---
title: "🧬 Pan-Genomic Saudi Edge-VCF"
emoji: 🧬
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: "5.29.0"
python_version: "3.12"
app_file: app.py
suggested_hardware: cpu-basic
tags:
  - genomics
  - clinical
  - variant-interpretation
  - bioinformatics
  - saudi-arabia
  - PAVS-2026
  - edge-ai
short_description: "Saudi clinical variant interpreter | PAVS 2026"
---

# 🧬 Pan-Genomic Saudi Edge-VCF

**Universal, 100% offline, mobile-native clinical variant interpreter for the Saudi population.**

Built by **BABA AI Solutions** using Teacher-Student Distillation with the **PAVS 2026** (Phenotype-Associated Variants in Saudi Arabia) Framework.

## Features

- **76,141 parameters** (<100k budget) — runs on mobile/edge devices
- **FiLM-conditioned** architecture: same model, different phenotype contexts → different clinical decisions
- **PAVS 2026** Saudi-specific allele frequency knowledge graph
- **ACMG 5-tier** pathogenicity classification (Benign → Pathogenic)
- **HPO ontology** integration for symptom-to-gene mapping
- **Consanguinity-aware** analysis (Saudi population F-coefficient baselines)

## Architecture

```
NTv3_650M_post (Teacher, 650M) → TimeKD Distillation → Edge-RGF-Pan (Student, 76K)
                                                         ├── 1D-CNN (24ch)
                                                         ├── Bi-GRU (hidden=28)
                                                         ├── FiLM [HPO_Emb, Saudi_AF, Consanguinity_F]
                                                         └── 5-class ACMG Head
```

## Try It

1. Enter patient symptoms (e.g., "blood in urine, hearing loss")
2. Select consanguinity status
3. Click **⚡ Run Full Analysis**
4. Explore HPO mapping, PAVS variants, and clinical report tabs