Search for: All records

ABSTRACT Protein structural fluctuations, measured by Debye‐Waller factors or B‐factors, are known to be closely associated with protein flexibility and function. Theoretical approaches have also been developed to predict B‐factor values, which reflect protein flexibility. Previous models have made significant strides in analyzing B‐factors by fitting experimental data. In this study, we propose a novel approach for B‐factor prediction using differential geometry theory, based on the assumption that the intrinsic properties of proteins reside on a family of low‐dimensional manifolds embedded within the high‐dimensional space of protein structures. By analyzing the mean and Gaussian curvatures of a set of low‐dimensional manifolds defined by kernel functions, we develop effective and robust multiscale differential geometry (mDG) models. Our mDG model demonstrates a 27% increase in accuracy compared to the classical Gaussian network model (GNM) in predicting B‐factors for a dataset of 364 proteins. Additionally, by incorporating both global and local protein features, we construct a highly effective machine‐learning model for the blind prediction of B‐factors. Extensive least‐squares approximations and machine learning‐based blind predictions validate the effectiveness of the mDG modeling approach for B‐factor predictions. 

While large language models (LLMs) have demonstrated impressive capabilities across tasks in language understanding and interactive decision making, their abilities for reasoning (e.g. chain-of-thought prompting) and acting (e.g. action plan generation) have primarily been studied as separate topics. In this paper, we explore the use of LLMs to generate both reasoning traces and task-specific actions in an interleaved manner, allowing for greater synergy between the two: reasoning traces help the model induce, track, and update action plans as well as handle exceptions, while actions allow it to interface with external sources, such as knowledge bases or environments, to gather additional information. We apply our approach, named ReAct, to a diverse set of language and decision making tasks and demonstrate its effectiveness over state-of-the-art baselines, as well as improved human interpretability and trustworthiness over methods without reasoning or acting components. Concretely, on question answering (HotpotQA) and fact verification (Fever), ReAct overcomes issues of hallucination and error propagation prevalent in chain-of-thought reasoning by interacting with a simple Wikipedia API, and generates human-like task-solving trajectories that are more interpretable than baselines without reasoning traces. On two interactive decision making benchmarks (ALFWorld and WebShop), ReAct outperforms imitation and reinforcement learning methods by an absolute success rate of 34% and 10% respectively, while being prompted with only one or two in-context examples.