[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-123103-en":3,"doc-seo-123103-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},123103,687197207639,"Asher","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data","Machine learning scoring models for binding affinity can be fast, yet they often lack robust evaluation and struggle with practical ranking tasks seen in hit-to-lead optimisation. This work introduces AEV-PLIG, an attention-based graph neural network (atomic environment vector–protein ligand interaction graph), and a more realistic out-of-distribution OOD Test. Benchmarks on CASF-2016 and an FEP-derived test set show improved performance, and augmented data generated via template-based modelling or molecular docking enhances FEP benchmark correlation and ranking while greatly reducing compute time versus FEP+.","A Nature Portfolio journal  \n[https://doi.org/10.1038/s42004-025-01428-y](https://doi.org/10.1038/s42004-025-01428-y)  \n\n| Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data\u003Cbr> Check for updates |  |\n| --- | --- |\n| Ísak Valsson 1,4, Matthew T. Warren 2,4, Charlotte M. Deane 1, Aniket Magarkar 3 , Garrett M. Morris 1  & Philip C. Biggin 2  |  |\n| 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 affinityofa 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 ofAEV-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. |  |\n| Predicting the change in free energy upon binding of a protein and a ligand represents a cornerstone ofcomputational small molecule drug discovery. It is essential during hit identification, where one aims to identify binders that demonstrate high affinity for a target, as well as in hit-to-lead and lead optimisation, where binding affinity must be optimised alongside a number of other properties pertinent to safety and biological efficacy. Given the vastness of chemical space—and hence possible drug candidates1—computer-aided drug design(CADD)can play an important role in accelerating the discovery process by enabling large numbers of compounds to be screened in silico, avoiding the significant cost and resources required for experimental measurements2. The need for fast and accurate virtual screening methods is self-evident.\u003Cbr>To predict binding affinity, computational methods typically employ knowledge-or physics-based approaches that depend on statistical potentialsorthose based on molecular mechanics forcefields, respectively3. These approaches offer a trade-off between computational cost and accuracy: fast | estimations of binding affinity, such as scoring functions used in molecular docking, rely on heuristics and physical approximations that can limit their accuracy4. In contrast, alchemical binding free energy (BFE) simulation methods using all-atom molecular dynamics(MD)in explicit solvent offeramore rigorous but more expensive approach to compute either the absolute binding free energy (ABFE) of a ligand and protein5 or the relative binding free energy (RBFE) between two similar ligands that bind to the same protein6. A popular class of alchemical method, namely free energy perturbation (FEP)7 theory, often performed using the FEP+ workflow8, has demonstrated performance approaching the limits of achievable chemical accuracy (~1 kcal/mol) for certain systems9, 10. However, FEP (herein used interchangeably with alchemical BFE calculations) suffers from several limitations, including: a strong dependence on the choice of MD force field9, 11; a need for extensive and custom preparation (although see ref. 12 and ref. 13 for recent progress in this area) and param","cbCaijhoitXx6AaZ","https://ap.wps.com/l/cbCaijhoitXx6AaZ","pdf",1423836,1,12,"English","en",105,"# Introduction\n## Binding affinity prediction in drug discovery\n## Physics-based methods and FEP limitations\n## Machine learning scoring functions and evaluation gap\n# Proposed approaches\n## AEV-PLIG model\n## Out-of-distribution OOD Test and benchmarks\n## Augmented data strategy and results","[{\"question\":\"What problem does the work address in machine learning binding affinity models?\",\"answer\":\"It addresses two issues: lack of robust evaluation and poor performance on ranking tasks encountered in hit-to-lead optimisation, such as ordering binding affinities within congeneric ligand series.\"},{\"question\":\"What is AEV-PLIG?\",\"answer\":\"AEV-PLIG is an attention-based graph neural network that models atomic environments and protein–ligand interactions using an interaction graph.\"},{\"question\":\"How does augmented data improve binding affinity predictions relative to FEP benchmarks?\",\"answer\":\"Augmented data generated via template-based modelling or molecular docking increases correlation and ranking performance on the FEP benchmark, narrowing the gap with physics-based FEP calculations while being much faster.\"}]","Narrowing the gap between machine learning scoring functions and free energy perturbation using augmented data | PDF",1785814653,30,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"narrowing-the-gap-between-machine-learning-scoring-functions-and-free-energy-perturbation-using-augmented-data","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/narrowing-the-gap-between-machine-learning-scoring-functions-and-free-energy-perturbation-using-augmented-data/123103/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05","2026-08-04",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What problem does the work address in machine learning binding affinity models?","Question",{"text":76,"@type":77},"It addresses two issues: lack of robust evaluation and poor performance on ranking tasks encountered in hit-to-lead optimisation, such as ordering binding affinities within congeneric ligand series.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"What is AEV-PLIG?",{"text":81,"@type":77},"AEV-PLIG is an attention-based graph neural network that models atomic environments and protein–ligand interactions using an interaction graph.",{"name":83,"@type":74,"acceptedAnswer":84},"How does augmented data improve binding affinity predictions relative to FEP benchmarks?",{"text":85,"@type":77},"Augmented data generated via template-based modelling or molecular docking increases correlation and ranking performance on the FEP benchmark, narrowing the gap with physics-based FEP calculations while being much faster.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":29,"slug":122},"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":107,"slug":138},19,"General","general"]