Powered by GNN + Transformer Hybrid AI

Your DNA. Decoded. Defended.

>Predicting Type 2 Diabetes risk at the genomic level
scroll to explore

Clinical-Grade Analysis Tools

every tool a genomics researcher needs, in one platform

GNN + Transformer AI

Hybrid deep learning pipeline combining Graph Attention Networks with DNABERT-2 embeddings for epistasis-aware risk scoring.

Interactive Knowledge Graph

Explore gene interactions powered by STRING DB with hand-sketched nodes, stitch-pattern edges, and pathway clustering.

SHAP Explainability

Waterfall and beeswarm visualizations showing exactly which genes push your risk score up or down, with biological annotations.

What-If Simulator

Toggle genes, swap alleles, and watch your risk score update in real-time. Counterfactual analysis at your fingertips.

Patient Comparison

Side-by-side comparison of two analyses with diff highlighting. Identify unique genetic risk factors between patients.

Clinical PDF Reports

Export publication-quality reports with risk gauges, SHAP charts, variant tables, and clinical methodology notes.

How It Works

from raw genes to actionable clinical insight in 5 steps

Step 1

Input Genes

Select genes of interest

Step 2

DNABERT-2 Embed

Generate sequence embeddings

Step 3

GAT Model

Graph attention inference

Step 4

SHAP Attribution

Explain predictions

Step 5

Risk Report

Generate clinical report

Integrated Stack

real APIs, real data, production-grade infrastructure

Supabase

Database, Auth & Row Level Security

Pinecone

Vector search for gene embeddings

Inngest

Background job orchestration

Stripe

Subscription billing (Free/Pro/Research)

Resend

Transactional email delivery

PostHog

Product analytics & event tracking

Sentry

Error tracking & monitoring

STRING DB

Real gene-gene interaction data

UniProt

Gene & protein annotation

ClinVar

Variant clinical significance

GWAS Catalog

Genome-wide association studies

Open Targets

Gene-disease evidence scores