BoltzGen: Toward Universal Binder Design

Codes for our paper "Full-Atom Peptide Design with Geometric Latent Diffusion" (NeurIPS 2024)

Codes for our paper "UniMoMo: Unified Generative Modeling of 3D Molecules for De Novo Binder Design" (ICML 2025)

Codes for our paper "Programming Biomolecular Interactions with All-Atom Generative Model"

Archived research prototype for immunogenic peptide design and docking.

De novo therapeutic peptide design pipeline using RFdiffusion and ESMFold to hallucinate NaV1.7 pain-channel blockers locally on CPU

A modular, extensible peptide design pipeline with target preparation, backbone generation, sequence design, scoring, and ranking. Full local CPU pipeline, and backend hooks for RFpeptides, ProteinMPNN/LigandMPNN, and ColabFold.

Source code for Targeting SARS-CoV-2 Receptor Binding Domain and Main Protease with D-peptides

The public repository for the CyclicCAE conformational autoencoder. CyclicCAE encodes peptide macrocycle conformations in a low-dimensional latent space, useful for rapid sampling, structural refinement, or denoising.

Evaluation other methods for macrocyclic design

AMP Forge is an antimicrobial peptide generation project based on PLM embeddings, VAE, and latent diffusion.

A Claude Code Skill for de novo D-peptide inhibitor design using the Boltz2 IC₅₀ prediction pipeline

AI-Driven De Novo Peptide Generator. Designing novel, highly diverse amino acid sequences for targeted biomedical applications.

AI pipeline that designs novel antimicrobial peptides and screens them for activity, potency, safety, and novelty into a ranked shortlist. In-silico, free compute, honest about uncertainty — predictions, not proof.

Multi-phase peptide ML pipeline for hemolysis prediction, generative design, and closed-loop optimization, built on transformers and modern PyTorch tooling.

AI-grounded peptide research reference — a curated catalog + Claude agent that cites PubChem, ClinicalTrials.gov, PubMed & UniProt, with a remote MCP server and offline bench calculators. Research reference, not a store.

Universal Peptide Drug Discovery — AI-driven cyclic peptide design pipeline with ncAA integration

A design framework for programmable genome engineering. Useful for designing peptide linkers that connect Transcription Activator-Like Effector (TALE) DNA-binding arrays to catalytic effector domains.