biolists / biodiffusion

A list of manuscripts/tools using diffusion on biological enttieis

📖 Table of contents

• Structure design
• Sequence design
• Structure reconstruction
• Sequence reconstruction
• Binding prediction
• Binder generation
• Molecule generation

💡 Notes

• The following lists are curated by humans, as such may be incomplete
• We do not wish to advertize one tool over any other, but simply list the tools we are aware of in a random order
• Any suggestions for improvements and additions are welcome as issues or pull requests
• Projects we identify as discontinued are marked with 💀 and in a section at the end
• If you wish to include this list in a publication do so by using the link "https://github.com/biolists/biodiffusion"

⚡️ Brought to you by:

• @sacdallago
• @MoreheadAlex

---

Structure Design

Methods that from noise sample protein structures, e.g. backbones.

• RFDiffusion
  • Backbone diffusion
• Chroma
  • Backbone diffusion
• FoldingDiff
  • Torsion angle diffusion
• FrameDiff
  • Backbone diffusion
• Genie
  • Backbone diffusion
• Anand/Achim
  • Backbone or rotamer diffusion
• ProteinSGM
  • diffusion on Ca, rotamer, and other measures

---

Sequence Design

Methods that from noise sample sequences, e.g. residue unmasking.

• NOS
  • Masked residue inpaining
• EvoDiff
  • Masked residue inpaining, conditioned on MSA or single sequence
• ProteinGenerator
  • Sequence space diffusion
• GraDe-IF
  • Inverse protein folding: from structure to sequence
• Anand/Achim
  • Masked residue inpaining
• ProteinDT
  • Conditional (from pLM & text) protein sequence generation
• Frey/Prescient
  • MCMC single step denoising sequence
• PepPrCLIP
  • Forward noising ESM2 representations, CLIP based sequence reconstrubction for peptide sequences

---

Structure Reconstruction

Methods that, given an input in modality X, impute a structure output, e.g. from image domain to PDB structure.

• CryoDRGN
  • CryoEM structure reconstruction

---

Sequence Reconstruction

Methods that, given an input in modality X, impute a sequence output, e.g. from frequency domain to sequence.

• InstaNovo
  • Given proteomics readout, reconstruct peptide sequence

---

Binding prediction

Methods that operate on two molecules and predict an aspect of bidning, e.g. pose

• DiffDock
  • Small molecule binding pose
• DiffDock-PP
  • Protein-protein binding pose
• NeuralPLexer
  • Small molecule binding
• DynamicBind
  • Find binding pockets on protein surfaces
  • ⚠️ no code available @ GitHub repo

---

Binding generation

Methods that generate some entity (e.g. small mol) linking or binding an input entity/entities (e.g. linker generation)

• DiffLinker
  • Linker generation
• DiffSBDD
  • Small molecule generation given pocket
• DiffAB
  • Design antibody given antigen structure
• DiffBP
  • Small molecule generation given pocket

---

Molecule generation

Methods that generate some entity (e.g. small mol) unconditionally or conditionally to target specific molecular properties (e.g. 3D molecule generation)

• EDM
• Diffusion Prior Bridges
• LDM
• GCDM

---

Discontinued

• ProtDiff_SMCDiff
  • C-alpha backbone diffusion
  • According to GitHub, superseeded by FrameDiff and RFdiffusion

biodiffusion's Issues

To triage @Hannes

• Dynamicbind
• Gradeif graph denoising diffusion for inverse folding
• Discrete guided diffusion for protein design
• Diffbp https://arxiv.org/abs/2211.11214

Diffusion Models for Protein Conformational Change Modelling

Hi,

Thank you for your efforts in compiling this resource! I just wanted to add our recent work on diffusion bridges, with applications both in cellular differentiation processes and protein conformational changes (from apo to holo structures)

Link: https://arxiv.org/abs/2302.11419
Code: https://github.com/vsomnath/aligned_diffusion_bridges

Antibody diffusion models

Antigen-Specific Antibody Design and Optimization with Diffusion-Based Generative Models for Protein Structures
https://openreview.net/forum?id=jSorGn2Tjg

Generative Diffusion Models for Antibody Design, Docking, and Optimization
https://doi.org/10.1101/2023.09.25.559190

To triage @Anthony

• Protein Structure and Sequence Generation... https://arxiv.org/abs/2205.15019
• Protein structure generation via folding diffusion https://arxiv.org/abs/2209.15611
• Score-based generative modeling... https://doi.org/10.1038/s43588-023-00440-3
• A Text-guided Protein Design Framework https://arxiv.org/abs/2302.04611

To triage @Nathan

https://arxiv.org/abs/2306.12360

To triage @Pranam

https://www.biorxiv.org/content/10.1101/2023.06.26.546591v1