Ali Alquraini, Mohammed Alahmari, Faisal K Alkholifi, Yahia Alghazwani, Marwa Qadri, Fadhel A. Alomar, Daniyah A. Almarghalani, Mohd Imran
{"title":"Machine Learning-Guided Discovery of Potential Natural KRAS-G12D Inhibitors for Pancreatic Carcinoma","authors":"Ali Alquraini, Mohammed Alahmari, Faisal K Alkholifi, Yahia Alghazwani, Marwa Qadri, Fadhel A. Alomar, Daniyah A. Almarghalani, Mohd Imran","doi":"10.1007/s12247-026-11123-7","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><p>The mutation KRAS-G12D found in pancreatic cancer is one of the biggest barriers to developing targeted therapy against pancreatic cancer. This research work tries to discover the potential KRAS-G12D inhibitors through natural products guided by machine learning approaches.</p><h3>Methods</h3><p>The screening process involved a library of 3,634 natural compounds sourced from the SelleckChem database, which were then filtered using the established QSAR model. Among these compounds, 176 potential compounds with appropriate predicted activities (pIC50) were chosen for subsequent testing. The 176 compounds were then docked against the switch-II pocket of the KRAS-G12D mutant protein. Compounds exhibiting docking scores equal to or lower than that of the reference inhibitor MRTX1133 (− 8.9 kcal/mol) were selected for subsequent ADMET and molecular dynamics analyses. The potential candidates were analyzed for their ADMET properties and binding interactions in silico. Further validation of the selected compounds was done by performing a molecular dynamics simulation for 300 ns along with a reference compound.</p><h3>Results</h3><p>KR1 and KR2 compounds were observed to have stable binding to the KRAS-G12D protein with RMSD values between 0.18 nm and 0.25 nm. MM-GBSA binding free energy calculations revealed a favorable interaction with energy levels of -13.57 kcal/mol and − 20.48 kcal/mol, respectively. The DFT analysis revealed that KR1 is more chemically stable due to high HOMO-LUMO gap values (4.669 eV) and low electrophilicity values (0.92 eV). However, KR2 was found to have balanced reactivity properties, with a relatively high electronegativity value (3.537 eV) and ionization potential value (4.828 eV).</p><h3>Conclusion</h3><p>Overall, these computations support the potential of KR1 and KR2 as valuable potential lead compounds of KRAS-G12D. Clearly, these results provide a robust basis for subsequent experimentation to validate their biological effects and therapeutic potential.</p></div>","PeriodicalId":656,"journal":{"name":"Journal of Pharmaceutical Innovation","volume":"22 1","pages":""},"PeriodicalIF":2.6000,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Pharmaceutical Innovation","FirstCategoryId":"3","ListUrlMain":"https://link.springer.com/article/10.1007/s12247-026-11123-7","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"PHARMACOLOGY & PHARMACY","Score":null,"Total":0}
引用次数: 0
Abstract
Background
The mutation KRAS-G12D found in pancreatic cancer is one of the biggest barriers to developing targeted therapy against pancreatic cancer. This research work tries to discover the potential KRAS-G12D inhibitors through natural products guided by machine learning approaches.
Methods
The screening process involved a library of 3,634 natural compounds sourced from the SelleckChem database, which were then filtered using the established QSAR model. Among these compounds, 176 potential compounds with appropriate predicted activities (pIC50) were chosen for subsequent testing. The 176 compounds were then docked against the switch-II pocket of the KRAS-G12D mutant protein. Compounds exhibiting docking scores equal to or lower than that of the reference inhibitor MRTX1133 (− 8.9 kcal/mol) were selected for subsequent ADMET and molecular dynamics analyses. The potential candidates were analyzed for their ADMET properties and binding interactions in silico. Further validation of the selected compounds was done by performing a molecular dynamics simulation for 300 ns along with a reference compound.
Results
KR1 and KR2 compounds were observed to have stable binding to the KRAS-G12D protein with RMSD values between 0.18 nm and 0.25 nm. MM-GBSA binding free energy calculations revealed a favorable interaction with energy levels of -13.57 kcal/mol and − 20.48 kcal/mol, respectively. The DFT analysis revealed that KR1 is more chemically stable due to high HOMO-LUMO gap values (4.669 eV) and low electrophilicity values (0.92 eV). However, KR2 was found to have balanced reactivity properties, with a relatively high electronegativity value (3.537 eV) and ionization potential value (4.828 eV).
Conclusion
Overall, these computations support the potential of KR1 and KR2 as valuable potential lead compounds of KRAS-G12D. Clearly, these results provide a robust basis for subsequent experimentation to validate their biological effects and therapeutic potential.
期刊介绍:
The Journal of Pharmaceutical Innovation (JPI), is an international, multidisciplinary peer-reviewed scientific journal dedicated to publishing high quality papers emphasizing innovative research and applied technologies within the pharmaceutical and biotechnology industries. JPI''s goal is to be the premier communication vehicle for the critical body of knowledge that is needed for scientific evolution and technical innovation, from R&D to market. Topics will fall under the following categories:
Materials science,
Product design,
Process design, optimization, automation and control,
Facilities; Information management,
Regulatory policy and strategy,
Supply chain developments ,
Education and professional development,
Journal of Pharmaceutical Innovation publishes four issues a year.