Resources & Further Reading¶

The primary sources this guide is built from, plus additional reading for anyone who wants to go deeper than a single guide can.

Official documentation & courses¶

• EBI Training: AlphaFold course — the most comprehensive free course covering AF2, AF-Multimer, AF3, and AlphaFold Server, including confidence-metric modules referenced throughout this guide.
• AlphaFold Server guides — official usage guidance for the AF3 web portal.
• google-deepmind/alphafold3 — docs/output.md — technical output format and confidence-head details.
• google-deepmind/alphafold — technical note v2.3.0 — AlphaFold-Multimer training-cutoff/parameter update details.
• ChopChopMF usage docs — the ChimeraX plug-in covered in this guide's ChopChopMF section.

Mechanism deep-dives¶

• The Illustrated AlphaFold — Elana Simon's detailed, diagram-heavy walkthrough of AlphaFold3's architecture (Pairformer, diffusion module, confidence heads); the single best source for AF3 internals.
• AlphaFold 2 is here: what's behind the structure prediction miracle — Oxford Protein Informatics Group's clear explainer of the Evoformer, IPA, and training losses.
• Deep Learning for Biology and AlphaFold — AI Summer's treatment of the attention mechanisms underlying AF2.

Confidence metrics & interpretation¶

• EBI: pLDDT — Understanding Local Confidence
• EBI: What AlphaFold 3 struggles with — spurious/hallucinated helices in disordered regions.
• AlphaFold2: A Role for Disordered Protein/Region Prediction? (PMC9104326) — pLDDT vs. conformational ensembles, transient helices.
• Systematic identification of conditionally folded intrinsically disordered regions by AlphaFold2 (PMC10622901) — conditional/binding-induced folding and deceptively high pLDDT.
• PAE Viewer — standalone tool for full-complex PAE heatmaps.

AlphaMissense¶

• Cheng et al., "Accurate proteome-wide missense variant effect prediction with AlphaMissense," Science 381, 2023 (DOI 10.1126/science.adg7492).
• AlphaMissense functional-assay correlation benchmark (PMC10634779) — comparison against REVEL, PrimateAI-3D and deep mutational scanning data.
• google-deepmind/alphamissense — reference code and data-access documentation.

Structure homology & Foldseek¶

• van Kempen et al., "Fast and accurate protein structure search with Foldseek", Nature Biotechnology 2023 — the 3Di structural-alphabet search method behind ChopChopMF's Foldseek tool.
• Foldseek GitHub repository
• EBI: AlphaFold DB + Foldseek integration announcement (Sept 2024)
• EBI training: Using the AlphaFold database for analysis
• Neurosnap: Foldseek Structural Clustering
• 310.ai: Hide and seek — structure similarities with Foldseek
• PAE Viewer paper — background on interface/PAE-based confidence for multimer predictions.

Running AlphaFold/ColabFold¶

• Boost AlphaFold2 Protein Structure Prediction with GPU-Accelerated MMseqs2 (NVIDIA)
• GPU-accelerated MSA preprint
• ColabFold: making protein folding accessible to all, Mirdita et al., Nature Methods 2022.
• Highly accurate protein structure prediction with AlphaFold, Jumper et al., Nature 2021.
• Protein complex prediction with AlphaFold-Multimer, Evans et al.
• From interaction networks to interfaces, scanning intrinsically disordered regions using AlphaFold2
• University of Chicago RCC: AlphaFold
• NSC Berzelius: AlphaFold
• Yale YCRC: AlphaFold
• ASBC: Best practices for presenting AlphaFold models
• ASBC: AlphaFold2 how-to guide
• Terwilliger et al., "AlphaFold predictions are valuable hypotheses, and accelerate but do not replace experimental structure determination", Nature Methods 2023/2024 — concrete evidence that confidently-predicted regions can still diverge from experimental structures.
• Sun et al., "AlphaFold2 for Protein Structure Prediction: Best Practices and Critical Analyses" (2024)

Beginner-friendly explainers¶

• How AlphaFold2 Revolutionized the Way We Do Biology (Pt 1)
• How to Use AlphaFold2 as a Wet Lab Biologist (Pt 2) / Pt 3
• thenode.biologists.com — AlphaFold protein interaction modeling tutorial — pointer to David Fay's tutorial/workshop video on modeling protein interactions and reading AlphaFold's confidence output.

Tooling used to build this guide¶

• Zensical — the static site generator this guide is built with (zensical serve / zensical build).