Skip to content

Lukas AlphaFold Guide

A practical guide to protein structure prediction with AlphaFold2, ColabFold and AlphaFold3 — what they are, how to tell whether a prediction is trustworthy, what AlphaMissense adds on top, and how to do all of this hands-on in ChimeraX with ChopChopMF.

Disclaimer

This guide was compiled to the best of my knowledge, from the primary documentation and literature linked throughout (see Resources & Further Reading). It comes with no guarantee of accuracy or completeness — tools, thresholds, licenses and web-service limits (AlphaFold Server quotas in particular) change over time, so always cross-check anything decision-critical against the current official source before relying on it.

Found this useful? Cite it

If this guide helped your work, citing it helps others find it too:

Bauer, L. W. (2026). Lukas AlphaFold Guide: A practical guide to AlphaFold2,
ColabFold, AlphaFold3, AlphaMissense and ChopChopMF.
https://github.com/LUKASinScience

Missing something, spotted an error, or have an idea? For the guide itself, corrections and suggestions are welcome — see below once it has a public home. For the tools it covers, report issues or feature ideas directly where they're maintained: ChopChopMF issues · ChimeraX-FigureStyle issues.

Not sure where to start?

Answer three quick questions and I'll point you to the right page to start on.

1. Have you run AlphaFold, ColabFold, or the AlphaFold Server before?

2. What does pLDDT tell you?

3. Have you had to judge whether a predicted interface or complex is trustworthy?

Who this is for

Every page is written to be read start-to-finish by a beginner, while still holding useful depth for people who already run these tools daily. Content is layered with three consistent markers:

Beginner

Always visible. The core idea, explained without assuming prior knowledge — read at least this on every page.

Advanced

Also always visible, but assumes you've read the beginner section. Practical detail for people already using these tools.

Expert deep-dive (click to expand)

Collapsed by default. Technical/architectural detail, edge cases, and caveats — skip this on a first read, come back when you need it.

What's in this guide

Try It: A Real Example Protein
An interactive 3D viewer on a real AlphaFold DB entry — see pLDDT coloring and PAE on an actual structure, not a diagram.
AF2 vs. ColabFold vs. AF3
What actually changed between the three, and which one you should reach for.
Structure vs. Sequence Homology
Why Foldseek finds relatives BLAST can't — and how to infer function from an unknown fold.
Reading Confidence Metrics
pLDDT, PAE, pTM/ipTM — what the numbers mean and when to distrust them.
What Influences a Prediction
Why the MSA dominates and a single point mutation usually doesn't.
AlphaMissense
What it predicts, how it's built, and how not to over-interpret a score.
Running ColabFold (local/HPC)
GPU-accelerated MSA, apptainer, SLURM — and how to present results afterwards.
Visualization Tools
The landscape (ChimeraX, PyMOL, VMD, Mol*) and why this guide picks ChimeraX.
ChopChopMF
Turning the concepts above into point-and-click ChimeraX workflows.
Glossary
Every acronym used in this guide, one click from wherever you're reading.

This guide is local-only for now

This is currently a local Zensical build (zensical serve / zensical build), not yet published anywhere. Structure and content are still evolving — see Resources & Further Reading for the full list of primary sources this guide is built from.