Can AI Predict a Protein's Shape? AlphaFold Explained With a Real Enzyme

Proteins do the work of every cell, and their 3D shape decides what they can do. Can AI work out that shape from a sequence of amino acids, and how would we know it is right? Follow one real enzyme, adenylate kinase, from sequence to AlphaFold model to a lab experiment. Terms are explained as they appear, so students and curious viewers can follow without a biology degree. You will see how protein shapes are measured (X-ray crystallography, cryo-EM, NMR), how evolutionary clues and residue-pair information feed the network, how AlphaFold is trained, what its confidence scores mean, and how a prediction becomes a testable experiment, from a single enzyme to the human nuclear pore. AlphaFold's developers shared the 2024 Nobel Prize in Chemistry. This explanation follows Highly accurate protein structure prediction with AlphaFold by Jumper and colleagues (DeepMind and collaborators, 2021), the nuclear-pore work by Mosalaganti and colleagues (EMBL, MPI of Biophysics and collaborators, 2022), and Accurate structure prediction of biomolecular interactions with AlphaFold 3 by Abramson and colleagues (Google DeepMind, Isomorphic Labs and collaborators, 2024). Chapters 0:00 One enzyme, one lysine 2:46 How we know a protein's shape 5:10 Clues from related sequences 6:43 How AlphaFold builds a model 9:00 How the model learns 10:50 How far to trust a prediction 13:24 Turning a model into an experiment 14:54 From one enzyme to a whole cell References AlphaFold2 — Jumper et al. (2021) — https://www.nature.com/articles/s4158... AlphaFold3 — Abramson et al. (2024) — https://www.nature.com/articles/s4158... Adenylate kinase — UniProt P69441 — https://www.uniprot.org/uniprotkb/P69... Closed adenylate kinase — PDB 1AKE — https://www.rcsb.org/structure/1AKE Open adenylate kinase — PDB 4AKE — https://www.rcsb.org/structure/4AKE AlphaFold adenylate kinase model — https://alphafold.ebi.ac.uk/entry/P69441 AlphaFold calmodulin model — https://alphafold.ebi.ac.uk/entry/P0DP23 Confidence scores — EMBL-EBI — https://www.ebi.ac.uk/training/online... Structural methods — RCSB PDB-101 — https://pdb101.rcsb.org/learn/guide-t... The Nobel Prize in Chemistry 2024 — NobelPrize.org — https://www.nobelprize.org/prizes/che... AlphaFold predictions and experimentally determined structures — PDBe — https://www.ebi.ac.uk/pdbe/node/22820 ATP — RCSB Chemical Component Dictionary — https://www.rcsb.org/ligand/ATP AMP — RCSB Chemical Component Dictionary — https://www.rcsb.org/ligand/AMP ADP — RCSB Chemical Component Dictionary — https://www.rcsb.org/ligand/ADP Ubiquitin NMR structure — PDB 2MOR — https://www.rcsb.org/structure/2MOR Ubiquitin chemical shifts — BMRB 6457 — https://bmrb.io/data_library/summary/... AI-based structure prediction empowers integrative structural analysis of human nuclear pores — Mosalaganti et al. (2022, EMBL, MPI of Biophysics and collaborators) — https://doi.org/10.1126/science.abm9506 Human nuclear pore scaffold — PDB 7R5J — https://doi.org/10.2210/pdb7r5j/pdb Traditional Chinese subtitles are available. #AlphaFold #ProteinStructure #AIinBiology