Truace
Subscribe

Integrating graph attention networks and deep learning with multi-stage computational screening: Bisdemethoxycurcumin as a potent A₂A receptor inhibitor for Parkinson's disease

Parkinson's disease (PD) remains a progressive neurodegenerative disorder with limited disease modifying therapies, as current treatments such as L-DOPA offer only symptomatic relief and are associated with severe adverse effects. This study presents an integrated multi-stage computational pipeline that combines quantitative structure activity relationship (QSAR) modelling enhanced by graph attention networks (GATs), molecular docking, density functional theory (DFT), molecular dynamics (MD) simulations, and ADM…

By the Truvace desk1 min read1 sourcePeer-reviewedTRV-2026-1990· cite & verify
Integrating graph attention networks and deep learning with multi-stage computational screening: Bisdemethoxycurcumin as a potent A₂A receptor inhibitor for Parkinson's disease
In-silico-mechanistic-analysis-of-IRF3-inactivation-and-high-risk-HPV-E6-species-dependent-drug-srep13446-s3 by Shah M, Anwar M, Park S, Jafri S, Choi S. CC BY 4.0 · https://creativecommons.org/licenses/by/4.0

In brief

Researchers built a multi-stage computational pipeline for Parkinson's disease drug discovery. Graph attention networks were used in QSAR modelling on 1393 compounds, followed by docking, density functional theory, molecular dynamics and toxicity profiling of six phytochemicals from turmeric and sage.

Among the compounds tested, bisdemethoxycurcumin stood out as the best predicted A2A receptor binder and appeared safer than L-DOPA in silico. All of this remains computational, so the findings only justify further laboratory and animal testing and do not show a working therapy.

Main points

  1. The graph attention network model outperformed Extra Trees, Random Forest and SVR-RBF in predictive accuracy, with R2 of 0.991 against 0.742, 0.761 and 0.65.
  2. Bisdemethoxycurcumin docked to the A2A receptor at -9.1 kcal/mol, compared with -7.2 kcal/mol for L-DOPA.
  3. A 200 ns molecular dynamics run showed stable binding with no dissociation events.
  4. ADMET tools predicted blood brain barrier permeability and toxicity Class 5 (LD50 2560 mg/kg), against Class 4 for L-DOPA.

The gain

A graph attention network QSAR model combined with docking, simulation and ADMET screening singled out bisdemethoxycurcumin as a predicted stronger A2A receptor binder with a better predicted toxicity profile than L-DOPA.

The rundown

The study targeted four Parkinson's-related proteins: A2A receptor, GABA-A receptor, Nrf2 and alpha-synuclein. The training set was derived from 6-hydroxybenzothiazole 2-carboxamide and related scaffolds, with 312 molecular descriptors in the feature pool.

Y-randomization with 100 permutations was used to check that the models were not capturing chance correlations. Docking covered ar-turmerone, alpha-turmerone, bisdemethoxycurcumin, camphor, 1,8-cineole and thujone isomers.

Quantum calculations showed bisdemethoxycurcumin had the lowest HOMO-LUMO gap (0.1428 eV) and the highest electrophilicity index (0.1477). The molecular dynamics run showed 3-8 persistent hydrogen bonds, averaging about 5.

Sources

  1. Peer-reviewedBioorganic Chemistry2026-10-14

The debate