Skip to content

TODOLIST.md

Latest commit

 

History

History
228 lines (183 loc) · 7.83 KB

TODOLIST.md

File metadata and controls

228 lines (183 loc) · 7.83 KB
Raw

Project TODO List

✅ Completed

Algorithm Notebooks (85+ Complete)

Model Notebooks Status
AlphaFold2 32/32 algorithms ✅ Complete
AlphaFold3 23/23 algorithms ✅ Complete
Boltz-1 20/20 algorithms ✅ Complete
Boltz-2 10/10 algorithms ✅ Complete

Reference Papers (229 Curated)

Model Papers Status
AlphaFold2 83 papers ✅ AF2REFPAPERS.md
AlphaFold3 50 papers ✅ AF3REFPAPERS.md
Boltz-1 46 papers ✅ BOLTZREFPAPERS.md
Boltz-2 50 papers ✅ BOLTZ2REFPAPERS.md

Project Organization

  • Reorganize directory structure (lowercase, consistent naming)
  • Create comprehensive README.md
  • Update .gitmodules for new paths
  • Create BLOG_POST.md
  • Add finetuning framework

🔄 In Progress

Fine-tuning Framework Enhancement

  • Add pre-trained LoRA weights for common tasks
  • Create Colab notebooks for cloud training
  • Build benchmark datasets for each task type
  • Implement model zoo with pre-fine-tuned models

Documentation

  • Add Chinese documentation
  • Create video tutorials
  • Write detailed API documentation for finetuning module

📋 Planned

New Model Support

  • ESMFold notebooks (protein language model approach)
  • Chai-1 notebooks (next-gen competitor)
  • OpenFold notebooks (PyTorch AlphaFold2)
  • RoseTTAFold notebooks

Advanced Features

  • Interactive 3D visualization in notebooks
  • Streamlit/Gradio demo apps
  • Docker containers for easy deployment
  • CI/CD pipeline for notebook testing

Fine-tuning Extensions

  • Domain-specific adapters (antibodies, enzymes, GPCRs)
  • Multi-task learning examples
  • Active learning integration
  • Uncertainty quantification

📊 AlphaFold2 Source Code Review

Location: alphafold2/source/

common/

  • confidence.py
  • protein.py
  • protein_test.py
  • residue_constants.py
  • residue_constants_test.py

data/

  • mmcif_parsing.py
  • parsers.py
  • pipeline.py
  • templates.py

data/tools/

  • hhblits.py
  • hhsearch.py
  • hmmbuild.py
  • hmmsearch.py
  • jackhmmer.py
  • kalign.py
  • utils.py

model/

  • all_atom.py
  • all_atom_test.py
  • common_modules.py
  • config.py
  • data.py
  • features.py
  • folding.py
  • layer_stack.py
  • layer_stack_test.py
  • lddt.py
  • lddt_test.py
  • mapping.py
  • model.py
  • modules.py
  • prng.py
  • prng_test.py
  • quat_affine.py
  • quat_affine_test.py
  • r3.py
  • utils.py

model/tf/

  • data_transforms.py
  • input_pipeline.py
  • protein_features.py
  • protein_features_test.py
  • proteins_dataset.py
  • shape_helpers.py
  • shape_helpers_test.py
  • shape_placeholders.py
  • utils.py

relax/

  • amber_minimize.py
  • amber_minimize_test.py
  • cleanup.py
  • cleanup_test.py
  • relax.py
  • relax_test.py
  • utils.py
  • utils_test.py

✅ AlphaFold2 Algorithm Notebooks (32/32 Complete)

# Algorithm Notebook Status
1 MSABlockDeletion algorithm-1 ✅
2 Inference algorithm-2 ✅
3 InputEmbedder algorithm-3 ✅
4 relpos algorithm-4 ✅
5 one_hot algorithm-5 ✅
6 EvoformerStack algorithm-6 ✅
7 MSARowAttentionWithPairBias algorithm-7 ✅
8 MSAColumnAttention algorithm-8 ✅
9 MSATransition algorithm-9 ✅
10 OuterProductMean algorithm-10 ✅
11 TriangleMultiplicationOutgoing algorithm-11 ✅
12 TriangleMultiplicationIncoming algorithm-12 ✅
13 TriangleAttentionStartingNode algorithm-13 ✅
14 TriangleAttentionEndingNode algorithm-14 ✅
15 PairTransition algorithm-15 ✅
16 TemplatePairStack algorithm-16 ✅
17 TemplatePointwiseAttention algorithm-17 ✅
18 ExtraMsaStack algorithm-18 ✅
19 MSAColumnGlobalAttention algorithm-19 ✅
20 StructureModule algorithm-20 ✅
21 rigidFrom3Points algorithm-21 ✅
22 InvariantPointAttention algorithm-22 ✅
23 BackboneUpdate algorithm-23 ✅
24 computeAllAtomCoordinates algorithm-24 ✅
25 makeRotX algorithm-25 ✅
26 renameSymmetricGroundTruthAtoms algorithm-26 ✅
27 torsionAngleLoss algorithm-27 ✅
28 computeFAPE algorithm-28 ✅
29 predictPerResidueLDDT algorithm-29 ✅
30 RecyclingInference algorithm-30 ✅
31 RecyclingTraining algorithm-31 ✅
32 RecyclingEmbedder algorithm-32 ✅

✅ AlphaFold3 Algorithm Notebooks (23/23 Complete)

# Algorithm Status
1-4 Input Preparation (MSA, Template, Atom, RelPos) ✅
5-7 MSA Module (OPM, Attention, Transition) ✅
8-14 Pairformer (Triangle Ops, Single Attn) ✅
15-19 Diffusion (Module, AdaLN, Transformer, CrossAttn) ✅
20-23 Confidence & Loss (Distogram, Confidence, Loss, LDDT) ✅

✅ Boltz Algorithm Notebooks (30/30 Complete)

Boltz-1 (20 algorithms)

# Category Status
1-3 Input Processing ✅
4-6 MSA Processing ✅
7-11 Pairformer Stack ✅
12-15 Diffusion Module ✅
16-20 Confidence & Loss ✅

Boltz-2 (10 new algorithms)

# Category Status
1-4 Affinity Module (NEW) ✅
5-10 Enhanced v2 Modules ✅

📈 Project Statistics

Metric Count
Total Notebooks 85+
Reference Papers 229
Fine-tuning Task Types 50+
Git Submodules 14
Source Code Files 60+

Last updated: January 2026