The Archive

Blitz distillation reduces AlphaFold3 and Boltz-2 denoising from 200 to 8 steps via geometric noise schedule and multistep consistency training.

Kohli (DeepMind) & Candido (Biohub) discuss limitations of AlphaFold and scaling approach to protein folding beyond structure prediction.

TurboPairFormer: optimized triangular attention kernel for AlphaFold3-style protein folding on NVIDIA Hopper with improved numerical stability.

IDiom protein language model trained on 54M intrinsically disordered regions from AlphaFold DB to enable functional design of unstructured protein domains.

Google's AlphaFold can help ID what parts of a gene editing protein enable mistakes.

DBMol leverages AlphaFold-3 and Boltz-2 structure prediction models for de novo small molecule drug design.

Interpretability study reveals AlphaFold2's learned protein conformational landscapes via Evoformer weight analysis and Gaussian smoothing.

When AlphaFold2 revolutionized drug discovery in 2020, its success relied entirely on the roughly 170,000 protein structures collected by scientists since 1971... When AlphaFold2 revolutionized drug discovery in 2020, its success relied entirely on the roughly 170,000 protein structures collected by scientists since 1971 and preserved in the Protein Data Bank. Measured data is the backbone for all AI models and workflows that process data as it’s created, act on what matters in real time, and analyzes data for deep insights. With the current rise of modern… Source

Sparse autoencoders (SAEs) decompose neural network activations into interpretable features, but many learned features never activate, a problem called feature death that wastes dictionary capacity and can reintroduce superposition. Death rates vary dramatically between models: near-zero on GPT-2, over 70% on AlphaFold3 with identical configurations. We find that dimension-level activation outliers (dimensions whose mean magnitude is large relative to per-token variation) cause this by shifting pre-activations at initialization based on each feature's alignment with the activation mean. Featu...

IBM Research releases MAMMAL, multimodal model integrating proteins/molecules/genes, achieves SOTA on 9/11 biological benchmarks including drug-target interaction and antibody-antigen binding.

ConforNets method targets latent perturbations in AlphaFold3 to reliably generate alternate protein conformations beyond single dominant structure.

Scientists are using AlphaFold to strengthen a photosynthesis enzyme for resilient, heat-tolerant crops.

Explore how AlphaFold has accelerated science and fueled a global wave of biological discovery.

AlphaFold has revealed the structure of a key protein behind heart disease

Berkeley BAIR's PLAID model generates proteins (sequence + 3D structure) via latent diffusion in AlphaFold2 latent space, enabling billion-scale training.