Title
Integrating Machine Learning with Computational Chemistry to Predict Xylan Chain Properties at Scale
Authors
[ 1 ] Instytut Technologii i Inżynierii Chemicznej, Wydział Technologii Chemicznej, Politechnika Poznańska | [ P ] employee
Scientific discipline (Law 2.0)
Year of publication
2025
Chapter type
chapter in monograph / paper
Publication language
english
Keywords
Abstract
EN This study demonstrates the integration of advanced computational frameworks, including machine learning (ML) and quantum chemical methods, to address challenges in molecular modeling of xylan chains. Trained ML model based on the SchNet architecture was validated against classical density functional theory (DFT) results for xylan homologs up to 15 units. The ML model is employed to investigate energetic trends in considerably longer chains, up to 80 units, which would be computationally challenging for DFT due to both memory and time constraints. The comparison demonstrate that the ML model accurately captures structural and energetic trends, thereby illustrating its potential as a viable tool for studying large polysaccharide systems. This work highlights the respective strengths and limitations of both approaches, thereby providing a foundation for further ML-based exploration of complex biopolymers and their interactions in biological and industrial perspectives, and contributes to the growing trend of leveraging machine learning frameworks to enhance system development and scalability in computational science.
Date of online publication
13.08.2025
Pages (from - to)
66 - 75
Ministry points / chapter
20
System created by Poznań University of Technology
and Poznan Supercomputing and Networking Center
Log in through eKonto to add to SIS