Success rates of single target RF models in predicting kinase inhibitory activity of selected eight compounds. Success rates represent the percentage of correctly predicted kinase activity obtained for each compound for 20 experimentally tested targets. Credit: International Journal of Molecular Sciences (2025). doi:10.3390/ijms26052157
Research on protein kinases provides an opportunity to explore molecular targets in the body to treat diseases such as cancer and autoimmune disorders. These enzymes bind to cell sites and can inhibit dysfunctional behavior such as overproduction of cancer cells and tumorigenesis.
With the vast combination of possible kinases and cell structures, scientists are turning to artificial intelligence (AI) to predict and create models where pairings can have therapeutic benefits.
A team of researchers has developed Kinasepred, a computational tool for predicting small molecule kinase targets. They published details of their project in the International Journal of Molecular Sciences.
The authors include Antonio Giordano, Dr. Maryland, professor at Temple University, scientists at the University of Pisa, and researchers working with the Sbarro Health Research Organization (SHRO) under the direction of other research institutes in Italy. This AI-based workflow can predict kinase activity, gain insight into molecular target interactions, and identify potential and combinations for cancer treatment.
The study, led by Dr. Miliana de Stefano of the University of Pisa Pharmacy, presents an advanced computational tool designed to enhance prediction of kinase interactions with small molecules.
Kinasepred is an example of a data-dependent computational tool developed to solve a specific problem, namely the selection of kinase inhibitors. This is achieved by applying predictive models using the molecular basis of kinase binding and selectivity.
Kinasepred uses machine learning (ML) and AI to make accurate predictions and describe molecular properties that promote interaction. Researchers hope that the tool will use new representations of molecules and various machine learning methods to lead to more accurate predictions and provide a more comprehensive knowledge of kinase interactions.
These advances are important to identify and minimize off-target effects and ultimately increase the safety and selectivity of therapeutic agents.
Details: Miriana Di Stefano et al, Kinasepred: A computational tool for predicting small molecule kinase targets, International Journal of Molecular Sciences (2025). doi:10.3390/ijms26052157
Provided by Sbarro Health Research Organization (SHRO)
Citation: The computational model enhances protein kinase target prediction (March 11, 2025) of treatments obtained from March 11, 2025 from https://phys.org/news/2025-03-protein-kinase-therapies.html.
This document is subject to copyright. Apart from fair transactions for private research or research purposes, there is no part that is reproduced without written permission. Content is provided with information only.
