A5000935700- Advanced Materials Characterization Techniques
- Protein Structure and Dynamics
- Electron and X-Ray Spectroscopy Techniques
- Bioinformatics and Genomic Networks
- Biomedical Text Mining and Ontologies
- Topic Modeling
- Natural Language Processing Techniques
- Semantic Web and Ontologies
- Computational Drug Discovery Methods
- AI-based Problem Solving and Planning
- Tuberculosis Research and Epidemiology
- Genomics and Phylogenetic Studies
- HIV/AIDS drug development and treatment
- Fire Detection and Safety Systems
- Pneumocystis jirovecii pneumonia detection and treatment
- Anomaly Detection Techniques and Applications
- Chemical Synthesis and Analysis
- Evolutionary Algorithms and Applications
- Machine Learning in Materials Science
- Innovative Microfluidic and Catalytic Techniques Innovation
- Reinforcement Learning in Robotics
- Scientific Computing and Data Management
- Neural Networks and Reservoir Computing
The Ohio State University
2024
University of Cincinnati
2016-2021
Georgia Institute of Technology
2021
Georgia Tech Research Institute
2021
Cincinnati Children's Hospital Medical Center
2016-2018
Abstract Background Proteins generally perform their function in a folded state. Residues forming an active site, whether it is catalytic center or interaction interface, are frequently distant protein sequence. Hence, traditional sequence-based prediction methods focusing on single residue (or short window of residues) at time may have difficulties identifying and clustering the residues constituting functional especially when has multiple functions. Evolutionary information encoded...
Chemistry plays a crucial role in many domains, such as drug discovery and material science. While large language models (LLMs) GPT-4 exhibit remarkable capabilities on natural processing tasks, existing work shows their performance chemistry tasks is discouragingly low. In this paper, however, we demonstrate that our developed LLMs can achieve very strong results comprehensive set of outperforming the most advanced across all by substantial margin approaching SoTA task-specific models. The...
Drug discovery is a long, expensive, and complex process, relying heavily on human medicinal chemists, who can spend years searching the vast space of potential therapies. Recent advances in artificial intelligence for chemistry have sought to expedite individual drug tasks; however, there remains critical need an intelligent agent that navigate process. Towards this end, we introduce LIDDiA, autonomous capable intelligently navigating process silico. By leveraging reasoning capabilities...
Similarity and distance matrices are general data structures that describe reciprocal relationships between the objects within a given dataset. Commonly used methods for representation of these include heatmaps, hierarchical trees, dimensionality reduction, various types networks. However, despite well-developed foundation visualization such representations, challenge creating an interactive view would allow quick navigation interpretation remains largely unaddressed. This problem becomes...
Proteins by and large carry out their molecular functions in a folded state when residues, distant sequence, assemble together 3D space to bind ligand, catalyze reaction, form channel, or exert another concerted macromolecular interaction. It has been long recognized that covariance of amino acids between positions within protein sequence allows for the inference range contacts facilitate structure modeling. In this work, we investigated whether analysis may reveal residues involved same...
Pneumocystis pneumonia (PCP) is an opportunistic infection that occurs in humans and other mammals with debilitated immune systems. These infections are caused by fungi the genus Pneumocystis, which not susceptible to standard antifungal agents. Despite decades of research drug development, primary treatment prophylaxis for PCP remains a combination trimethoprim (TMP) sulfamethoxazole (SMX) targets two enzymes folic acid biosynthesis, dihydrofolate reductase (DHFR) dihydropteroate synthase...
Retrosynthesis is the process of determining set reactant molecules that can react to form a desired product. Semitemplate-based retrosynthesis methods, which imitate reverse logic synthesis reactions, first predict reaction centers in products and then complete resulting synthons back into reactants. We develop new offline–online reinforcement learning method RLSynC for synthon completion semitemplate-based methods. assigns one agent each synthon, all by conducting actions step synchronized...
The advancements of language models (LLMs) have piqued growing interest in developing LLM-based agents to automate scientific discovery end-to-end, which has sparked both excitement and skepticism about the true capabilities such agents. In this work, we argue that for an agent fully discovery, it must be able complete all essential tasks workflow. Thus, call rigorous assessment on individual a workflow before making bold claims end-to-end automation. To end, present ScienceAgentBench, new...
To enhance large language models (LLMs) for chemistry problem solving, several LLM-based agents augmented with tools have been proposed, such as ChemCrow and Coscientist. However, their evaluations are narrow in scope, leaving a gap understanding the benefits of across diverse tasks. bridge this gap, we develop ChemAgent, an enhanced agent over ChemCrow, conduct comprehensive evaluation its performance on both specialized tasks general questions. Surprisingly, ChemAgent does not consistently...
Retrosynthesis is the process of determining set reactant molecules that can react to form a desired product. Semi-template-based retrosynthesis methods, which imitate reverse logic synthesis reactions, first predict reaction centers in products, and then complete resulting synthons back into reactants. We develop new offline-online reinforcement learning method RLSynC for synthon completion semi-template-based methods. assigns one agent each synthon, all by conducting actions step...