Related Experiment Video
Updated: Apr 18, 2026
09:49
In Vitro Assay for Studying the Aggregation of Tau Protein and Drug Screening
Published on: November 20, 2018
20.1K
Machine Learning-Driven Ensemble Screening of Multitarget Kinase Inhibitors for Tauopathy-Associated
Arunabh Choudhury1, Mohammad Umar Saeed1, Sneh Prabha1
1Centre for Interdisciplinary Research in Basic Sciences, Jamia Millia Islamia, Jamia Nagar, New Delhi 110025, India.
ACS Chemical Neuroscience
|April 17, 2026
Summary
Researchers identified two natural compounds, CNP0591834.1 and CNP0484145.0, as potential treatments for tauopathies. These compounds target key kinases like DYRK1A, TTBK1, and ABL1, aiming to reduce tau hyperphosphorylation and neurodegeneration.
Area of Science:
- Neuroscience
- Pharmacology
- Computational Biology
Background:
- Tauopathies result from disrupted tau protein function due to kinase/phosphatase imbalance.
- Dysregulation of kinases like DYRK1A, TTBK1, and ABL1 leads to excessive tau phosphorylation and neurofibrillary tangles.
- These kinases contribute to tau pathology through various mechanisms, including altered splicing, aggregation, and Aβ-induced synaptic dysfunction.
Purpose of the Study:
- To discover natural product-derived multitarget inhibitors for key kinases implicated in tauopathies.
- To develop and apply a machine learning workflow for screening natural compounds against DYRK1A, TTBK1, and ABL1.
- To identify and validate potent natural compounds with the potential to mitigate tau-hyperphosphorylation-driven neurodegeneration.
Main Methods:
- Developed a machine learning workflow using five classifiers (CatBoost, SVM, KNN, Naive Bayes, XGBoost) trained on bioactivity data.
- Employed stratified sampling and SMOTE for class imbalance, and Bemis-Murcko scaffold splitting for data-scarce sets.
- Utilized a soft-voting ensemble model, consensus molecular docking, deep learning rescoring (GNINA), and molecular dynamics simulations for compound identification and validation.
Main Results:
- An ensemble model integrating CatBoost, XGBoost, and SVM demonstrated superior performance in kinase inhibition prediction.
- Screening of ~695,000 natural compounds identified two high-potential lead molecules: CNP0591834.1 and CNP0484145.0.
- Molecular dynamics simulations confirmed the stability and strong binding affinities of these lead compounds, particularly against DYRK1A and ABL1.
Conclusions:
- The identified natural compounds, CNP0591834.1 and CNP0484145.0, show significant potential as multitarget inhibitors for tauopathy treatment.
- This integrative computational framework successfully identified potent leads for mitigating tau-hyperphosphorylation-driven neurodegeneration.
- Further investigation of these compounds could lead to novel therapeutic strategies for tauopathies.
