Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data
Abstract Machine learning offers great promise for fast and accurate binding affinity predictions. However, current models lack robust evaluation and fail on tasks encountered in (hit-to-) lead optimisation, such as ranking the binding affinity of a congeneric series of ligands, thereby limiting the...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-02-01
|
| Series: | Communications Chemistry |
| Online Access: | https://doi.org/10.1038/s42004-025-01428-y |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1825197578646454272 |
|---|---|
| author | Ísak Valsson Matthew T. Warren Charlotte M. Deane Aniket Magarkar Garrett M. Morris Philip C. Biggin |
| author_facet | Ísak Valsson Matthew T. Warren Charlotte M. Deane Aniket Magarkar Garrett M. Morris Philip C. Biggin |
| author_sort | Ísak Valsson |
| collection | DOAJ |
| description | Abstract Machine learning offers great promise for fast and accurate binding affinity predictions. However, current models lack robust evaluation and fail on tasks encountered in (hit-to-) lead optimisation, such as ranking the binding affinity of a congeneric series of ligands, thereby limiting their application in drug discovery. Here, we address these issues by first introducing a novel attention-based graph neural network model called AEV-PLIG (atomic environment vector–protein ligand interaction graph). Second, we introduce a new and more realistic out-of-distribution test set called the OOD Test. We benchmark our model on this set, CASF-2016, and a test set used for free energy perturbation (FEP) calculations, that not only highlights the competitive performance of AEV-PLIG, but provides a realistic assessment of machine learning models with rigorous physics-based approaches. Moreover, we demonstrate how leveraging augmented data (generated using template-based modelling or molecular docking) can significantly improve binding affinity prediction correlation and ranking on the FEP benchmark (weighted mean PCC and Kendall’s τ increases from 0.41 and 0.26 to 0.59 and 0.42). These strategies together are closing the performance gap with FEP calculations (FEP+ achieves weighted mean PCC and Kendall’s τ of 0.68 and 0.49 on the FEP benchmark) while being ~400,000 times faster. |
| format | Article |
| id | doaj-art-29105d7176d647a788aa531784053348 |
| institution | Kabale University |
| issn | 2399-3669 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Communications Chemistry |
| spelling | doaj-art-29105d7176d647a788aa5317840533482025-02-09T12:16:22ZengNature PortfolioCommunications Chemistry2399-36692025-02-018111210.1038/s42004-025-01428-yNarrowing the gap between machine learning scoring functions and free energy perturbation using augmented dataÍsak Valsson0Matthew T. Warren1Charlotte M. Deane2Aniket Magarkar3Garrett M. Morris4Philip C. Biggin5Oxford Protein Informatics Group, Department of Statistics, University of OxfordStructural Bioinformatics and Computational Biochemistry, Department of Biochemistry, University of OxfordOxford Protein Informatics Group, Department of Statistics, University of OxfordBoehringer Ingelheim Pharma GmbH & Co. KGOxford Protein Informatics Group, Department of Statistics, University of OxfordStructural Bioinformatics and Computational Biochemistry, Department of Biochemistry, University of OxfordAbstract Machine learning offers great promise for fast and accurate binding affinity predictions. However, current models lack robust evaluation and fail on tasks encountered in (hit-to-) lead optimisation, such as ranking the binding affinity of a congeneric series of ligands, thereby limiting their application in drug discovery. Here, we address these issues by first introducing a novel attention-based graph neural network model called AEV-PLIG (atomic environment vector–protein ligand interaction graph). Second, we introduce a new and more realistic out-of-distribution test set called the OOD Test. We benchmark our model on this set, CASF-2016, and a test set used for free energy perturbation (FEP) calculations, that not only highlights the competitive performance of AEV-PLIG, but provides a realistic assessment of machine learning models with rigorous physics-based approaches. Moreover, we demonstrate how leveraging augmented data (generated using template-based modelling or molecular docking) can significantly improve binding affinity prediction correlation and ranking on the FEP benchmark (weighted mean PCC and Kendall’s τ increases from 0.41 and 0.26 to 0.59 and 0.42). These strategies together are closing the performance gap with FEP calculations (FEP+ achieves weighted mean PCC and Kendall’s τ of 0.68 and 0.49 on the FEP benchmark) while being ~400,000 times faster.https://doi.org/10.1038/s42004-025-01428-y |
| spellingShingle | Ísak Valsson Matthew T. Warren Charlotte M. Deane Aniket Magarkar Garrett M. Morris Philip C. Biggin Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data Communications Chemistry |
| title | Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| title_full | Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| title_fullStr | Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| title_full_unstemmed | Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| title_short | Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| title_sort | narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data |
| url | https://doi.org/10.1038/s42004-025-01428-y |
| work_keys_str_mv | AT isakvalsson narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata AT matthewtwarren narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata AT charlottemdeane narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata AT aniketmagarkar narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata AT garrettmmorris narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata AT philipcbiggin narrowingthegapbetweenmachinelearningscoringfunctionsandfreeenergyperturbationusingaugmenteddata |