Gene Explorer: Drug Repurposer App

1st Place – PoliHack v17 (AppDev Division)

---

Overview

Gene Explorer is a cross-platform application developed as part of PoliHack v17, under the AppDev division, where it proudly earned 1st place.

Its main objective is to support researchers, bioinformaticians, and healthcare professionals in exploring the complex relationships between human genes and pharmaceutical compounds.

The application is fully responsive and can be accessed seamlessly across desktop, tablet, and mobile devices, making it adaptable for both field work and research labs.

By integrating interactive data visualizations and AI-powered tools, the platform enables users to identify promising opportunities for drug repurposing — the process of discovering new therapeutic uses for existing medications.

---

Team Members

We are a team of five students passionate about technology, innovation, and biomedical research. Our group includes three 3rd-year students and two 2nd-year students from the Faculty of Computer Science.

Team Lead: Ruse Teodor – responsible for coordinating development, UI/UX direction, and integration between components.

Name	Year of Study
Balahura Vlad	2nd year – Computer Science
Moga Antonia - Teodora	2nd year – Computer Science
Ruse Teodor	3rd year – Computer Science
Rusu Dana	3rd year – Computer Science
Stan Ariana - Maria	3rd year – Computer Science

---

Key Features

• Gene Search – Enter any human gene to begin exploring its biological and pharmaceutical context.
• Drug Repurposing Engine – Discover existing drugs with high interaction potential based on gene-specific data.
• Interactive 3D Graph – Visualize complex gene–drug networks through a dynamic, node-based interface.
• BioGPT-powered Chatbot – Ask biomedical questions in natural language and receive real-time answers from an AI assistant.
• Gene Details Panel – Access structured data including nomenclature, functional role, genomic location, and known aliases.
• Drug Insights – View enriched information about suggested drugs, including usage context, interaction score, and scientific summaries.
• Cross-Platform Ready – Fully responsive and optimized for both desktop and mobile devices.

---

Application Architecture

root/
│
├── public/
│   ├── datasets
│   ├── video_prima_pg.mp4
│   ├── logo-shift-happens.png
│   ├── chatbot-icon.png
│   └── img_prima_pg.jpg
│
├── imgReadMe/                        # Images for README.md visual flow  
│
├── src/
│   ├── index.js                      # Entry point – renders App
│   ├── App.js                        # Main app logic & state
│   ├── App.test.js                   # Basic test setup
│
│   ├── DA/                           # Frontend components
│   │   ├── Page1.js                  # Landing page
│   │   ├── Page2.js                  # Graph + drug view
│   │   ├── Chatbot.js                # AI chatbot (BioGPT)
│   │   └── formatBioGPTResponse.js   # Chat output formatter
│
│   ├── TV/                           # Data & logic layer
│   │   ├── GeneInfo.ts               # Gene info (NCBI)
│   │   ├── GeneCode.ts               # KEGG gene codes
│   │   ├── GenetoDrugDetails.ts      # Drug interactions (DGIdb)
│   │   ├── GeneToNetwork.ts          # Gene to network (KEGG)
│   │   ├── GraphData.ts              # Graph builder (nodes/links)
│   │   ├── moreDrug.ts               # Drug info (FDA)
│   │   ├── MyGraph.js                # 3D graph component
│   │   └── NeworkNameMembers.ts      # Parse network relationships
│
│
├── proxy.js              # API proxy for backend requests
├── proxy_bigBoss.js      # Main backend proxy (KEGG endpoints)
├── package.json          # Project dependencies and scripts
└── README.md             # Project documentation and setup guide

Flow Vizual

Homepage	Gene & Graph View	Gene Details	Drug Details	Chatbot

---

AI Chatbot – Shifty

The application integrates an intelligent biomedical assistant called Shifty, powered by BioGPT. Users can interact in natural language to retrieve information about genes, conditions, and preventative measures.

Example Prompts:

• "How do I prevent breast cancer?"
• "What is the function of BRCA1?"
• "What drugs interact with gene SRC?"

---

Technologies Used

Layer	Stack / Tools
Frontend	React.js, CSS Modules, ForceGraph3D
Backend	Node.js, Express, custom proxy servers
AI Component	BioGPT (via local HTTP endpoint)
Data Sources	KEGG REST API, DGIdb GraphQL, FDA API
Build Tools	Vite, npm
Dev Tools	Visual Studio Code, Jest, React Testing Library

---

Setup & Local Installation

1. Clone the Repository

git clone https://github.com/AriiSM/PoliHack_V17.git
cd poli_hack

2. Install Dependencies

npm install --legacy-peer-deps

3. Run the Frontend

npm run dev

4. Install & Configure LM Studio (for Shifty AI features)

To enable AI functionality in Shifty, you need to run a local LLM API using LM Studio.

Steps:

1. Download LM Studio

2. Open the application and go to the Models tab.

3. In the search bar, find and download the following model: akhilanilkumar_-_biogpt-baseline-gguf/biogpt-baseline.Q4_K_S.gguf

4. Once the model is downloaded, navigate to the Server tab.

5. Select the downloaded model and click Start Server.

   Configuration Notes:

   • Make sure the API server runs on: http://localhost:1234
   • You can leave the default settings, unless you need a custom port.

   Shifty will now be able to connect and interact with the local model using this endpoint.

---

Demo Video

Watch a full demo of the Gene Explorer application in action: