Protein Structure Space Explorer
Overview
PSpace allows for exploration of structural relationships between proteins utilizing structure space mapping. Mapping is achieved by performing all-against-all alignments on representative protein structures, then embedding the pairwise structural similarities into a low dimensional space suitable for visualization and algorithmic analysis. This transformation from high to low dimensionality is performed through multi-dimensional scaling. Alignments are generated from the Astral and PDB Select reference sets, optionally including user selected or uploaded structures as well.
Exploration
Generated space maps of all-by-all structural alignments can be visualized in a custom 3D scatter-plot. Chains of interest can be selected either through keyboard entry of the chain identifier or through visual navigation of the space map. Structures can be tagged for visual reference and annotated with CATH or SCOP; the structural classification is visualized as a color dimension. K-Nearest neighbors can be selected dynamically for any structure as a sorted list of annotated structures.
Alignment Data
Alignments are performed using any of three algorithms for protein structure alignment: Dali, CE or FATCAT. The alignments are run on the Pspace server, using the DaliLite workbench or the BioJava library.
Authors
David Foote
San Francisco State University,
Department of Computer Science
Daniel Asarnow
San Francisco State University,
Department of Chemistry and Biochemistry
Rahul Singh
San Francisco State University,
Department of Computer Science
Reference
Publication
D. Foote, D. Asarnow and R. Singh, "The Protein Structure Space Explorer", in submission.
Corresponding Author
Rahul Singh: rahul @ sfsu . edu
Funding
Creation of PSPACE was funded in part by the NSF through grant IIS-0644418 (CAREER).