ABC), tool designed to streamline workflows enzyme engineering research. LM-ABC integrates large domain-specific modules facilitate research through natural inputs. Its architecture employs the Reasoning Acting (ReACT) framework dynamic selection chaining, enabling functionalities like binding site extraction molecular dynamics simulations. can interpret process user queries form of language, interface existing resources generate relevant results engineering. Additionally, is available via both command-line web-based interfaces, which lowers barriers its usage integration various disciplines. Provided as open-source software, contributes application models biology, potentially accelerating processes.

Язык: Английский

---

Modeling protein motions through reinforcement learning

Proceedings of the National Academy of Sciences, Год журнала: 2024, Номер 121(51)

Опубликована: Дек. 9, 2024

Язык: Английский

---

The Application of Machine Learning on Antibody Discovery and Optimization

Molecules, Год журнала: 2024, Номер 29(24), С. 5923 - 5923

Опубликована: Дек. 16, 2024

Antibodies play critical roles in modern medicine, serving as diagnostics and therapeutics for various diseases due to their ability specifically bind target antigens. Traditional antibody discovery optimization methods are time-consuming resource-intensive, though they have successfully generated antibodies diagnosing treating diseases. The advancements protein data, computational hardware, machine learning (ML) models the opportunity disrupt research. Machine demonstrated abilities design. These enable rapid silico design of candidates within a few days, achieving approximately 60% reduction time 50% cost compared traditional methods. This review focuses on latest learning-based developments. We briefly discuss limitations then explore methodologies. also focus future research directions, including developing Antibody Design AI Agents data foundries, alongside ethical regulatory considerations essential adopting learning-driven designs.

Язык: Английский

---

ProtChat: An AI Multi-Agent for Automated Protein Analysis Leveraging GPT-4 and Protein Language Model

Journal of Chemical Information and Modeling, Год журнала: 2024, Номер 65(1), С. 62 - 70

Опубликована: Дек. 17, 2024

Large language models (LLMs) have transformed natural processing, enabling advanced human-machine communication. Similarly, in computational biology, protein sequences are interpreted as language, facilitating the creation of large (PLLMs). However, applying PLLMs requires specialized preprocessing and script development, increasing complexity their use. Researchers integrated LLMs with to develop automated analysis tools address these challenges, simplifying analytical workflows. Existing technologies often require substantial human intervention for specific protein-related tasks, maintaining high barriers implementing systems. Here, we propose ProtChat, an AI multiagent system that integrates inference capabilities task-planning abilities LLMs. ProtChat GPT-4 multiple PLLMs, like ESM MASSA, automate tasks such property prediction protein–drug interactions without intervention. This agent enables users input instructions directly, significantly improving efficiency usability, making it suitable researchers a background. Experiments demonstrate can complex accurately, avoiding manual delivering results rapidly. advancement opens new research avenues biology drug discovery. Future applications may extend ProtChat's broader biological data analysis. Our code publicly available at github.com/SIAT-code/ProtChat.

Язык: Английский

---

Molecular analysis and design using generative artificial intelligence via multi-agent modeling

Molecular Systems Design & Engineering, Год журнала: 2024, Номер unknown

Опубликована: Янв. 1, 2024

We report the use of a multiagent generative artificial intelligence framework, X-LoRA-Gemma large language model (LLM), to analyze, design and test molecular design. The model, inspired by biological...

Язык: Английский

---