MUMBAI, India, Feb. 6 -- Intellectual Property India has published a patent application (202541126613 A) filed by Gitam, Visakhapatnam, Andhra Pradesh, on Dec. 14, 2025, for 'graph-attention-network-based computational framework for predicting high-efficacy drug candidates using pharmacophore fingerprints and similarity-driven label generation.'
Inventor(s) include Mr. Sandhi Kranthi Reddy; Dr. S V G Reddy; Dr. V Valli Kumari; Prof. K. Thammi Reddy; and Dr. Revelle Akshara.
The application for the patent was published on Feb. 6, under issue no. 06/2026.
According to the abstract released by the Intellectual Property India: "The invention provides a computational framework for predicting the therapeutic efficacy of novel chemical entities by integrating pharmacophore fingerprint generation, similarity-based quartile labelling, molecular graph construction and graph-attention-network-driven classification. A reference molecule exhibiting known biological activity is used as the basis for computing Jaccard similarity values across large molecular datasets, which are then statistically transformed into four categorical efficacy labels. Each molecule is converted into a graph representation preserving atomic and bonding characteristics, enabling the Graph Attention Network to learn structural features relevant to biological activity through attention-based aggregation. The framework demonstrates high predictive capability, achieving an accuracy of approximately eighty-eight percent on large-scale datasets. The invention significantly accelerates early-stage drug discovery by providing an automated, scalable and data-driven system for screening vast chemical libraries and identifying high-potential drug candidates for further experimental evaluation."
Disclaimer: Curated by HT Syndication.
