Machine learning model for predicting detergent properties
A machine learning model utilizing molecular dynamics simulations to predict novel detergent properties for applications in textiles, hair, and skin care. The model employs methods like Lasso regression to analyze detergent-lipid, protein, and polymer interactions.
Overview
This innovative solution leverages machine learning (ML) to predict the properties of novel detergents by analyzing their interactions with various substrates such as lipids, proteins, and polymers. By integrating molecular dynamics (MD) simulations and quantitative structure-property relationship (QSPR) studies, this model aims to accurately forecast detergent characteristics, enhancing formulations for applications in textiles, hair, and skin care.
Technical specifications
- Molecular Dynamics Simulations: Conducts 300-ns long simulations with up to 120 detergent molecules, assessing interactions with substrates such as 18-MEA monolayers, keratin, and cellulose.
- Machine Learning Methods: Utilizes Lasso regression and tree-based methods to train models on simulation data, requiring relatively low computational resources.
- Similarity Metrics: Employs the Tanimoto metric for comparing chemical structures, aiding in the prediction of untested molecule properties.
- Properties Analyzed: Includes viscosity changes, film formation, and polymer-surfactant complexation.
Technology readiness level
Currently at Technology Readiness Level 3, this model has been validated through extensive simulations, and the next steps involve experimental validation to confirm the theoretically calculated properties. The research is structured into three work packages, focusing on simulations, QSPR studies, and model application to unseen compounds.
About National Institute of Chemistry (Kemijski inštitut)
Slovenia’s National Institute of Chemistry (Kemijski inštitut) is a leading public research institute based in Ljubljana, bringing together multidisciplinary teams, advanced infrastructure, and close ties with universities to address industry‑relevant challenges. Industry partners engage through contract research, analytical services, collaborative projects, and access to shared facilities such as high‑field NMR, mass spectrometry, and advanced microscopy. Its location near established pharmaceutical and specialty‑chemical manufacturers in Slovenia and Central Europe supports agile partnering, talent pipelines, and rapid technology validation. Research is supported by competitive European Commission programs and national research funding. A dedicated technology transfer office manages IP, licensing, and spinout support, offering clear pathways from discovery to deployment.
delivered to your inbox.