Skip to content

xiet02/TADA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

82 Commits
 
 
 
 
 
 

Repository files navigation

TADA: Target and Antibody Discovery via AI

TADA is an AI-powered pipeline for accelerating ADC (Antibody-Drug Conjugate) and TCE (T-cell Engager) drug development, covering both target discovery and antibody engineering.


Overview

TADA integrates multi-source data and AI to streamline the discovery and optimization of therapeutic antibodies. The repository is organized into two parts:

  • Part I: Antigen target identification and prioritization.
  • Part II: In silico antibody design and developability optimization.

Part I: TAA Discovery System

Overview

The TAA Discovery System is an innovative framework for identifying and prioritizing tumor-associated antigens for TCE or ADC development. It integrates diverse datasets—including multi-omics repositories and scientific publications—using a graph retrieval-augmented generation (RAG)-enhanced language model to extract insights from biological/clinical literature and curated oncology-related omics databases (TCGA, GTEx, single-cell atlases, etc.).

Features

  • Graph RAG-enhanced Language Model: Efficiently extracts relevant information from scientific literature.
  • Omics Database Integration: Combines data from TCGA, GTEx, and other sources for comprehensive analysis.
  • Safety Score: Prioritizes TAAs with high tumor selectivity and low on-target/off-tumor risk.

Reference

For methodology and findings, see: Xie & Huang,bioRxiv 2025


Part II: In Silico Antibody Design

Overview

After target nomination, antibody engineering remains a major bottleneck. TADA provides an integrated, AI-driven framework for in silico antibody design and multi-objective optimization, combining sequence diversification, structure prediction, docking confidence assessment, and developability scoring.

Features

  • Sequence Diversification: Generates CDR-focused antibody variants.
  • Structure Prediction: Uses AlphaFold-Multimer and AlphaFold3 for accurate modeling.
  • Developability Scoring: Evaluates solubility, CDR liability risks, and interface confidence.
  • Unbiased Ranking: Produces a unified score for candidate prioritization.

Reference

For methodology and findings, see: Xie, bioRxiv 2026


Reference & Contact

For questions or collaboration, contact: txie@neoomics.com

About

Target and Antibody Discovery via AI

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages