Details
Proteins with little detectable sequence similarity can still share a three-dimensional fold because structure is often more conserved than sequence over evolutionary time. Searching by structure can therefore reveal remote relationships, suggest possible functions, and identify structural neighbours that sequence searches may miss.
Structure search compares a query protein’s three-dimensional model with structures in a database. Methods may superimpose atomic coordinates and measure geometric similarity, compare patterns of distances between residues, or encode local structural features as descriptors that can be searched efficiently at large scale. For example, Foldseek represents residue interactions with a structural alphabet and uses sequence-search techniques to find structurally similar proteins. Learn more about Foldseek.
The module is being developed through a collaboration between members of the Australian Structural Biology Computing community and the team from Protein Data Bank in Europe (PDBe).
Completed
- Define the module scope relating to different methods of protein structure search:
- Introduction and background context.
- Superposition-based search.
- Contact-based search.
- Token-based search.
- Shape-based search.
- Embedding-based search.
- Create initial draft of module content.
- Internal review of module content.
In progress
- Send draft to external reviewers.
- Incorporate feedback from external reviewers
- Prepare the module for publication