Gihan Panapitiya
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Gihan Panapitiya has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Emily Saldanha
Pacific Northwest National Laboratory
8 shared publications - Peiyuan Gao
Pacific Northwest National Laboratory
2 shared publications Aaron Hollas
Pacific Northwest National Laboratory, Richland, Washington 99352, United States
1 shared publication- Andrew David McNaughton
Pacific Northwest National Laboratory, Richland, Washington 99354, United States
1 shared publication C. Mark Maupin
Pacific Northwest National Laboratory
1 shared publication- Carter Knuston
Pacific Northwest National Laboratory
1 shared publication
Latest Publications
- Open Access
- 25 June 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
AutoLabs: cognitive multi-agent systems with self-correction for autonomous chemical experimentation
The automation of chemical research through self-driving laboratories (SDLs) promises to accelerate scientific discovery, yet the reliability and granular performance of the underlying AI agents remain critical, under-examined challenges. In this work, we introduce AutoLabs, a self-correcting, multi-agent architecture designed to autonomously translate natural-language instructions into executable protocols for a high-throughput liquid handler. The system engages users in dialogue, decomposes experimental goals into discrete tasks for specialized agents, performs tool-assisted stoichiometric calculations, and iteratively self-corrects its output before generating a hardware-ready file. We present a comprehensive evaluation framework featuring five benchmark experiments of increasing complexity, from simple sample preparation to multi-plate timed syntheses. Through a systematic ablation study of 20 agent configurations, we assess the impact of reasoning capacity, architectural design (single- vs. multi-agent), tool use, and self-correction mechanisms. Our results demonstrate that agent reasoning capacity is the most critical factor for success, reducing quantitative errors in chemical amounts (nRMSE) by over 85% in complex tasks. When combined with a multi-agent architecture and iterative self-correction, AutoLabs approaches expert-authored reference procedures on the benchmark (F1-score > 0.89) on challenging multi-plate syntheses. These findings establish a clear blueprint for developing robust and trustworthy AI partners for autonomous laboratories, highlighting the synergistic effects of modular design, advanced reasoning, and self-correction to ensure both performance and reliability in high-stakes scientific applications. Code: https://github.com/pnnl/autolabs
- 25 June 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessAutoLabs: cognitive multi-agent systems with self-correction for autonomous chemical experimentation
The automation of chemical research through self-driving laboratories (SDLs) promises to accelerate scientific discovery, yet the reliability and granular performance of the underlying AI agents remain critical, under-examined challenges. In this work, we introduce AutoLabs, a self-correcting, multi-agent architecture designed to autonomously translate natural-language instructions into executable protocols for a high-throughput liquid handler. The system engages users in dialogue, decomposes experimental goals into discrete tasks for specialized agents, performs tool-assisted stoichiometric calculations, and iteratively self-corrects its output before generating a hardware-ready file. We present a comprehensive evaluation framework featuring five benchmark experiments of increasing complexity, from simple sample preparation to multi-plate timed syntheses. Through a systematic ablation study of 20 agent configurations, we assess the impact of reasoning capacity, architectural design (single- vs. multi-agent), tool use, and self-correction mechanisms. Our results demonstrate that agent reasoning capacity is the most critical factor for success, reducing quantitative errors in chemical amounts (nRMSE) by over 85% in complex tasks. When combined with a multi-agent architecture and iterative self-correction, AutoLabs approaches expert-authored reference procedures on the benchmark (F1-score > 0.89) on challenging multi-plate syntheses. These findings establish a clear blueprint for developing robust and trustworthy AI partners for autonomous laboratories, highlighting the synergistic effects of modular design, advanced reasoning, and self-correction to ensure both performance and reliability in high-stakes scientific applications. Code: https://github.com/pnnl/autolabs
- Top Cited Papers
- 25 February 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
- Vol. 148 (9) , 9930-9950
FragNet: A Graph Neural Network for Molecular Property Prediction with Four Levels of Interpretability
Molecular property prediction is essential in a variety of contemporary scientific fields, such as drug development and the design of energy storage materials. Although there are many machine learning models available for this purpose, those that achieve high accuracy while also offering the interpretability of predictions are uncommon. We present a graph neural network that not only matches the prediction accuracies of leading models but also provides insights into four levels of molecular substructures. This model helps identify which atoms, bonds, molecular fragments, and connections among fragments are significant for predicting a specific molecular property. Understanding the importance of connections between fragments is particularly valuable for molecules with substructures that do not connect through standard bonds. The model additionally can quantify the impact of specific fragments on the prediction, allowing for the identification of fragments that may improve or degrade a property value. These interpretable features are essential for deriving scientific insights from the model's learned relationships between molecular structures and properties.
- 25 February 2026
- Published by American Chemical Society (ACS) in Journal of the American Chemical Society
- Vol. 148 (9) , 9930-9950
Top Cited PapersFragNet: A Graph Neural Network for Molecular Property Prediction with Four Levels of Interpretability
Molecular property prediction is essential in a variety of contemporary scientific fields, such as drug development and the design of energy storage materials. Although there are many machine learning models available for this purpose, those that achieve high accuracy while also offering the interpretability of predictions are uncommon. We present a graph neural network that not only matches the prediction accuracies of leading models but also provides insights into four levels of molecular substructures. This model helps identify which atoms, bonds, molecular fragments, and connections among fragments are significant for predicting a specific molecular property. Understanding the importance of connections between fragments is particularly valuable for molecules with substructures that do not connect through standard bonds. The model additionally can quantify the impact of specific fragments on the prediction, allowing for the identification of fragments that may improve or degrade a property value. These interpretable features are essential for deriving scientific insights from the model's learned relationships between molecular structures and properties.
- Open Access
- 21 January 2026
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 16 (6) , 5281-5295
Out-of-distribution evaluation of active learning pipelines for molecular property prediction
Evidential deep learning–driven active learning beats random sampling for OOD solvation free-energy prediction, delivering lower test RMSE at the same labeling budget and improving generalization to unseen chemical space.
- 21 January 2026
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 16 (6) , 5281-5295
Open AccessOut-of-distribution evaluation of active learning pipelines for molecular property prediction
Evidential deep learning–driven active learning beats random sampling for OOD solvation free-energy prediction, delivering lower test RMSE at the same labeling budget and improving generalization to unseen chemical space.
- Open Access
AutoLabs: Cognitive Multi-Agent Systems with Self-Correction for Autonomous Chemical Experimentation
The automation of chemical research through self-driving laboratories (SDLs) promises to accelerate scientific discovery, yet the reliability and granular performance of the underlying AI agents remain critical, under-examined challenges. In this work, we introduce AutoLabs, a self-correcting, multi-agent architecture designed to autonomously translate natural-language instructions into executable protocols for a high-throughput liquid handler. The system engages users in dialogue, decomposes experimental goals into discrete tasks for specialized agents, performs tool-assisted stoichiometric calculations, and iteratively self-corrects its output before generating a hardware-ready file. We present a comprehensive evaluation framework featuring five benchmark experiments of increasing complexity, from simple sample preparation to multi-plate timed syntheses. Through a systematic ablation study of 20 agent configurations, we assess the impact of reasoning capacity, architectural design (single- vs. multi-agent), tool use, and self-correction mechanisms. Our results demonstrate that agent reasoning capacity is the most critical factor for success, reducing quantitative errors in chemical amounts (nRMSE) by over 85% in complex tasks. When combined with a multi-agent architecture and iterative self-correction, AutoLabs achieves near-expert procedural accuracy (F1-score > 0.89) on challenging multi-step syntheses. These findings establish a clear blueprint for developing robust and trustworthy AI partners for autonomous laboratories, highlighting the synergistic effects of modular design, advanced reasoning, and self-correction to ensure both performance and reliability in high-stakes scientific applications. Code: https://github.com/pnnl/autolabs
Open AccessAutoLabs: Cognitive Multi-Agent Systems with Self-Correction for Autonomous Chemical Experimentation
The automation of chemical research through self-driving laboratories (SDLs) promises to accelerate scientific discovery, yet the reliability and granular performance of the underlying AI agents remain critical, under-examined challenges. In this work, we introduce AutoLabs, a self-correcting, multi-agent architecture designed to autonomously translate natural-language instructions into executable protocols for a high-throughput liquid handler. The system engages users in dialogue, decomposes experimental goals into discrete tasks for specialized agents, performs tool-assisted stoichiometric calculations, and iteratively self-corrects its output before generating a hardware-ready file. We present a comprehensive evaluation framework featuring five benchmark experiments of increasing complexity, from simple sample preparation to multi-plate timed syntheses. Through a systematic ablation study of 20 agent configurations, we assess the impact of reasoning capacity, architectural design (single- vs. multi-agent), tool use, and self-correction mechanisms. Our results demonstrate that agent reasoning capacity is the most critical factor for success, reducing quantitative errors in chemical amounts (nRMSE) by over 85% in complex tasks. When combined with a multi-agent architecture and iterative self-correction, AutoLabs achieves near-expert procedural accuracy (F1-score > 0.89) on challenging multi-step syntheses. These findings establish a clear blueprint for developing robust and trustworthy AI partners for autonomous laboratories, highlighting the synergistic effects of modular design, advanced reasoning, and self-correction to ensure both performance and reliability in high-stakes scientific applications. Code: https://github.com/pnnl/autolabs
- 13 May 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (10) , 4906-4917
Extracting Material Property Measurements from Scientific Literature with Limited Annotations
Extracting material property data from scientific text is pivotal for advancing data-driven research in chemistry and materials science; however, the extensive annotation effort required to produce training data for named entity recognition (NER) models for this task often makes it a barrier to extracting specialized data sets. In this work, we present a comparative study of the conventional, supervised NER methodology to alternative few-shot learning architectures and large language model (LLM)-based approaches that mitigate the need to label large training data sets. We find that the best-performing LLM (GPT-4o) not only excels in directly extracting relevant material properties based on limited examples but also enhances supervised learning through data augmentation. We supplement our findings with error and data quality assessments to provide a nuanced understanding of factors that impact property measurement extraction.
- 13 May 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (10) , 4906-4917
Extracting Material Property Measurements from Scientific Literature with Limited Annotations
Extracting material property data from scientific text is pivotal for advancing data-driven research in chemistry and materials science; however, the extensive annotation effort required to produce training data for named entity recognition (NER) models for this task often makes it a barrier to extracting specialized data sets. In this work, we present a comparative study of the conventional, supervised NER methodology to alternative few-shot learning architectures and large language model (LLM)-based approaches that mitigate the need to label large training data sets. We find that the best-performing LLM (GPT-4o) not only excels in directly extracting relevant material properties based on limited examples but also enhances supervised learning through data augmentation. We supplement our findings with error and data quality assessments to provide a nuanced understanding of factors that impact property measurement extraction.
- Open Access
- 8 November 2024
- Published by Springer Nature in Research Square
FragNet: A Graph Neural Network for Molecular Property Prediction with Four Layers of Interpretability
Molecular property prediction is a crucial step in many modern-day scientific applications including drug discovery and energy storage material design. Despite the availability of numerous machine learning models for this task, we are lacking in models that provide both high accuracies and interpretability of the predictions. We introduce the FragNet architecture, a graph neural network not only capable of achieving prediction accuracies comparable to the current state-of-the-art models, but also able to provide insight on four levels of molecular substructures. This model enables understanding of which atoms, bonds, molecular fragments, and molecular fragment connections are critical in the prediction of a given molecular property. The ability to interpret the importance of connections between fragments is of particular interest for molecules which have substructures that are not connected with regular covalent bonds. The interpretable capabilities of FragNet are key to gaining scientific insights from the model’s learned patterns between molecular structure and molecular properties.
- 8 November 2024
- Published by Springer Nature in Research Square
Open AccessFragNet: A Graph Neural Network for Molecular Property Prediction with Four Layers of Interpretability
Molecular property prediction is a crucial step in many modern-day scientific applications including drug discovery and energy storage material design. Despite the availability of numerous machine learning models for this task, we are lacking in models that provide both high accuracies and interpretability of the predictions. We introduce the FragNet architecture, a graph neural network not only capable of achieving prediction accuracies comparable to the current state-of-the-art models, but also able to provide insight on four levels of molecular substructures. This model enables understanding of which atoms, bonds, molecular fragments, and molecular fragment connections are critical in the prediction of a given molecular property. The ability to interpret the importance of connections between fragments is of particular interest for molecules which have substructures that are not connected with regular covalent bonds. The interpretable capabilities of FragNet are key to gaining scientific insights from the model’s learned patterns between molecular structure and molecular properties.
- Open Access
- 2 September 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Impact of Molecular Representations on Deep Learning Model Comparisons in Drug Response Predictions
Deep learning (DL) plays a crucial role in tackling the complexity and heterogeneity of cancer, particularly in predicting drug response. However, the effectiveness of these models is often hindered by inconsistent benchmarks and disparate data sources. To address the gaps in comparisons, we introduce CoMParison workflow for Cross Validation (CMP-CV), an automated cross-validation framework that trains multiple models with user-specified parameters and evaluation metrics. The effectiveness of DL models in predicting drug responses is closely tied to the methods used to represent drugs at the molecular level. In this contribution, we benchmarked commonly leveraged drug representations (graph, molecular descriptors, molecular fingerprints, and SMILES) to learn and understand the predictive capabilities of the models. We compare the ability of different drug representations to encode different structural properties of the drugs by using prediction errors made by models in different drug descriptor domains. We find that, in terms of the average prediction error over the entire test set, molecular descriptors and Morgan fingerprints perform slightly better than the others. However, we also observe that the rankings of the model performance vary in different regions over the descriptor space studied in this work, emphasizing the importance of domain-based model comparison when selecting a model for a specific application. Our efforts are part of CANcer Distributed Learning Environment (CANDLE), enhancing the model comparison capabilities in cancer research and driving the development of more effective strategies for drug response prediction and optimization.
- 2 September 2024
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessImpact of Molecular Representations on Deep Learning Model Comparisons in Drug Response Predictions
Deep learning (DL) plays a crucial role in tackling the complexity and heterogeneity of cancer, particularly in predicting drug response. However, the effectiveness of these models is often hindered by inconsistent benchmarks and disparate data sources. To address the gaps in comparisons, we introduce CoMParison workflow for Cross Validation (CMP-CV), an automated cross-validation framework that trains multiple models with user-specified parameters and evaluation metrics. The effectiveness of DL models in predicting drug responses is closely tied to the methods used to represent drugs at the molecular level. In this contribution, we benchmarked commonly leveraged drug representations (graph, molecular descriptors, molecular fingerprints, and SMILES) to learn and understand the predictive capabilities of the models. We compare the ability of different drug representations to encode different structural properties of the drugs by using prediction errors made by models in different drug descriptor domains. We find that, in terms of the average prediction error over the entire test set, molecular descriptors and Morgan fingerprints perform slightly better than the others. However, we also observe that the rankings of the model performance vary in different regions over the descriptor space studied in this work, emphasizing the importance of domain-based model comparison when selecting a model for a specific application. Our efforts are part of CANcer Distributed Learning Environment (CANDLE), enhancing the model comparison capabilities in cancer research and driving the development of more effective strategies for drug response prediction and optimization.
- Open Access
- 2 February 2023
- Published by American Chemical Society (ACS) in ChemRxiv
Outlier-Based Domain of Applicability Identification for Materials Property Prediction Models
Machine learning models have been widely applied for material property prediction. However, practical application of these models can be hindered by a lack of information about how well they will perform on previously unseen types of materials. Because machine learning model predictions depend on the quality of the available training data, different domains of the material feature space are predicted with different accuracy levels by such models. The ability to identify such domains enables the ability to find the confidence level of each prediction, to determine when and how the model should be employed depending on the prediction accuracy requirements of different tasks, and to improve the model for domains with high errors. In this work, we propose a method to find domains of applicability using a large feature space and also introduce analysis techniques to gain more insight into the detected domains and subdomains.
- 2 February 2023
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessOutlier-Based Domain of Applicability Identification for Materials Property Prediction Models
Machine learning models have been widely applied for material property prediction. However, practical application of these models can be hindered by a lack of information about how well they will perform on previously unseen types of materials. Because machine learning model predictions depend on the quality of the available training data, different domains of the material feature space are predicted with different accuracy levels by such models. The ability to identify such domains enables the ability to find the confidence level of each prediction, to determine when and how the model should be employed depending on the prediction accuracy requirements of different tasks, and to improve the model for domains with high errors. In this work, we propose a method to find domains of applicability using a large feature space and also introduce analysis techniques to gain more insight into the detected domains and subdomains.
- Open Access
Outlier-Based Domain of Applicability Identification for Materials Property Prediction Models
Machine learning models have been widely applied for material property prediction. However, practical application of these models can be hindered by a lack of information about how well they will perform on previously unseen types of materials. Because machine learning model predictions depend on the quality of the available training data, different domains of the material feature space are predicted with different accuracy levels by such models. The ability to identify such domains enables the ability to find the confidence level of each prediction, to determine when and how the model should be employed depending on the prediction accuracy requirements of different tasks, and to improve the model for domains with high errors. In this work, we propose a method to find domains of applicability using a large feature space and also introduce analysis techniques to gain more insight into the detected domains and subdomains.
Open AccessOutlier-Based Domain of Applicability Identification for Materials Property Prediction Models
Machine learning models have been widely applied for material property prediction. However, practical application of these models can be hindered by a lack of information about how well they will perform on previously unseen types of materials. Because machine learning model predictions depend on the quality of the available training data, different domains of the material feature space are predicted with different accuracy levels by such models. The ability to identify such domains enables the ability to find the confidence level of each prediction, to determine when and how the model should be employed depending on the prediction accuracy requirements of different tasks, and to improve the model for domains with high errors. In this work, we propose a method to find domains of applicability using a large feature space and also introduce analysis techniques to gain more insight into the detected domains and subdomains.
- Open Access
- 25 April 2022
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 7 (18) , 15695-15710
Evaluation of Deep Learning Architectures for Aqueous Solubility Prediction
Determining the aqueous solubility of molecules is a vital step in many pharmaceutical, environmental, and energy storage applications. Despite efforts made over decades, there are still challenges associated with developing a solubility prediction model with satisfactory accuracy for many of these applications. The goals of this study are to assess current deep learning methods for solubility prediction, develop a general model capable of predicting the solubility of a broad range of organic molecules, and to understand the impact of data properties, molecular representation, and modeling architecture on predictive performance. Using the largest currently available solubility data set, we implement deep learning-based models to predict solubility from the molecular structure and explore several different molecular representations including molecular descriptors, simplified molecular-input line-entry system strings, molecular graphs, and three-dimensional atomic coordinates using four different neural network architectures-fully connected neural networks, recurrent neural networks, graph neural networks (GNNs), and SchNet. We find that models using molecular descriptors achieve the best performance, with GNN models also achieving good performance. We perform extensive error analysis to understand the molecular properties that influence model performance, perform feature analysis to understand which information about the molecular structure is most valuable for prediction, and perform a transfer learning and data size study to understand the impact of data availability on model performance.
- 25 April 2022
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 7 (18) , 15695-15710
Open AccessEvaluation of Deep Learning Architectures for Aqueous Solubility Prediction
Determining the aqueous solubility of molecules is a vital step in many pharmaceutical, environmental, and energy storage applications. Despite efforts made over decades, there are still challenges associated with developing a solubility prediction model with satisfactory accuracy for many of these applications. The goals of this study are to assess current deep learning methods for solubility prediction, develop a general model capable of predicting the solubility of a broad range of organic molecules, and to understand the impact of data properties, molecular representation, and modeling architecture on predictive performance. Using the largest currently available solubility data set, we implement deep learning-based models to predict solubility from the molecular structure and explore several different molecular representations including molecular descriptors, simplified molecular-input line-entry system strings, molecular graphs, and three-dimensional atomic coordinates using four different neural network architectures-fully connected neural networks, recurrent neural networks, graph neural networks (GNNs), and SchNet. We find that models using molecular descriptors achieve the best performance, with GNN models also achieving good performance. We perform extensive error analysis to understand the molecular properties that influence model performance, perform feature analysis to understand which information about the molecular structure is most valuable for prediction, and perform a transfer learning and data size study to understand the impact of data availability on model performance.