Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning

The SARS-CoV-2 main protease (Mpro) is a validated therapeutic target for inhibiting viral replication. Few compounds have advanced clinically, underscoring the difficulty in optimizing both target affinity and drug-like properties. To address this challenge, we integrated machine learning (ML), mol...

Full description

Saved in:
Bibliographic Details
Main Authors: Anacleto Silva de Souza, Vitor Martins de Freitas Amorim, Eduardo Pereira Soares, Robson Francisco de Souza, Cristiane Rodrigues Guzzo
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Viruses
Subjects:
Online Access:https://www.mdpi.com/1999-4915/17/7/935
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849713325031555072
author Anacleto Silva de Souza
Vitor Martins de Freitas Amorim
Eduardo Pereira Soares
Robson Francisco de Souza
Cristiane Rodrigues Guzzo
author_facet Anacleto Silva de Souza
Vitor Martins de Freitas Amorim
Eduardo Pereira Soares
Robson Francisco de Souza
Cristiane Rodrigues Guzzo
author_sort Anacleto Silva de Souza
collection DOAJ
description The SARS-CoV-2 main protease (Mpro) is a validated therapeutic target for inhibiting viral replication. Few compounds have advanced clinically, underscoring the difficulty in optimizing both target affinity and drug-like properties. To address this challenge, we integrated machine learning (ML), molecular docking, and molecular dynamics (MD) simulations to investigate the balance between pharmacodynamic (PD) and pharmacokinetic (PK) properties in Mpro inhibitor design. We developed ML models to classify Mpro inhibitors based on experimental IC<sub>50</sub> data, combining molecular descriptors with structural insights from MD simulations. Our Support Vector Machine (SVM) model achieved strong performance (training accuracy = 0.84, ROC AUC = 0.91; test accuracy = 0.79, ROC AUC = 0.86), while our Logistic Regression model (training accuracy = 0.78, ROC AUC = 0.85; test accuracy = 0.76, ROC AUC = 0.83). Notably, PK descriptors often exhibited opposing trends to binding affinity: hydrophilic features enhanced binding affinity but compromised PK properties, whereas hydrogen bonding, hydrophobic, and π–π interactions in Mpro subsites S2 and S3/S4 are fundamental for binding affinity. Our findings highlight the need for a balanced approach in Mpro inhibitor design, strategically targeting these subsites may balance PD and PK properties. For the first time, we demonstrate antagonistic trends between pharmacokinetic (PK) and pharmacodynamic (PD) features through the integrated application of ML/MD. This study provides a computational framework for rational Mpro inhibitors, combining ML and MD to investigate the complex interplay between enzyme inhibition and drug likeness. These insights may guide the hit-to-lead optimization of the novel next-generation Mpro inhibitors of SARS-CoV-2 with preclinical and clinical potential.
format Article
id doaj-art-e89817838d204c0287ac27ddc33499e0
institution DOAJ
issn 1999-4915
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Viruses
spelling doaj-art-e89817838d204c0287ac27ddc33499e02025-08-20T03:13:59ZengMDPI AGViruses1999-49152025-06-0117793510.3390/v17070935Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine LearningAnacleto Silva de Souza0Vitor Martins de Freitas Amorim1Eduardo Pereira Soares2Robson Francisco de Souza3Cristiane Rodrigues Guzzo4Department of Microbiology, Institute of Biomedical Sciences, University of São Paulo, Sao Paulo 05508-000, BrazilDepartment of Microbiology, Institute of Biomedical Sciences, University of São Paulo, Sao Paulo 05508-000, BrazilDepartment of Microbiology, Institute of Biomedical Sciences, University of São Paulo, Sao Paulo 05508-000, BrazilDepartment of Microbiology, Institute of Biomedical Sciences, University of São Paulo, Sao Paulo 05508-000, BrazilDepartment of Microbiology, Institute of Biomedical Sciences, University of São Paulo, Sao Paulo 05508-000, BrazilThe SARS-CoV-2 main protease (Mpro) is a validated therapeutic target for inhibiting viral replication. Few compounds have advanced clinically, underscoring the difficulty in optimizing both target affinity and drug-like properties. To address this challenge, we integrated machine learning (ML), molecular docking, and molecular dynamics (MD) simulations to investigate the balance between pharmacodynamic (PD) and pharmacokinetic (PK) properties in Mpro inhibitor design. We developed ML models to classify Mpro inhibitors based on experimental IC<sub>50</sub> data, combining molecular descriptors with structural insights from MD simulations. Our Support Vector Machine (SVM) model achieved strong performance (training accuracy = 0.84, ROC AUC = 0.91; test accuracy = 0.79, ROC AUC = 0.86), while our Logistic Regression model (training accuracy = 0.78, ROC AUC = 0.85; test accuracy = 0.76, ROC AUC = 0.83). Notably, PK descriptors often exhibited opposing trends to binding affinity: hydrophilic features enhanced binding affinity but compromised PK properties, whereas hydrogen bonding, hydrophobic, and π–π interactions in Mpro subsites S2 and S3/S4 are fundamental for binding affinity. Our findings highlight the need for a balanced approach in Mpro inhibitor design, strategically targeting these subsites may balance PD and PK properties. For the first time, we demonstrate antagonistic trends between pharmacokinetic (PK) and pharmacodynamic (PD) features through the integrated application of ML/MD. This study provides a computational framework for rational Mpro inhibitors, combining ML and MD to investigate the complex interplay between enzyme inhibition and drug likeness. These insights may guide the hit-to-lead optimization of the novel next-generation Mpro inhibitors of SARS-CoV-2 with preclinical and clinical potential.https://www.mdpi.com/1999-4915/17/7/935drug discoveryhit-to-lead optimization challengesmain proteaseSARS-CoV-2machine learningmolecular dynamics simulations
spellingShingle Anacleto Silva de Souza
Vitor Martins de Freitas Amorim
Eduardo Pereira Soares
Robson Francisco de Souza
Cristiane Rodrigues Guzzo
Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
Viruses
drug discovery
hit-to-lead optimization challenges
main protease
SARS-CoV-2
machine learning
molecular dynamics simulations
title Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
title_full Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
title_fullStr Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
title_full_unstemmed Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
title_short Antagonistic Trends Between Binding Affinity and Drug-Likeness in SARS-CoV-2 Mpro Inhibitors Revealed by Machine Learning
title_sort antagonistic trends between binding affinity and drug likeness in sars cov 2 mpro inhibitors revealed by machine learning
topic drug discovery
hit-to-lead optimization challenges
main protease
SARS-CoV-2
machine learning
molecular dynamics simulations
url https://www.mdpi.com/1999-4915/17/7/935
work_keys_str_mv AT anacletosilvadesouza antagonistictrendsbetweenbindingaffinityanddruglikenessinsarscov2mproinhibitorsrevealedbymachinelearning
AT vitormartinsdefreitasamorim antagonistictrendsbetweenbindingaffinityanddruglikenessinsarscov2mproinhibitorsrevealedbymachinelearning
AT eduardopereirasoares antagonistictrendsbetweenbindingaffinityanddruglikenessinsarscov2mproinhibitorsrevealedbymachinelearning
AT robsonfranciscodesouza antagonistictrendsbetweenbindingaffinityanddruglikenessinsarscov2mproinhibitorsrevealedbymachinelearning
AT cristianerodriguesguzzo antagonistictrendsbetweenbindingaffinityanddruglikenessinsarscov2mproinhibitorsrevealedbymachinelearning