MUMBAI, India, Oct. 5 -- Intellectual Property India has published a patent application (202641115484 A) filed by Rns Institute Of Technology; Samarth M G; Santosh D; Shashank V; Shree Ashwin; and Ms. Rashmi B C on September 26, 2026, for Moleculex- An Ai- Powered Genomic Diagnostics Using Large Dna Language Models Evo2 For Mutation Pathogenicity Prediction.
Inventors include Samarth M G; Santosh D; Shashank V; Shree Ashwin; and Ms. Rashmi B C.
The application for the patent was published on October 02, 2026, under issue no. 40/2026.
Abstract: This project presents MoleculeX, a long-context, LLM-powered variant analysis system built to evaluate the pathogenicity of genomic Single Nucleotide Variants (SNVs). The platform leverages Evo2, a state-of-the-art nucleotide foundation model trained on trillions of bases, to compute likelihood-based delta scores for mutated genomic sequences. A serverless GPU pipeline deployed on Modal enables scalable, high-throughput inference, while FastAPI manages variant requests and UCSC Genome Browser APIs supply precise genomic context. The frontend, developed in Next.js, provides an interactive interface for gene browsing, chromosome-level navigation, and variant-level analysis. Comprehensive validation experiments on BRCA1 and ClinVar datasets show robust predictive accuracy (AUROC 0.92) and stable calibration across functional classes. MoleculeX demonstrates that transformerbased genome models can generalize effectively to real-world variant interpretation tasks and forms a template for future AI-assisted genomic diagnostics.
Disclaimer: Curated by HT Syndication.
