When a protein picture misses the dance

Introduction

AlphaFold helped transform structural biology by making protein structure prediction dramatically faster and more accessible. For many proteins, researchers can now obtain a useful three-dimensional model in minutes rather than spending months or years solving a structure experimentally. That is a major scientific advance and one reason AlphaFold is often cited as evidence that AI could accelerate discovery across medicine and biology.[Nature]nature.comHighly accurate protein structure prediction with AlphaFoldby J Jumper · 2021 · Cited by 50175 — The AlphaFold network directly pre…

But a protein structure is not the same thing as a protein’s behaviour. Living proteins are not rigid sculptures. They bend, twist, open, close, bind to other molecules, change shape in response to their environment, and sometimes exist in several different forms at once. A single predicted structure can therefore be highly accurate and still miss some of the most important biology. Understanding that gap is essential for understanding both the promise and the limits of AI-driven protein prediction.[Science News]sciencenews.orgalphafold ai protein structure folding predictionA static “structure doesn't tell you everything about how a protein works…Read more…[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Why proteins change shape inside cells

A common mental picture of proteins is that they resemble tiny molecular machines with fixed shapes. In reality, many proteins behave more like moving mechanical systems whose function depends on motion.

Cells are crowded, chemically active environments. Proteins are constantly being jostled by surrounding molecules and water. Their atoms vibrate, their loops flex, and entire domains can swing between different positions. These movements are not merely noise. Often they are central to what the protein actually does.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Enzymes provide a simple example. Many enzymes alter their shape as they bind a target molecule and carry out a chemical reaction. A structure showing only the “before” state may not reveal how the reaction proceeds. Likewise, transport proteins can shift between open and closed forms as they move substances across membranes. Regulatory proteins may switch conformations when they receive a chemical signal.[Science News]sciencenews.orgalphafold ai protein structure folding predictionA static “structure doesn't tell you everything about how a protein works…Read more…

Biologists increasingly describe proteins not as single structures but as conformational ensembles: collections of related shapes that exist with different probabilities. Function often depends on movement between these states rather than on any one state alone.[ScienceDirect]sciencedirect.comstatic structure prediction achieved by AlphaFold…[OUP Academic]academic.oup.comOUP AcademicBeyond static structures: protein dynamic conformations…by X Cui · 2025 · Cited by 34 — This review outlines the fundament…

This matters because disease can emerge when those motions change. A mutation might not destroy a protein’s overall structure yet still alter how easily it shifts between functional states. Looking only at a static snapshot can hide that effect.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Where static predictions help and fail

AlphaFold’s success comes partly from its ability to infer a highly plausible structure from patterns learned across enormous numbers of experimentally determined proteins. For many research questions, that is exactly what scientists need. A reliable structural model can reveal active sites, suggest protein functions, guide laboratory experiments, and help researchers identify promising drug targets. Nature AlphaFold The limitation is that most protein structure databases contain snapshots rather than movies. As a result[pmc.ncbi.nlm.nih.gov]pmc.ncbi.nlm.nih.govAlphaFold's static predictions do not reflect the reality of conformational landscapes…, AlphaFold and related systems are generally optimised to predict a likely structure, not a complete map of a protein’s motions through time. ScienceDirect[EMBL-EBI]ebi.ac.ukWhat AlphaFold 3 struggles withA key limitation of protein structure prediction models is that they typically predict static structures a…

Several kinds of biology are particularly difficult to capture with a single model:

• Proteins with multiple functional states. Some proteins deliberately switch between distinct shapes during their activity cycle. A prediction may capture one state while missing others.[Cell]cell.comRecognizing this limitation, several groups have now…Read more…
• Ligand-dependent conformations. A protein can adopt one shape when unbound and another when attached to a drug, metabolite, or signalling molecule.[EMBL-EBI]ebi.ac.ukWhat AlphaFold 3 struggles withA key limitation of protein structure prediction models is that they typically predict static structures a…
• Intrinsically disordered proteins. These proteins do not settle into one stable structure and instead exist as shifting ensembles. They are important in signalling, regulation, neurodegeneration, and cancer biology.[ScienceDirect]sciencedirect.comstatic structure prediction achieved by AlphaFold…[Nature]nature.comAlphaFold predictions are valuable hypotheses and…by TC Terwilliger · 2024 · Cited by 408 — Artificial intelligence-based protei…
• Allosteric regulation. A molecule binding at one site can trigger structural changes elsewhere in the protein. The critical biology may be the transition itself rather than either endpoint.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

The problem is not that AlphaFold’s predictions are wrong. Often they are remarkably accurate. The problem is that biology frequently depends on states that are absent from the prediction or on the movement between states.[Nature]nature.comAlphaFold prediction of structural ensembles of disordered…by ZF Brotzakis · 2025 · Cited by 125 — We introduce the AlphaFold-Me…

One example highlighted in EMBL-EBI’s AlphaFold training material is hexokinase, an enzyme that adopts different conformations depending on whether it has bound sugar. A prediction matching one state may not reveal the structural shift associated with activity.[EMBL-EBI]ebi.ac.ukWhat AlphaFold 3 struggles withA key limitation of protein structure prediction models is that they typically predict static structures a…

Why drug discovery often needs the missing motions

The importance of protein motion becomes especially clear in drug development.

Many medicines work by fitting into a protein’s binding pocket. Yet binding pockets are not always permanently visible. Some appear only briefly when a protein flexes into a particular conformation. Others change size or shape depending on what molecules are nearby. A static model may miss these transient opportunities entirely.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Drug resistance can also involve dynamics. Two proteins might appear nearly identical in a static comparison while behaving differently because one samples alternative conformations more frequently. That difference can alter how well a medicine binds.[ScienceDirect]sciencedirect.comstatic structure prediction achieved by AlphaFold…

This is one reason why structure prediction alone does not automatically translate into rapid cures. Knowing a protein’s shape is extremely useful, but turning that knowledge into a safe and effective therapy still requires understanding how the protein behaves in realistic biological conditions.[Nature]nature.comAccurate structure prediction of biomolecular interactions…by J Abramson · 2024 · Cited by 14654 — Here we describe our AlphaFol…

For advocates of AI-enabled scientific acceleration, this distinction is important. AlphaFold removed a major bottleneck, but it did not remove the need to study living molecular systems. Much of the remaining challenge lies in understanding dynamics, interactions, and biological context.[Nature]nature.comModeling protein conformational ensembles by guiding…by T Wu · 2025 · Cited by 16 — We find that the intrinsic performance of AlphaFol…

How experiments reveal motion that models miss

The continuing importance of experimental biology becomes clear when researchers try to observe molecular motion directly.

NMR can watch proteins in solution

Nuclear magnetic resonance spectroscopy, usually called NMR, is especially valuable for studying protein dynamics. Unlike techniques that often capture a single structure, NMR can provide information about motions occurring across a range of timescales while proteins remain in solution.[Frontiers]frontiersin.orgFrontiersAlphaFold 2 and NMR Spectroscopy: Partners to…by DV Laurents · 2022 · Cited by 107 — This experimental approach provides info…

Researchers use NMR to investigate flexible regions, transient interactions, folding processes, and disordered proteins that may be difficult to describe with a single structure. This makes it highly complementary to AlphaFold rather than a competitor to it.[Frontiers]frontiersin.orgFrontiersAlphaFold 2 and NMR Spectroscopy: Partners to…by DV Laurents · 2022 · Cited by 107 — This experimental approach provides info…

Cryo-EM increasingly captures multiple states

Cryo-electron microscopy, or cryo-EM, has become another powerful tool for observing conformational diversity. Modern cryo-EM workflows can separate different structural states from large experimental datasets, allowing researchers to reconstruct multiple configurations of the same molecular complex.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Instead of revealing only one structure, cryo-EM can sometimes expose the sequence of states involved in a molecular process. Researchers effectively gain a set of snapshots that can be assembled into a partial picture of motion.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

Molecular dynamics simulations create approximate movies

Another approach is molecular dynamics simulation. These simulations use physics-based models to estimate how atoms move over time.

Researchers increasingly combine AlphaFold predictions with molecular dynamics, using the predicted structure as a starting point and then simulating possible motions. This can reveal alternative conformations, transient interactions, and flexible regions that a static prediction does not show.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig… Nature The catch is that molecular dynamics is computationally expensive and remains an approximation. Simulating biologically meaningful timescales[nature.com]nature.comHighly accurate protein structure prediction with AlphaFoldby J Jumper · 2021 · Cited by 50175 — The AlphaFold network directly pre… for large proteins is still challenging.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

The next frontier is predicting protein ensembles, not just structures

A growing area of research aims to move beyond predicting one structure toward predicting entire conformational landscapes. Instead of asking, “What shape does this protein have?”, researchers increasingly ask, “What collection of shapes does this protein explore, and how often?”[ScienceDirect]sciencedirect.comstatic structure prediction achieved by AlphaFold…

Recent work combines AlphaFold with experimental measurements, molecular dynamics simulations, diffusion models, and ensemble-based approaches designed to generate multiple plausible states rather than a single answer. Researchers are developing systems that can model disordered proteins, alternative conformations, and dynamic trajectories more directly.[Nature]nature.comConformational ensembles for protein structure predictionby J Yang · 2025 · Cited by 9 — The results demonstrated an effective algorithm…[Nature]nature.comProc. Natl. Acad. Sci. USA 121…Read more…

This shift reflects a broader lesson about scientific AI. Solving one bottleneck often reveals the next one. Once researchers gained rapid access to protein structures, attention naturally moved toward interactions, dynamics, cellular context, and biological function.[PMC]pmc.ncbi.nlm.nih.govPMCAlpha Fold and protein folding: Not dead yet!The frontier is…by GR Bowman · 2024 · Cited by 31 — AlphaFold is revolutionizing structural biology by providing routine access to hig…

In the context of AI bloom, that is not a failure story. It is a reminder of how scientific progress usually works. Powerful AI systems can remove major obstacles and expand what researchers can investigate, but living systems remain extraordinarily complex. AlphaFold showed that AI could compress years of structural work into minutes. The challenge now is to understand not only what molecular machines look like, but how they move, adapt, and behave inside living cells.[Nature]nature.comA comprehensive application of FiveFold for conformation…by SK Niazi · 2025 · Cited by 3 — Building upon the FiveFold methodology for…[Nature]nature.comprotein binder design and conformational state predictionby LA Abriata · 2026 — A recent review describes several systems developed recen…

Amazon book picks

Further Reading

Books and field guides related to When a protein picture misses the dance. Use these as the next step if you want deeper reading beyond the article.

Book

Life's Edge

By Carl Zimmer

Explains why biology is dynamic and context-dependent, matching the page's static-versus-moving-protein theme.

See on Amazon

Book

The Song of the Cell

By Siddhartha Mukherjee

Places molecular behaviour inside cellular systems where motion and interaction matter.

See on Amazon

Book

The Code Breaker

By Walter Isaacson

Shows how molecular tools accelerate discovery while still requiring experiments and interpretation.

See on Amazon

Book

Transformer

By Nick Lane

Helps explain why living molecular systems are active processes rather than static diagrams.

See on Amazon

Browse more on Amazon:Life's EdgeThe Song of the CellThe Code Breaker

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

UsingUSA USA USAUKCanadaAustraliaIreland

Example eBay listing

Protein Modelling, Gamble, Andrew

SearcheBay.co.uk: protein model

Browse similar oneBay.co.uk

Example eBay listing

Protein Toxins in Modeling Biochemistry (SpringerBriefs in (New)

SearcheBay.co.uk: protein model

Browse similar oneBay.co.uk

Example eBay listing

British chemist Dorothy Hodgkin showing model of protein molecular- 1970 4x6 Pho

SearcheBay.co.uk: protein model

Browse similar oneBay.co.uk

Example eBay listing

Dna Models Classroom Human Dna Model Protein Model Kit Dna Molecule Model

SearcheBay.co.uk: protein model

Browse similar oneBay.co.uk

Browse more oneBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: nature.com
Link:https://www.nature.com/articles/s41586-021-03819-2

2. Source: pmc.ncbi.nlm.nih.gov
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC12956938/

3. Source: ebi.ac.uk
Link:https://www.ebi.ac.uk/training/online/courses/alphafold/alphafold-3-and-alphafold-server/introducing-alphafold-3/what-alphafold-3-struggles-with/

4. Source: pmc.ncbi.nlm.nih.gov
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC12262120/

5. Source: academic.oup.com
Link:https://academic.oup.com/bib/article/26/4/bbaf340/8202937

6. Source: sciencedirect.com
Link:https://www.sciencedirect.com/science/article/abs/pii/S0959440X26000357

7. Source: pmc.ncbi.nlm.nih.gov
Title: PMCAlpha Fold and protein folding: Not dead yet!
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC11892350/

8. Source: nature.com
Link:https://www.nature.com/articles/s41592-023-02087-4

9. Source: sciencedirect.com
Link:https://www.sciencedirect.com/science/article/am/pii/S0010482524007054

10. Source: cell.com
Link:https://www.cell.com/cell-systems/fulltext/S2405-4712%2825%2900341-2

11. Source: ebi.ac.uk
Title: strengths and limitations of alphafold
Link:https://www.ebi.ac.uk/training/online/courses/alphafold/an-introductory-guide-to-its-strengths-and-limitations/strengths-and-limitations-of-alphafold/

12. Source: sciencedirect.com
Link:https://www.sciencedirect.com/science/article/abs/pii/S014181302404618X

13. Source: nature.com
Link:https://www.nature.com/articles/s41467-025-56572-9

14. Source: pmc.ncbi.nlm.nih.gov
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC11956457/

15. Source: nature.com
Link:https://www.nature.com/articles/s41586-024-07487-w

16. Source: sciencedirect.com
Link:https://www.sciencedirect.com/science/article/abs/pii/S0022283624005308

17. Source: pmc.ncbi.nlm.nih.gov
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC11090348/

18. Source: nature.com
Link:https://www.nature.com/articles/s41467-025-62582-4

19. Source: nature.com
Link:https://www.nature.com/articles/s41598-024-84066-z

20. Source: nature.com
Link:https://www.nature.com/articles/s41467-026-69172-y

21. Source: nature.com
Link:https://www.nature.com/articles/s41598-025-17022-0

22. Source: nature.com
Link:https://www.nature.com/articles/s42003-026-10112-3

23. Source: nature.com
Link:https://www.nature.com/articles/s41467-024-46715-9

24. Source: sciencenews.org
Title: alphafold ai protein structure folding prediction
Link:https://www.sciencenews.org/article/alphafold-ai-protein-structure-folding-prediction

25. Source: frontiersin.org
Link:https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2022.906437/full

Additional References

26. Source: alphafold.ebi.ac.uk
Link:https://alphafold.ebi.ac.uk/

27. Source: alphafoldserver.com
Link:https://alphafoldserver.com/

28. Source: medium.com
Link:https://medium.com/advances-in-biological-science/can-we-predict-not-only-static-protein-structures-but-also-their-structural-diversity-fa1d9380fc34

29. Source: linkedin.com
Link:https://www.linkedin.com/posts/deniz-kavi_a-warning-that-proteins-are-highly-dynamic-activity-7351989182782754816-_QlR

30. Source: ouci.dntb.gov.ua
Link:https://ouci.dntb.gov.ua/en/works/4rG68kXl/

31. Source: youtu.be
Link:https://youtu.be/nIvsM8_GGlc?si=oBZshX_v1zM1MbaN

32. Source: youtube.com
Link:https://www.youtube.com/watch?v=1bXzOgUZ3PQ

33. Source: annualreviews.org
Title: [Alpha Fold]({{ ‘alpha-fold/’ | relative_url }}) and Protein Folding: Not Dead Yet!
Link:https://www.annualreviews.org/docserver/fulltext/biodatasci/7/1/annurev-biodatasci-102423-011435.pdf?accname=guest&checksum=11B4492DA6BE97A4FECAB9B72E8AB31F&expires=1750580512&id=id

34. Source: orbion.life
Title: alphafold s limitations what it can t predict (and how to fill the gaps)
Link:https://www.orbion.life/blog/alphafold-s-limitations-what-it-can-t-predict-%28and-how-to-fill-the-gaps%29

35. Source: link.springer.com
Link:https://link.springer.com/article/10.1007/s10930-025-10310-8