ImmunoTwin-Sim v2.0: A Universal In-Silico Platform for Transcriptomic Drug Repurposing

ImmunoTwin-Sim-v2.0 (https://immunotwin-sim-v2-dashboard.streamlit.app/#immuno-twin-sim-v2-0) ImmunoTwin-Sim is a high-performance in-silico simulation tool designed to identify drug repurposing candidates for rare and complex diseases. By utilizing a Reverse-Transcriptomics approach, the platform compares a patient’s disease-specific gene expression profile against a global database of drug-induced perturbation signatures.The goal of this tool is to help researchers and clinicians identify existing FDA-approved drugs that can "flip" or "reverse" a disease's genetic signature back toward a healthy state. 🚀 Key Features- Universal Discovery Engine: Powered by the Enrichr L1000 database, providing access to thousands of drug signatures. - Fail-Safe Architecture: Combines a local high-speed knowledge base with a live API fallback for 100% stability. - Multi-Domain Application: Successfully tested across Autoimmune (Dermatomyositis), Oncology (Neuroblastoma), and Neurology (Huntington’s Disease). - Real-time Visualization: Interactive Plotly-based dashboards for calculating "Repurposing Potential" scores. - Fuzzy Search Technology: Intelligent drug name matching to handle database naming variations. 🛠️ The Concept: Signature Reversal- The core logic of ImmunoTwin-Sim follows a "Lock and Key" mechanism: - The Lock (Disease Profile): Upregulated genes driving a rare disease. - The Key (Drug Signature): Genes suppressed by a specific drug. - The Match: The tool calculates the overlap.If a drug's suppression signature targets the disease's driver genes, it has high repurposing potential. 📂 Installation & Usage1. PrerequisitesEnsure you have Python installed, then install the required libraries:pip install streamlit pandas requests plotly 2. Run the Applicationstreamlit run ImmunoTwin-Sim.py 3. How to Use- Sidebar: Select a drug from the dropdown or type a custom name (e.g., Baricitinib, Anifrolumab).Sidebar: Select or type the target disease (e.g., Dermatomyositis). - Step 1: Upload a CSV file of the disease profile. The file must have a column named gene_symbol. - Step 2: View the Repurposing Potential score and the specific molecular targets being reversed. 📊 Sample Test DataTo demonstrate universality, you can test the following pairs using data from NCBI GEO: | Disease | GEO Accession | Recommended Test Drug || --- | --- | --- || **Dermatomyositis** | GSE142807 | Tofacitinib / Baricitinib || **Neuroblastoma** | GSE120572 | Vincristine || **Huntington’s Disease** | GSE3790 | Memantine | 🧬 Technology Stack- Language: Python 3.x - Frontend: Streamlit - Data Analysis: Pandas, NumPy - Visualization: Plotly Express 🌍 Real-World Problem Solving & Applications 1. Accelerating Rare Disease Treatment Developing new drugs for rare diseases is often financially unfeasible for pharmaceutical companies due to small patient populations. ImmunoTwin-Sim solves this by identifying FDA-approved drugs that can be repurposed, significantly reducing the time and cost of bringing treatments to patients. 2. Personalized Precision Medicine Clinicians can upload a specific patient's transcriptomic profile (derived from RNA-Seq or Microarray data) to find the drug with the highest "Repurposing Potential" for that specific individual’s genetic makeup. 3. Oncology & Rare Autoimmunity The tool is uniquely suited for: Rare Cancers: Identifying compounds that suppress tumor-driver genes (e.g., MYCN in Neuroblastoma). Autoimmune Storms: Finding JAK-inhibitors or Interferon antagonists to settle "cytokine storms" in diseases like Dermatomyositis. ✅ Advantages Hybrid Stability: Unlike other tools that crash when APIs are busy, ImmunoTwin-Sim uses a fail-safe architecture combining a local master knowledge base with live API discovery. Multi-Domain Versatility: The logic is "Universal"—it works for neurology, oncology, and immunology without requiring code changes. Low Computational Barrier: Designed for accessibility; doctors and researchers can run complex bioinformatic simulations without writing a single line of code. Fuzzy Name Matching: Automatically handles different drug naming conventions, ensuring high search success rates. ⚠️ Limitations Transcriptomic Bias: The tool currently focuses on mRNA expression. It does not account for post-translational modifications or protein-level interactions. Directionality Assumption: The tool assumes that reversing an upregulated gene is always beneficial. In some biological contexts, certain gene activations may be compensatory/protective. Data Quality Dependence: The accuracy of the "Repurposing Potential" score is directly dependent on the quality of the user’s uploaded CSV file and the statistical significance of the input gene list. 🔮 Future Work 1. Integration of Side-Effect Profiling Future versions will include a safety-filter that flags drugs with high toxicity profiles for specific patient demographics, ensuring "Repurposing Potential" is balanced with "Patient Safety." 2. Support for Down-regulated Genes The next update will allow users to upload "Down-regulated" disease genes and match them with drugs that act as "Up-regulators," providing a complete 360-degree signature reversal. 3. AI-Driven Synergy Prediction Incorporating machine learning models to predict how drug combinations (polytherapy) might work together to reverse signatures that a single drug cannot fix alone.