POSA: a user-driven, interactive multiple protein structure alignment server

Abstract

POSA (Partial Order Structure Alignment), available at http://posa.godziklab.org, is a server for multiple protein structure alignment introduced in 2005 (Ye,Y. and Godzik,A. (2005) Multiple flexible structure alignment using partial order graphs. Bioinformatics, 21, 2362-2369). It is free and open to all users, and there is no login requirement, albeit there is an option to register and store results in individual, password-protected directories. In the updated POSA server described here, we introduce two significant improvements. First is an interface allowing the user to provide additional information by defining segments that anchor the alignment in one or more input structures. This interface allows users to take advantage of their intuition and biological insights to improve the alignment and guide it toward a biologically relevant solution. The second improvement is an interactive visualization with options that allow the user to view all superposed structures in one window (a typical solution for visualizing results of multiple structure alignments) or view them individually in a series of synchronized windows with extensive, user-controlled visualization options. The user can rotate structure(s) in any of the windows and study similarities or differences between structures clearly visible in individual windows.

Figure 1.

A typical POSA server workflow: (A) Input interface to start MPStrA jobs on the POSA server. Users fill in an input table providing PDB IDs of the structures (alternatively uploading coordinates from a file) and the chain IDs. Additionally, the user has an option to define segments in one or more structures to anchor alignment on these specific parts. Specifying ‘Other Chains’ will not affect the alignment result but will allow these chains to be visualized consistently with the chain or segment(s) being aligned. (B) The job home page contains links to the MPStrA result visualization, information about the common core and the RMSD value detected for the alignment. It also displays the alignment guide tree that indicates similarity by branch distance. Clicking on the respective sub-branches will display the alignment of proteins in the chosen sub-branch only. An editable input table with the current settings makes it easy to iteratively refine the current alignment. (C) Visualization page of the alignment result. The upper menu provides links to specific result information such as the POG or the sequence of the common core. Below are the Jmol views for visualization of the superposed or individual structures. All views are synchronized by default but can be viewed independently if required. The menu on the left of the main view controls multiple visualization modes. POSA visualization on a large monitor and the arrangement of the Jmol windows may vary for smaller screens.

Figure 2.

Multiple display modes of the result visualization: (A) All zinc finger structures superposed after POSA alignment. All ligands and other chains specified are visible. (B) View on the same scene displaying the aligned chains only with all ligands (zinc atoms) visible. (C) Only structures for which ‘Other Chains’ have been specified are displayed. This view mode allows direct comparison of multiple protein–nucleic acid and protein–protein interactions. (D) The common core matches the first zinc finger domain of all aligned structures. The highlight button will set all the non-core segments to transparent.