A Shortcut to Structure in AlphaFold 3

Key points

• Here, through causal interventions on internal representations and direct probing of every Pairformer block, we trace the formation of global protein geometry and identify the multiple sequence alignment (MSA) as a structural shortcut to the fold.1
• Restoring the MSA-enriched pair representation at only forty residues recovers most of this lost organization, including at pairs never directly modified.1
• The Pairformer rapidly converts this signal into global geometry: the final fold becomes recoverable by approximately block 9 of 48 for a majority of proteins, roughly twenty-seven blocks before the model's decoder can render it, whereas without the MSA it remains inaccessible for most proteins throughout the pass.1
• What AlphaFold 3 reads from an alignment is therefore a description of the fold itself, transferable between proteins that share one, rather than the query's own evolutionary history.1

Sources (1)

• [1]A Shortcut to Structure in AlphaFold 3
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 06:04 PM

  AlphaFold 3 predicts protein structures with remarkable accuracy, yet how structural information emerges within the model remains poorly understood.

  Here, through causal interventions on internal representations and direct probing of every Pairformer block, we trace the formation of global protein geometry and identify the multiple sequence alignment (MSA) as a structural shortcut to the fold.

Extractive summary: sentences quoted from the sources.