Susanne Winiwarter
Susanne Winiwarter has a total of 134 co-authors. Here are the top co-authors by number of shared publications:
- Filip Miljković
Uppsala University
7 shared publications - Ya Chen
Department of Pharmaceutical Sciences, Division of Pharmaceutical Chemistry, Faculty of Life Sciences, University of Vienna, Josef-Holaubek-Platz 2, 1090 Vienna, Austria
4 shared publications Christopher Keefer
Translational Modeling and Simulation
3 shared publicationsIoana Oprisiu
Imaging and Data Analytics, Clinical Pharmacology & Safety Sciences, R&D, AstraZeneca, Gothenburg SE-43183, Sweden
3 shared publications- Rocío Mercado
Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Chalmersplatsen 1, 412 96 Gothenburg, Sweden
3 shared publications - Angelica Mazzolari
University of Milan
2 shared publications
Latest Publications
- 17 September 2026
- Published by Elsevier in Drug Metabolism and Disposition
- Vol. 54 (11) , 100409
A practical framework to predict human hepatic clearance for Extended Clearance Classification System class 1B acidic and zwitterionic drugs
- 17 September 2026
- Published by Elsevier in Drug Metabolism and Disposition
- Vol. 54 (11) , 100409
A practical framework to predict human hepatic clearance for Extended Clearance Classification System class 1B acidic and zwitterionic drugs
- Open Access
- 18 June 2026
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 23 (7) , 3816-3824
Computational Identification of Active Drug Metabolites for Human Protein Targets
Understanding drug metabolism helps mitigate toxicity risks and anticipate pharmacological effects beyond the parent compound. Here, we explore high-confidence in vitro human bioactivity annotations from the public domain associated with a curated data set of drug-metabolite pairs. Our findings show that 26% of all drug-metabolite-target combinations contain metabolites with retained or increased bioactivity relative to their parent drugs, showing a high proportion of active metabolites with potencies below 100 nM. Moreover, these active metabolites display similar predicted physicochemical and ADME properties to their parent drugs, indicating that their biological activity may be preserved in vivo. Two key features of the identified active metabolites are noted: (1) the metabolite is more likely to be formed via an O-demethylation or an N-demethylation at a tertiary nitrogen, and (2) these drug-metabolite pairs are frequently active against the membrane receptor target class. To support further research in this area, we make our data compilation, including pairs with identified active metabolites and recorded bioactivity data, openly available.
- 18 June 2026
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 23 (7) , 3816-3824
Open AccessComputational Identification of Active Drug Metabolites for Human Protein Targets
Understanding drug metabolism helps mitigate toxicity risks and anticipate pharmacological effects beyond the parent compound. Here, we explore high-confidence in vitro human bioactivity annotations from the public domain associated with a curated data set of drug-metabolite pairs. Our findings show that 26% of all drug-metabolite-target combinations contain metabolites with retained or increased bioactivity relative to their parent drugs, showing a high proportion of active metabolites with potencies below 100 nM. Moreover, these active metabolites display similar predicted physicochemical and ADME properties to their parent drugs, indicating that their biological activity may be preserved in vivo. Two key features of the identified active metabolites are noted: (1) the metabolite is more likely to be formed via an O-demethylation or an N-demethylation at a tertiary nitrogen, and (2) these drug-metabolite pairs are frequently active against the membrane receptor target class. To support further research in this area, we make our data compilation, including pairs with identified active metabolites and recorded bioactivity data, openly available.
- 29 May 2026
- Published by American Chemical Society (ACS) in Journal of Medicinal Chemistry
- Vol. 69 (11) , 12973-12981
Developing Predictive Models by Sharing Predictions – An Investigation of a Federated Learning Approach for ADMET Predictions
Machine learning models for ADMET prediction benefit from large, diverse data sets, yet such data are typically siloed across organizations. Federated learning (FL) enables collaborative modeling while preserving data privacy. Here, we investigate a student–teacher model (STM) framework in which organizations train internal models on proprietary data and share predictions on a public data set to generate pseudolabels for a centralized student model. As a proof of concept, 11 pharmaceutical companies contributed predictions for rat steady-state volume of distribution, yielding a pseudolabeled data set of ∼133,000 compounds. The resulting student model achieved performance comparable to individual teacher models on an external test set (RMSE ≈ 0.51 vs 0.47–0.61). Compared with FL approaches such as MELLODY and Effiris, STM offers a simpler workflow that avoids direct data sharing or iterative collaboration, providing a practical and scalable framework for secure cross-company model development.
- 29 May 2026
- Published by American Chemical Society (ACS) in Journal of Medicinal Chemistry
- Vol. 69 (11) , 12973-12981
Developing Predictive Models by Sharing Predictions – An Investigation of a Federated Learning Approach for ADMET Predictions
Machine learning models for ADMET prediction benefit from large, diverse data sets, yet such data are typically siloed across organizations. Federated learning (FL) enables collaborative modeling while preserving data privacy. Here, we investigate a student–teacher model (STM) framework in which organizations train internal models on proprietary data and share predictions on a public data set to generate pseudolabels for a centralized student model. As a proof of concept, 11 pharmaceutical companies contributed predictions for rat steady-state volume of distribution, yielding a pseudolabeled data set of ∼133,000 compounds. The resulting student model achieved performance comparable to individual teacher models on an external test set (RMSE ≈ 0.51 vs 0.47–0.61). Compared with FL approaches such as MELLODY and Effiris, STM offers a simpler workflow that avoids direct data sharing or iterative collaboration, providing a practical and scalable framework for secure cross-company model development.
- 31 March 2026
- Published by American Chemical Society (ACS) in ACS Medicinal Chemistry Letters
- Vol. 17 (4) , 890-899
Evaluation of Oral PROTAC Guidelines: Efflux Ratio Outweighs Chameleonicity Descriptors
Oral bioavailability of PROTACs, which often fall outside the Rule-of-Five space, is still not perfectly understood. Thus, the design of orally bioavailable PROTACs remains challenging. Chameleonicity, the ability of a compound to adapt its conformation to different environments (solvent or membrane), has been suggested as a key molecular property. Here, using a diverse set of PROTACs from our portfolio, we evaluate whether published guidelines including predicted or experimental chameleonicity descriptors could refine AstraZeneca’s internal guidelines for designing orally absorbed PROTACs. We did not find such a trend. Instead, reducing the efflux ratio in Caco-2 cells emerged as a useful addition to our guidelines.
- 31 March 2026
- Published by American Chemical Society (ACS) in ACS Medicinal Chemistry Letters
- Vol. 17 (4) , 890-899
Evaluation of Oral PROTAC Guidelines: Efflux Ratio Outweighs Chameleonicity Descriptors
Oral bioavailability of PROTACs, which often fall outside the Rule-of-Five space, is still not perfectly understood. Thus, the design of orally bioavailable PROTACs remains challenging. Chameleonicity, the ability of a compound to adapt its conformation to different environments (solvent or membrane), has been suggested as a key molecular property. Here, using a diverse set of PROTACs from our portfolio, we evaluate whether published guidelines including predicted or experimental chameleonicity descriptors could refine AstraZeneca’s internal guidelines for designing orally absorbed PROTACs. We did not find such a trend. Instead, reducing the efflux ratio in Caco-2 cells emerged as a useful addition to our guidelines.
- Open Access
- 24 February 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Systematic Computational Identification of Active Drug Metabolites Across the Human Proteome
Understanding drug metabolism helps mitigate toxicity risks and anticipate pharmacological effects beyond the parent compound. Here, we explore high-confidence in vitro human bioactivity annotations from the public domain associated with a curated dataset of drug-metabolite pairs. Our findings show that 26% of all drug-metabolite-target combinations contain metabolites with retained or increased bioactivity relative to their parent drugs, showing a high proportion of active metabolites with potencies below 100 nM. Moreover, these active metabolites display similar predicted physicochemical and ADME properties to their parent drugs, indicating that their biological activity may be preserved in vivo. Two key features of the identified active metabolites are noted: 1) the metabolite is more likely to be formed via an O-demethylation or an N-demethylation at a tertiary nitrogen, and 2) these drug-metabolite pairs are frequently active against the membrane receptor target class. To support further research in this area, we make our data compilation, including pairs with identified active metabolites and recorded bioactivity data, openly available.
- 24 February 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessSystematic Computational Identification of Active Drug Metabolites Across the Human Proteome
Understanding drug metabolism helps mitigate toxicity risks and anticipate pharmacological effects beyond the parent compound. Here, we explore high-confidence in vitro human bioactivity annotations from the public domain associated with a curated dataset of drug-metabolite pairs. Our findings show that 26% of all drug-metabolite-target combinations contain metabolites with retained or increased bioactivity relative to their parent drugs, showing a high proportion of active metabolites with potencies below 100 nM. Moreover, these active metabolites display similar predicted physicochemical and ADME properties to their parent drugs, indicating that their biological activity may be preserved in vivo. Two key features of the identified active metabolites are noted: 1) the metabolite is more likely to be formed via an O-demethylation or an N-demethylation at a tertiary nitrogen, and 2) these drug-metabolite pairs are frequently active against the membrane receptor target class. To support further research in this area, we make our data compilation, including pairs with identified active metabolites and recorded bioactivity data, openly available.
- Open Access
- 5 November 2025
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 10 (45) , 54148-54159
Prediction of Permeability and Efflux Using Multitask Learning
In silico prediction of cell membrane permeability is crucial in drug discovery, since a compound’s permeation through membranes influences parameters such as its in vivo efficacy, bioavailability, and pharmacokinetics. This study investigates the use of multitask graph neural networks to predict a selection of permeability-related endpoints. The study utilized a harmonized, single-laboratory internal data set of over 10K compounds measured in human colorectal adenocarcinoma (Caco-2) and Madin–Darby canine kidney (MDCK) cell lines, routinely employed in experimental assays for drug permeability and efflux. This data set is an order of magnitude larger than comparable public collections, thus providing greater statistical power and a consistent error profile for model development. A series of multitask learning (MTL) models trained on such data were benchmarked against single-task approaches and evaluated on an external public data set to investigate the model’s applicability domain. The comparison between the performance of single- and multitask models suggests that MTL can achieve higher accuracy by leveraging shared information across endpoints. MTL is also shown to perform better when augmented with molecular features. In particular, the inclusion of pKa and LogD, is shown to improve the accuracy of both permeability and efflux endpoints. This work presents benchmarking results of models utilizing different data splitting strategies, accompanied by guidelines for optimal validation in the context of MTL.
- 5 November 2025
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 10 (45) , 54148-54159
Open AccessPrediction of Permeability and Efflux Using Multitask Learning
In silico prediction of cell membrane permeability is crucial in drug discovery, since a compound’s permeation through membranes influences parameters such as its in vivo efficacy, bioavailability, and pharmacokinetics. This study investigates the use of multitask graph neural networks to predict a selection of permeability-related endpoints. The study utilized a harmonized, single-laboratory internal data set of over 10K compounds measured in human colorectal adenocarcinoma (Caco-2) and Madin–Darby canine kidney (MDCK) cell lines, routinely employed in experimental assays for drug permeability and efflux. This data set is an order of magnitude larger than comparable public collections, thus providing greater statistical power and a consistent error profile for model development. A series of multitask learning (MTL) models trained on such data were benchmarked against single-task approaches and evaluated on an external public data set to investigate the model’s applicability domain. The comparison between the performance of single- and multitask models suggests that MTL can achieve higher accuracy by leveraging shared information across endpoints. MTL is also shown to perform better when augmented with molecular features. In particular, the inclusion of pKa and LogD, is shown to improve the accuracy of both permeability and efflux endpoints. This work presents benchmarking results of models utilizing different data splitting strategies, accompanied by guidelines for optimal validation in the context of MTL.
- Open Access
- 21 October 2025
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 22 (11) , 6772-6787
Metabolite Identification Data in Drug Discovery, Part 2: Site-of-Metabolism Annotation, Analysis, and Exploration for Machine Learning
The ability to pinpoint and predict sites of metabolism (SoMs) is essential for designing and optimizing effective and safe bioactive small molecules. However, the number of molecules with annotated SoMs is limited, hindering the advancement of data-driven methods such as machine learning for metabolism prediction. Here, we provide a comprehensive characterization of SoM data obtained from the readouts of a human hepatocyte assay conducted at AstraZeneca Gothenburg. We explore a new strategy for SoM annotation that accounts for uncertainty in the experimental data, and we relate our findings to the most comprehensive SoM data collection available to date. Our study includes entropy analysis of SoM annotations, accompanied by representative examples that highlight the complexities of interpreting and working with metabolism data. Furthermore, we demonstrate the impact and value of the new metabolism data on SoM prediction. Importantly, a substantial portion of the data generated and analyzed as part of this work is made publicly available.
- 21 October 2025
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 22 (11) , 6772-6787
Open AccessMetabolite Identification Data in Drug Discovery, Part 2: Site-of-Metabolism Annotation, Analysis, and Exploration for Machine Learning
The ability to pinpoint and predict sites of metabolism (SoMs) is essential for designing and optimizing effective and safe bioactive small molecules. However, the number of molecules with annotated SoMs is limited, hindering the advancement of data-driven methods such as machine learning for metabolism prediction. Here, we provide a comprehensive characterization of SoM data obtained from the readouts of a human hepatocyte assay conducted at AstraZeneca Gothenburg. We explore a new strategy for SoM annotation that accounts for uncertainty in the experimental data, and we relate our findings to the most comprehensive SoM data collection available to date. Our study includes entropy analysis of SoM annotations, accompanied by representative examples that highlight the complexities of interpreting and working with metabolism data. Furthermore, we demonstrate the impact and value of the new metabolism data on SoM prediction. Importantly, a substantial portion of the data generated and analyzed as part of this work is made publicly available.
- Open Access
- 14 October 2025
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 22 (11) , 6788-6802
Metabolite Identification Data in Drug Discovery, Part 1: Data Generation and Trend Analysis
In drug discovery, metabolite identification data are used to identify metabolic soft spots in research molecules to facilitate reduced metabolism in subsequently designed compounds. In addition, knowledge about exact metabolite structures enables the assessment of risks associated with active, reactive, or toxic metabolites. In the present work, we exemplify how metabolite identification data are generated and utilized at AstraZeneca. We share metabolite transformation scheme data derived from incubations in human hepatocytes for a set of 120 compounds. Comparison with other in-house generated metabolite identification data, both in terms of chemical space analysis and in vitro properties, is performed, including the characterization of observed metabolic pathways. For selected compounds, the correlation between in vitro and in vivo metabolite data in animal species is provided. Finally, the usage of shared metabolite identification data for drug metabolism prediction using machine learning and artificial intelligence approaches is discussed.
- 14 October 2025
- Published by American Chemical Society (ACS) in Molecular Pharmaceutics
- Vol. 22 (11) , 6788-6802
Open AccessMetabolite Identification Data in Drug Discovery, Part 1: Data Generation and Trend Analysis
In drug discovery, metabolite identification data are used to identify metabolic soft spots in research molecules to facilitate reduced metabolism in subsequently designed compounds. In addition, knowledge about exact metabolite structures enables the assessment of risks associated with active, reactive, or toxic metabolites. In the present work, we exemplify how metabolite identification data are generated and utilized at AstraZeneca. We share metabolite transformation scheme data derived from incubations in human hepatocytes for a set of 120 compounds. Comparison with other in-house generated metabolite identification data, both in terms of chemical space analysis and in vitro properties, is performed, including the characterization of observed metabolic pathways. For selected compounds, the correlation between in vitro and in vivo metabolite data in animal species is provided. Finally, the usage of shared metabolite identification data for drug metabolism prediction using machine learning and artificial intelligence approaches is discussed.
- Open Access
- 9 July 2025
- Published by American Chemical Society (ACS) in Journal of Medicinal Chemistry
- Vol. 68 (14) , 14195-14222
Discovery of AZD4144, a Selective and Potent NLRP3 Inhibitor for the Treatment of Inflammatory Diseases
Inflammatory disease is a major burden on health care, and new treatments are needed to manage exaggerated inflammatory responses. The NLRP3 inflammasome is a key player in inflammation, the activation of which is believed to be associated with an increased cardiovascular and renal risk and play a major role in several common noncommunicable diseases. The emergence of evidence that small molecules can work as NLRP3 inhibitors has initiated a surge of drug discovery activities within the field. In this paper, we describe the discovery of the clinical candidate AZD4144. The compound inhibits NLRP3 selectively with high potency, has beneficial pharmacokinetic properties, and shows a favorable off-target pharmacology profile. We also show that AZD4144 binds directly to NLRP3 and competes with MCC950 for binding to the protein, indicating that the likely mode of inhibition of AZD4144 is to stabilize the inactive form of NLRP3.
- 9 July 2025
- Published by American Chemical Society (ACS) in Journal of Medicinal Chemistry
- Vol. 68 (14) , 14195-14222
Open AccessDiscovery of AZD4144, a Selective and Potent NLRP3 Inhibitor for the Treatment of Inflammatory Diseases
Inflammatory disease is a major burden on health care, and new treatments are needed to manage exaggerated inflammatory responses. The NLRP3 inflammasome is a key player in inflammation, the activation of which is believed to be associated with an increased cardiovascular and renal risk and play a major role in several common noncommunicable diseases. The emergence of evidence that small molecules can work as NLRP3 inhibitors has initiated a surge of drug discovery activities within the field. In this paper, we describe the discovery of the clinical candidate AZD4144. The compound inhibits NLRP3 selectively with high potency, has beneficial pharmacokinetic properties, and shows a favorable off-target pharmacology profile. We also show that AZD4144 binds directly to NLRP3 and competes with MCC950 for binding to the protein, indicating that the likely mode of inhibition of AZD4144 is to stabilize the inactive form of NLRP3.
- Open Access
- 21 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Metabolite Identification Data in Drug Discovery: Site-of-Metabolism Annotation, Analysis, and Exploration for Machine Learning
The ability to pinpoint and predict sites of metabolism (SoMs) is essential for designing and optimizing effective and safe bioactive small molecules, such as drugs. However, the number of molecules with annotated SoMs in the public domain and pharmaceutical industry is limited, hindering the advancement of data-driven methods like machine learning for metabolism prediction. Here, we provide a comprehensive characterization of SoM data obtained from the readouts of a human hepatocyte assay conducted at AstraZeneca Gothenburg. We explore a new strategy for SoM annotation that accounts for uncertainty in the experimental data, and we relate our findings to the most comprehensive SoM data collection available to date. Our study includes entropy analysis of SoM annotations, accompanied by representative examples that highlight the complexities of interpreting and working with metabolism data. Furthermore, we demonstrate the impact and value of the new metabolism data on SoM prediction. Importantly, a substantial portion of the data generated and analyzed as part of this work is made publicly available to support further research.
- 21 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessMetabolite Identification Data in Drug Discovery: Site-of-Metabolism Annotation, Analysis, and Exploration for Machine Learning
The ability to pinpoint and predict sites of metabolism (SoMs) is essential for designing and optimizing effective and safe bioactive small molecules, such as drugs. However, the number of molecules with annotated SoMs in the public domain and pharmaceutical industry is limited, hindering the advancement of data-driven methods like machine learning for metabolism prediction. Here, we provide a comprehensive characterization of SoM data obtained from the readouts of a human hepatocyte assay conducted at AstraZeneca Gothenburg. We explore a new strategy for SoM annotation that accounts for uncertainty in the experimental data, and we relate our findings to the most comprehensive SoM data collection available to date. Our study includes entropy analysis of SoM annotations, accompanied by representative examples that highlight the complexities of interpreting and working with metabolism data. Furthermore, we demonstrate the impact and value of the new metabolism data on SoM prediction. Importantly, a substantial portion of the data generated and analyzed as part of this work is made publicly available to support further research.