Rare disease patients often wait years for a diagnosis, and even after one arrives, the path to a therapeutic strategy is fragmented across disconnected databases. Clinicians and researchers must manually cross-reference OMIM, ClinVar, UniProt, gnomAD, PubMed, and Open Targets to build even a basic picture of a gene's therapeutic potential, a process that demands significant bioinformatics expertise and can take days to weeks. RareSight was built around a single clinical question: given a patient's sequencing report, what do we do next?

The platform accepts a gene symbol and an optional variant in HGVS notation and orchestrates calls across six databases to return a structured report. This includes disease mechanism classification, therapy modality recommendations spanning AAV gene replacement, ASO/siRNA silencing, CRISPR base editing, and mRNA therapy, ranked CRISPR guide RNA candidates, ClinVar variant interpretation with UniProt protein-level feature intersection, gnomAD population frequency, and known drugs from Open Targets. From there, the pipeline extends into experimental design: users can generate customised wet lab validation protocols tailored to their cell model, equipment, and replicate structure, which are then translated into robot-executable Python code for the Opentrons liquid handling platform for fully autonomous assay execution. The backend is a Python Flask application with a custom multi-database orchestration pipeline, and the frontend is a single-page interface with interactive report rendering and ReportLab-powered PDF export.

Planned extensions include Cas12a PAM support, SpliceAI integration for splice-region variant assessment, and a patient case history module for tracking variants across a family. Longer term, RareSight is intended to serve as the front end for a fully automated assay parameter pipeline, from sequencing report to ready-to-order oligonucleotide sequences for ASO or siRNA synthesis.