v1.0.3: The Four-Step Federation Debate Loop

🚀 The Ultimate Autonomous Research Loop (v1.0.3)

In this update, we have perfected the autonomous reasoning core of GeneX-AI. The system now features a highly rigorous, closed-loop Four-Step Multi-Agent Federation Debate.

When tackling complex biomedical problems (e.g., curing Glioblastoma or designing extremophiles), the AI will autonomously orchestrate the following pipeline without human intervention:

1. 🏗️ The Architect (Blueprint Design): Queries cloud databases (OpenTargets/UniProt) to propose a macroscopic biological or pharmaceutical blueprint.
2. 👨‍🔬 The Theoretical Physicist (First-Principles Check): Employs pure physical tools (Gibbs Free Energy, Brownian diffusion, Lorentz force) to forcefully find thermodynamic or kinetic dead-ends in the Architect's design.
3. ⚖️ The Review Committee (Clinical Risk): Evaluates clinical and ecological risks by running Virtual Clinical Trials (PK/PD) and checking for evolutionary tumor escape routes.
4. 🧪 The Experimentalist (Wet-Lab Execution): Once consensus is reached, this agent translates the theoretical solution into reality, generating wet-lab execution scripts (e.g., Opentrons robotic protocols) and structural formulas.

(Note: The dynamic ontology tree and all first-principles calculators are fully integrated into this new debate loop. The core gene_knowledge_full_20k.db database remains unchanged. New users should download it from the v1.0.0 release assets.)

v1.0.2: Dynamic Cloud Ontology & Hardcore Federation Debate

Redefining General-Purpose Biomedical AI (v1.0.2)

GeneX-AI is built to be a comprehensive, end-to-end computational biology research system—covering everything from molecular target discovery and drug design to virtual clinical trials and ecological modeling. In v1.0.2, we introduce two massive architectural upgrades to solidify this vision:

• 🌐 Dynamic Medical Ontology (Infinite Scalability)
  A biomedical AI is only as good as its knowledge base. GeneX-AI is no longer limited by a hardcoded local database. When tasked with researching a rare disease or novel pathway, the AI now autonomously queries the EBI OLS (Ontology Lookup Service) cloud. It dynamically maps disease ontologies (e.g., MONDO), fetches real gene targets via UniProt, and builds its own local SQLite cache in real-time. The system's medical knowledge is now infinite and self-expanding.

• 🏛️ Upgraded Multi-Agent Federation Debate
  The debate mode has been overhauled to simulate a rigorous, multi-disciplinary pharmaceutical review board:

  • The Theoretical Physicist strictly evaluates the thermodynamic and kinetic viability of biological pathways.
  • The Synthetic Biologist engineers workarounds (e.g., covalent inhibitors or XNA) to bypass physical dead-ends.
  • The Review Committee acts as the final gatekeeper, executing Virtual Clinical Trials (Population PK/PD) and assessing tumor evolutionary escape risks before approving any drug or genetic modification.

(Note: The gene_knowledge_full_20k.db database remains unchanged from v1.0.0. If you are a new user, please download it from the v1.0.0 release assets.)

GeneX AI v1.0.1: First-Principles Physics & Macro Engines

Core Architecture Overhaul

• First-Principles Engine: Transitioned core biological simulations from empirical formulas to strict physical and chemical laws (Thermodynamics, Arrhenius kinetics, Energy Balance Models).

New Built-in AI Capabilities

• 5 Frontier Macro-Engines:

  1. Virtual Clinical Trials (Population PK/PD simulation)
  2. Epigenetic Reprogramming Simulator (Waddington landscape thermodynamics)
  3. Pangenome & Horizontal Gene Transfer Dynamics
  4. Neuro-Genomic Topology Model
  5. Astrobiology Panspermia Engine (Radiation survival in deep space)

• Advanced Physics Calculators: Added raw physics tools for the built-in AI: Gibbs Free Energy, Quantum Tunneling, Brownian Diffusion, Lorentz Force, Bragg Diffraction, and Capillary Action.

(Note: The gene_knowledge_full_20k.db database remains unchanged from v1.0.0. If you are a new user, please download it from the v1.0.0 release assets.)

GeneX AI v1.0.0 Released

Key Capabilities:

• Ontology-Guided Lazy Loading: Navigates AI queries via local GO/MONDO topology trees, dynamically fetching and caching authentic genomic data from Ensembl/UniProt.
• Deterministic Biological Analysis: Provides reproducible, mathematically constrained proxies for CRISPR off-target scoring and HLA immunogenicity.
• Multi-Agent Protocol: Implements an Architect -> Reviewer -> Experimentalist pipeline to ensure scientific rigor in de novo designs.
• Physics & Complex Systems Solvers: Integrates scipy.integrate.solve_ivp (LSODA) for geological-timescale Lotka-Volterra ecological ODEs, and thermodynamic $\Delta G$ evaluators.

Installation:

# Windows
install.bat
# Mac/Linux
bash install.sh

License:
Licensed under AGPL-3.0. For commercial deployment (closed-source cloud/on-premise), please contact 2141595982@qq.com.