Local lab/research prototype
J Thrust
Making cancer research flow smoothly, so your life can too. Built for researchers, made for the world.
The problem
Variant uncertainty slows research teams down.
In DNA testing, many DNA variants are classified as uncertain, meaning researchers can’t determine if they’re harmless or harmful. Most of these are VUSs, or variants of uncertain significance. Unfortunately, VUSs don’t provide crucial information for research, acting as a bottleneck to the research. While 90% of VUSs are later reclassified as benign, the remaining 10% are vital for most studies. However, reclassifying and finding evidence for each variant takes weeks to months. Multiply this by the number of variants in data sets, which can reach 340 for a small control group of 100 people, and you can see the problem: time.
Our solution
A routing system for missing evidence.
Our solution is to build an AI system that looks at genetic variants and organizes the evidence by priority. It would then suggest the next best steps: whether a literature review is needed, whether splicing or RNA evidence could reclassify the variant, or whether other evidence factors should be collected. It would also back up every claim with references from trusted datasets like ClinVar, gnomAD, ClinGen, MaveDB, and PubMed. On top of that, there would be an AI tool specifically programmed to summarize and explain the reasoning clearly, and in a way researchers can actually use.
What J Thrust does
Accepts curated variant datasets and validates the required columns.
Uses transparent rules to prioritize variants by evidence-resolution opportunity.
Maps each variant to RNA/splicing, family, population, literature, functional, or reanalysis paths.
Uses constrained AI only to explain already-structured routing results.
Download and run locally
Download the prototype ZIP, extract it, and follow the separate setup page for Mac or Windows.