Structural similarity network tool

[AlexanderPico]

Background

Protein structure determination methodologies create sizable databases of X-ray solved structures (http://www.rcsb.org/pdb/). Similarities between these structures are invaluable for understanding of structure and function evolution, of both specific proteins of interest, or members of the same class of proteins. One way to represent these similarities is by structural similarity networks, in which nodes represent protein structures and edges represent pairwise similarities between them. These have proved to be very useful in the past by our lab in demonstrating structure-based grouping of proteins and understanding remote homologies between them. Users can greatly benefit from a standalone software package that inputs structure identifiers and outputs the similarity network, especially if their specific proteins of interest are highlighted in the “structure space” context.

Goal

The goal of this project is to create a python-based software package that inputs structure identifiers, computes pairwise similarities (TM-align/chimera matchmaker), reduce redundancy by grouping structures that are very similar to each other, and finally outputs a structural similarity network with some node metadata retrieved from the RCSB database. Such networks can be visualized using Cytoscape (local installation) or cytoscapejs.

Difficulty level: 1

Technology and Skills

• Web-based data retrieval and manipulation, simple GUI implementation

  Potential Mentors

• Eyal Akiva
• Shoshana Brown

  Contact

• babbitt@cgl.ucsf.edu
• eyal.akiva@ucsf.edu

[AlexanderPico]

postponed until mentors are available