AlphaFold

AlphaFold uses deep learning to solve the "protein folding problem" — predicting how a linear sequence of amino acids folds into a stable, functional 3D structure. This is a major breakthrough in computational biology.

How It Works

• Trained on thousands of known protein structures and sequences.
• Uses evolutionary information (like multiple sequence alignments) and neural networks to predict:
  • Inter-residue distances
  • Angles and torsions
  • Final 3D atomic coordinates

Limitations

• Less accurate for:
  • Flexible or disordered regions
  • Multimeric complexes (partially addressed by AlphaFold-Multimer)
  • Dynamic conformational states
• Predictions need validation for critical applications (e.g., drug design)

What we offer

We provide support with interpretation of AlphFold models, and extracting relevant information from predicted models.