Joseph A. Morrone
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Joseph A. Morrone has a total of 109 co-authors. Here are the top co-authors by number of shared publications:
Tien Huynh
Computational Biology, IBM Thomas J. Watson Research Center, 1101 Kitchawan Road, Yorktown Heights, New York 10594, United States
6 shared publications- Roberto Car
Department of Chemistry, Princeton University, Princeton, USA
5 shared publications - Seung-Gu Kang
Cornell Computational Biology Center IBM Thomas J. Watson Research 1101 Kitchawan Road 10594 Yorktown Heights NY USA
5 shared publications - Alberto Perez
Department of Chemistry and Quantum Theory Project, University of Florida, Gainesville, FL, USA
4 shared publications B. J. Berne
Department of Chemistry, Columbia University, 3000 Broadway, MC 3103, New York, New York 10027, United States
4 shared publications- Bruce J. Berne
Department of Chemistry, Columbia University, New York, New York 10027, USA
4 shared publications
Latest Publications
- Open Access
- 28 October 2026
- Published by Royal Society of Chemistry (RSC) in Chemical Science
- Vol. 17 (41) , 20118-20139
Syndirella: synthesis-directed fragment elaboration enables extensive and cost-effective binding site exploration beyond catalog compounds
Syndirella proposes elaborations of fragment-derived scaffolds for high-throughput synthesis using in-catalog reactants for SAR exploration.
- 28 October 2026
- Published by Royal Society of Chemistry (RSC) in Chemical Science
- Vol. 17 (41) , 20118-20139
Open AccessSyndirella: synthesis-directed fragment elaboration enables extensive and cost-effective binding site exploration beyond catalog compounds
Syndirella proposes elaborations of fragment-derived scaffolds for high-throughput synthesis using in-catalog reactants for SAR exploration.
- 4 March 2026
- Published by American Chemical Society (ACS) in The Journal of Physical Chemistry B
- Vol. 130 (11) , 2965-2973
In-Pocket 3D Graphs Enhance Ligand–Target Compatibility in Generative Small-Molecule Creation: A Dopamine D2 Receptor Model System
Proteins in complex with small-molecule ligands represent the core of structure-based drug discovery. However, three-dimensional representations are absent from most deep-learning-based generative models. Here, we present a graph-based generative modeling technology that encodes explicit 3D protein-ligand contacts within a relational graph architecture and evaluate its behavior using the dopamine D2 receptor (DD2R) as a model system. The models combine a conditional variational autoencoder that allows for activity-specific molecule generation with putative contact generation that provides predictions of molecular interactions within the target-binding pocket. We show that molecules generated with our 3D procedure are more compatible with the DD2R-binding pocket than those produced by a comparable ligand-based 2D generative method, as measured by docking scores, expected stereochemistry, and recoverability in commercial chemical databases. Predicted protein-ligand contacts were found to be among the highest-ranked docking poses with a high recovery rate. Overall, this work shows how the structural context of a protein target can enhance the generation of small molecules within a realistic binding environment.
- 4 March 2026
- Published by American Chemical Society (ACS) in The Journal of Physical Chemistry B
- Vol. 130 (11) , 2965-2973
In-Pocket 3D Graphs Enhance Ligand–Target Compatibility in Generative Small-Molecule Creation: A Dopamine D2 Receptor Model System
Proteins in complex with small-molecule ligands represent the core of structure-based drug discovery. However, three-dimensional representations are absent from most deep-learning-based generative models. Here, we present a graph-based generative modeling technology that encodes explicit 3D protein-ligand contacts within a relational graph architecture and evaluate its behavior using the dopamine D2 receptor (DD2R) as a model system. The models combine a conditional variational autoencoder that allows for activity-specific molecule generation with putative contact generation that provides predictions of molecular interactions within the target-binding pocket. We show that molecules generated with our 3D procedure are more compatible with the DD2R-binding pocket than those produced by a comparable ligand-based 2D generative method, as measured by docking scores, expected stereochemistry, and recoverability in commercial chemical databases. Predicted protein-ligand contacts were found to be among the highest-ranked docking poses with a high recovery rate. Overall, this work shows how the structural context of a protein target can enhance the generation of small molecules within a realistic binding environment.
- Open Access
- 28 January 2026
- Published by Wiley in Advanced Science
- Vol. 13 (14) , e17840
Multi‐View Biomedical Foundation Models for Molecule‐Target and Property Prediction
Molecular foundation models hold promise to provide accurate predictions for a large and diverse set of downstream tasks in bio‐medical research. Quality molecular representations are key and foundation model development has typically focused on a single representation or molecular view, which may have strengths or weaknesses on a given task. We develop Multi‐view Molecular Embedding with Late Fusion (MMELON), an approach that integrates pre‐trained graph, image and text foundation models and may be readily extended to additional views and models. The multi‐view model performs robustly and is validated on over 120 tasks, including molecular solubility, ADME properties, and activity against G Protein‐Coupled receptors (GPCRs). The GPCR model array is leveraged to perform a virtual screen in search of ligands binding to Alzheimer's disease related GPCRs. We identify a number of such targets and employ the multi‐view model to select strong binders from a compound screen. Predictions are validated through structure‐based modeling and identification of key binding motifs.
- 28 January 2026
- Published by Wiley in Advanced Science
- Vol. 13 (14) , e17840
Open AccessMulti‐View Biomedical Foundation Models for Molecule‐Target and Property Prediction
Molecular foundation models hold promise to provide accurate predictions for a large and diverse set of downstream tasks in bio‐medical research. Quality molecular representations are key and foundation model development has typically focused on a single representation or molecular view, which may have strengths or weaknesses on a given task. We develop Multi‐view Molecular Embedding with Late Fusion (MMELON), an approach that integrates pre‐trained graph, image and text foundation models and may be readily extended to additional views and models. The multi‐view model performs robustly and is validated on over 120 tasks, including molecular solubility, ADME properties, and activity against G Protein‐Coupled receptors (GPCRs). The GPCR model array is leveraged to perform a virtual screen in search of ligands binding to Alzheimer's disease related GPCRs. We identify a number of such targets and employ the multi‐view model to select strong binders from a compound screen. Predictions are validated through structure‐based modeling and identification of key binding motifs.
- Open Access
Predicting pyrazinamide resistance in Mycobacterium tuberculosis using a graph convolutional network
Background: Pyrazinamide is an important first-line antibiotic for treating tuberculosis, with resistance primarily driven by mutations in the pncA gene. Traditional machine learning models are able to predict pyrazinamide resistance with some success, but are limited in their ability to incorporate 3-dimensional protein structural information. Graph neural networks offer the potential to integrate protein structure and residue-level features to better predict the impact of mutations on drug resistance. Results: We trained a graph convolutional network on PncA variants containing missense mutations and evaluated its ability to classify resistance to pyrazinamide. Each PncA variant was represented as an amino acid-level graph, with edges calculated from 3-dimensional spatial proximity, and node features derived from chemical properties and mutation meta-predictors. We used AlphaFold2 to generate predicted structures of the PncA variants, which we used to create the protein graphs. The predicted structures of resistant PncA variants showed greater deviation from the wild-type structure compared to susceptible variants. Our model achieved an F1 score of 81.6 %, sensitivity of 81.6 % and specificity of 80.4 % on the test set and either matched or exceeded the performance of a published set of traditional machine learning models. We show that both structural graph connectivity and node features contribute significantly to model performance. Furthermore, we employ additional train/test dataset splits to demonstrate the GCN’s ability to generalise and predict resistance in samples with mutations in unseen positions and structural regions. Conclusions: Our study demonstrates that graph-based deep learning can leverage protein structure and biochemical features to accurately predict antimicrobial resistance, despite being trained on a small dataset with little variation. We present this as a proof-of-concept for these methods to be applied to resistance phenotype prediction in more genetically diverse pathogens to predict the more complex observed patterns of antimicrobial resistance.
Open AccessPredicting pyrazinamide resistance in Mycobacterium tuberculosis using a graph convolutional network
Background: Pyrazinamide is an important first-line antibiotic for treating tuberculosis, with resistance primarily driven by mutations in the pncA gene. Traditional machine learning models are able to predict pyrazinamide resistance with some success, but are limited in their ability to incorporate 3-dimensional protein structural information. Graph neural networks offer the potential to integrate protein structure and residue-level features to better predict the impact of mutations on drug resistance. Results: We trained a graph convolutional network on PncA variants containing missense mutations and evaluated its ability to classify resistance to pyrazinamide. Each PncA variant was represented as an amino acid-level graph, with edges calculated from 3-dimensional spatial proximity, and node features derived from chemical properties and mutation meta-predictors. We used AlphaFold2 to generate predicted structures of the PncA variants, which we used to create the protein graphs. The predicted structures of resistant PncA variants showed greater deviation from the wild-type structure compared to susceptible variants. Our model achieved an F1 score of 81.6 %, sensitivity of 81.6 % and specificity of 80.4 % on the test set and either matched or exceeded the performance of a published set of traditional machine learning models. We show that both structural graph connectivity and node features contribute significantly to model performance. Furthermore, we employ additional train/test dataset splits to demonstrate the GCN’s ability to generalise and predict resistance in samples with mutations in unseen positions and structural regions. Conclusions: Our study demonstrates that graph-based deep learning can leverage protein structure and biochemical features to accurately predict antimicrobial resistance, despite being trained on a small dataset with little variation. We present this as a proof-of-concept for these methods to be applied to resistance phenotype prediction in more genetically diverse pathogens to predict the more complex observed patterns of antimicrobial resistance.
- Open Access
- 29 September 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Syndirella: Synthesis-directed fragment elaboration enables extensive binding site exploration beyond catalog compounds
Fragment screens provide an information-rich starting point for designing derivative compounds that recapitulate key protein-ligand interactions in structure-based drug discovery. Maximizing the compound diversity and interaction sampling of the derivatives is critical to avoid premature convergence on unproductive chemical series, yet traditional catalog procurement typically limits campaigns to <100 compounds due to budget constraints. We developed Syndirella (Synthesis Directed Elaborations), which enables the sampling of 100s–1000s of compounds by generating congeneric series through digitized multi-step synthesis routes for reactant-based purchasing, in-house robotic synthesis, and direct-to-biology testing. Applied to three viral protein targets, Syndirella demonstrated comprehensive pharmacophore diversity, recapitulating known fragment interactions while identifying novel catalytic ones and accessing distinct chemical regions compared to traditional superstructure searches. Experimental validation on a calcium-binding protein fragment screen yielded 13 X-ray crystal structures with binding kinetics from 252 synthesized compounds, achieving 84% cost reduction versus catalog purchasing. By shifting from product to reactant procurement, Syndirella enables order-of-magnitude increases in accessible compounds for structure-activity relationship exploration within typical budgets, fundamentally expanding the breadth-to-depth ratio of fragment-based series.
- 29 September 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessSyndirella: Synthesis-directed fragment elaboration enables extensive binding site exploration beyond catalog compounds
Fragment screens provide an information-rich starting point for designing derivative compounds that recapitulate key protein-ligand interactions in structure-based drug discovery. Maximizing the compound diversity and interaction sampling of the derivatives is critical to avoid premature convergence on unproductive chemical series, yet traditional catalog procurement typically limits campaigns to <100 compounds due to budget constraints. We developed Syndirella (Synthesis Directed Elaborations), which enables the sampling of 100s–1000s of compounds by generating congeneric series through digitized multi-step synthesis routes for reactant-based purchasing, in-house robotic synthesis, and direct-to-biology testing. Applied to three viral protein targets, Syndirella demonstrated comprehensive pharmacophore diversity, recapitulating known fragment interactions while identifying novel catalytic ones and accessing distinct chemical regions compared to traditional superstructure searches. Experimental validation on a calcium-binding protein fragment screen yielded 13 X-ray crystal structures with binding kinetics from 252 synthesized compounds, achieving 84% cost reduction versus catalog purchasing. By shifting from product to reactant procurement, Syndirella enables order-of-magnitude increases in accessible compounds for structure-activity relationship exploration within typical budgets, fundamentally expanding the breadth-to-depth ratio of fragment-based series.
- Open Access
- 1 August 2024
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 9, 8713-8716
SPARK: Harnessing Human-Centered Workflows with Biomedical Foundation Models for Drug Discovery
Biomedical foundation models, trained on diverse sources of small molecule data, hold great potential for accelerating drug discovery. However, their complex nature often presents a barrier for researchers seeking scientific insights and drug candidate generation. SPARK addresses this challenge by providing a user-friendly, web-based interface that empowers researchers to leverage these powerful models in their scientific workflows. Through SPARK, users can specify target proteins and desired molecule properties, adjust pre-trained models for tailored inferences, generate lists of potential drug candidates, analyze and compare molecules through interactive visualizations, and filter candidates based on key metrics (e.g., toxicity). By seamlessly integrating human knowledge and biomedical AI models' capabilities through an interactive web-based system, SPARK can improve the efficiency of collaboration between human experts and AI, thereby accelerating drug candidate discovery and ultimately leading to breakthroughs in finding cures for various diseases.
- 1 August 2024
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 9, 8713-8716
Open AccessSPARK: Harnessing Human-Centered Workflows with Biomedical Foundation Models for Drug Discovery
Biomedical foundation models, trained on diverse sources of small molecule data, hold great potential for accelerating drug discovery. However, their complex nature often presents a barrier for researchers seeking scientific insights and drug candidate generation. SPARK addresses this challenge by providing a user-friendly, web-based interface that empowers researchers to leverage these powerful models in their scientific workflows. Through SPARK, users can specify target proteins and desired molecule properties, adjust pre-trained models for tailored inferences, generate lists of potential drug candidates, analyze and compare molecules through interactive visualizations, and filter candidates based on key metrics (e.g., toxicity). By seamlessly integrating human knowledge and biomedical AI models' capabilities through an interactive web-based system, SPARK can improve the efficiency of collaboration between human experts and AI, thereby accelerating drug candidate discovery and ultimately leading to breakthroughs in finding cures for various diseases.
- Open Access
- 22 November 2023
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (1)
Unsupervised and supervised AI on molecular dynamics simulations reveals complex characteristics of HLA-A2-peptide immunogenicity
Immunologic recognition of peptide antigens bound to class I major histocompatibility complex (MHC) molecules is essential to both novel immunotherapeutic development and human health at large. Current methods for predicting antigen peptide immunogenicity rely primarily on simple sequence representations, which allow for some understanding of immunogenic features but provide inadequate consideration of the full scale of molecular mechanisms tied to peptide recognition. We here characterize contributions that unsupervised and supervised artificial intelligence (AI) methods can make toward understanding and predicting MHC(HLA-A2)-peptide complex immunogenicity when applied to large ensembles of molecular dynamics simulations. We first show that an unsupervised AI method allows us to identify subtle features that drive immunogenicity differences between a cancer neoantigen and its wild-type peptide counterpart. Next, we demonstrate that a supervised AI method for class I MHC(HLA-A2)-peptide complex classification significantly outperforms a sequence model on small datasets corrected for trivial sequence correlations. Furthermore, we show that both unsupervised and supervised approaches reveal determinants of immunogenicity based on time-dependent molecular fluctuations and anchor position dynamics outside the MHC binding groove. We discuss implications of these structural and dynamic immunogenicity correlates for the induction of T cell responses and therapeutic T cell receptor design.
- 22 November 2023
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (1)
Open AccessUnsupervised and supervised AI on molecular dynamics simulations reveals complex characteristics of HLA-A2-peptide immunogenicity
Immunologic recognition of peptide antigens bound to class I major histocompatibility complex (MHC) molecules is essential to both novel immunotherapeutic development and human health at large. Current methods for predicting antigen peptide immunogenicity rely primarily on simple sequence representations, which allow for some understanding of immunogenic features but provide inadequate consideration of the full scale of molecular mechanisms tied to peptide recognition. We here characterize contributions that unsupervised and supervised artificial intelligence (AI) methods can make toward understanding and predicting MHC(HLA-A2)-peptide complex immunogenicity when applied to large ensembles of molecular dynamics simulations. We first show that an unsupervised AI method allows us to identify subtle features that drive immunogenicity differences between a cancer neoantigen and its wild-type peptide counterpart. Next, we demonstrate that a supervised AI method for class I MHC(HLA-A2)-peptide complex classification significantly outperforms a sequence model on small datasets corrected for trivial sequence correlations. Furthermore, we show that both unsupervised and supervised approaches reveal determinants of immunogenicity based on time-dependent molecular fluctuations and anchor position dynamics outside the MHC binding groove. We discuss implications of these structural and dynamic immunogenicity correlates for the induction of T cell responses and therapeutic T cell receptor design.
- 16 September 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62 (19) , 4660-4671
Roughness of Molecular Property Landscapes and Its Impact on Modellability
In molecular discovery and drug design, structure-property relationships and activity landscapes are often qualitatively or quantitatively analyzed to guide the navigation of chemical space. The roughness (or smoothness) of these molecular property landscapes is one of their most studied geometric attributes, as it can characterize the presence of activity cliffs, with rougher landscapes generally expected to pose tougher optimization challenges. Here, we introduce a general, quantitative measure for describing the roughness of molecular property landscapes. The proposed roughness index (ROGI) is loosely inspired by the concept of fractal dimension and strongly correlates with the out-of-sample error achieved by machine learning models on numerous regression tasks.
- 16 September 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62 (19) , 4660-4671
Roughness of Molecular Property Landscapes and Its Impact on Modellability
In molecular discovery and drug design, structure-property relationships and activity landscapes are often qualitatively or quantitatively analyzed to guide the navigation of chemical space. The roughness (or smoothness) of these molecular property landscapes is one of their most studied geometric attributes, as it can characterize the presence of activity cliffs, with rougher landscapes generally expected to pose tougher optimization challenges. Here, we introduce a general, quantitative measure for describing the roughness of molecular property landscapes. The proposed roughness index (ROGI) is loosely inspired by the concept of fractal dimension and strongly correlates with the out-of-sample error achieved by machine learning models on numerous regression tasks.
- 6 August 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62 (16) , 3854-3862
Self-Focusing Virtual Screening with Active Design Space Pruning
High-throughput virtual screening is an indispensable technique utilized in the discovery of small molecules. In cases where the library of molecules is exceedingly large, the cost of an exhaustive virtual screen may be prohibitive. Model-guided optimization has been employed to lower these costs through dramatic increases in sample efficiency compared to random selection. However, these techniques introduce new costs to the workflow through the surrogate model training and inference steps. In this study, we propose an extension to the framework of model-guided optimization that mitigates inference costs using a technique we refer to as design space pruning (DSP), which irreversibly removes poor-performing candidates from consideration. We study the application of DSP to a variety of optimization tasks and observe significant reductions in overhead costs while exhibiting similar performance to the baseline optimization. DSP represents an attractive extension of model-guided optimization that can limit overhead costs in optimization settings where these costs are non-negligible relative to objective costs, such as docking.
- 6 August 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62 (16) , 3854-3862
Self-Focusing Virtual Screening with Active Design Space Pruning
High-throughput virtual screening is an indispensable technique utilized in the discovery of small molecules. In cases where the library of molecules is exceedingly large, the cost of an exhaustive virtual screen may be prohibitive. Model-guided optimization has been employed to lower these costs through dramatic increases in sample efficiency compared to random selection. However, these techniques introduce new costs to the workflow through the surrogate model training and inference steps. In this study, we propose an extension to the framework of model-guided optimization that mitigates inference costs using a technique we refer to as design space pruning (DSP), which irreversibly removes poor-performing candidates from consideration. We study the application of DSP to a variety of optimization tasks and observe significant reductions in overhead costs while exhibiting similar performance to the baseline optimization. DSP represents an attractive extension of model-guided optimization that can limit overhead costs in optimization settings where these costs are non-negligible relative to objective costs, such as docking.
- Open Access
- 15 June 2022
- Published by Springer Nature in Research Square
In-Pocket 3D Graphs Enhance Ligand-Target Compatibility in Generative Small-Molecule Creation
Proteins in complex with small molecule ligands represent the core of structure-based drug discovery. However, three-dimensional representations are absent from most deep-learning-based generative models. We here present a graph-based generative modeling technology that encodes explicit 3D protein-ligand contacts within a relational graph architecture. The models combine a conditional variational autoencoder that allows for activity-specific molecule generation with putative contact generation that provides predictions of molecular interactions within the target binding pocket. We show that molecules generated with our 3D procedure are more compatible with the binding pocket of the dopamine D2 receptor than those produced by a comparable ligand-based 2D generative method, as measured by docking scores, expected stereochemistry, and recoverability in commercial chemical databases. Predicted protein-ligand contacts were found among highest-ranked docking poses with a high recovery rate. This work shows how the structural context of a protein target can be used to enhance molecule generation.
- 15 June 2022
- Published by Springer Nature in Research Square
Open AccessIn-Pocket 3D Graphs Enhance Ligand-Target Compatibility in Generative Small-Molecule Creation
Proteins in complex with small molecule ligands represent the core of structure-based drug discovery. However, three-dimensional representations are absent from most deep-learning-based generative models. We here present a graph-based generative modeling technology that encodes explicit 3D protein-ligand contacts within a relational graph architecture. The models combine a conditional variational autoencoder that allows for activity-specific molecule generation with putative contact generation that provides predictions of molecular interactions within the target binding pocket. We show that molecules generated with our 3D procedure are more compatible with the binding pocket of the dopamine D2 receptor than those produced by a comparable ligand-based 2D generative method, as measured by docking scores, expected stereochemistry, and recoverability in commercial chemical databases. Predicted protein-ligand contacts were found among highest-ranked docking poses with a high recovery rate. This work shows how the structural context of a protein target can be used to enhance molecule generation.