JMiV User's Guide Version 1.0 by Paul L. Brown, Duke University
The JMiV User’s Guide describes how to use the protein docking interface visualization features that have been added by the Duke University BioGeometry group to the Java Molecular Viewer (JMV) application. In its original form, JMV was developed by the Theoretical and Computational Biophysics Group in the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign with NIH support (http://www.ks.uiuc.edu/Research/jmv/). This guide will be limited to the new features we have added but a tutorial of the basic JMV features can be viewed at JMV Online HTML.
Changes to JMV
JMV has been extended to visualize protein-docking interfaces (thus JMiV) that were created with tools from computational geometry, including weighted Voronoi diagrams, weighted Delauney triangulations, and topological persistence. Simply described, the interface surface is a wrinkly (non-flat) sheet halfway between two proteins, trimmed where it protrudes beyond the interface. The interface is represented in 3D space imbedded within the original JMV visualization and also in a planarl parameterization that, although somewhat distorted, preserves the connectedness and biochemical characteristics at the docking interface.
Tiling Paradigm
The interface surface consists of Voronoi polygons, each one being associated with a pair of atoms – one from each “side” of the docking. The parameterization preserves this atom-pairing association and thus makes it possible to colorize the visualization based on characteristics directly or indirectly related to atomic structure at the protein docking interface. By convention, contiguous polygons having the same characteristic value will be drawn with the same color and the boundary between differing characteristic values will be denoted with a black line. Polygons with identical characteristics are fused to form tiles. Usually this means the polygons have the same color but same color polygons can be in different tiles, e.g. if they are not contiguous.
Primary and Secondary Tiles
Two levels of tiles can be visualized simultaneously - this is called Primary Tiling and Secondary Tiling. Colorization applies only to the characteristics chosen for the primary tiles. To help distinguish between primary and secondary tile boundaries, the thickness of the tile-boundary lines differs – a primary tile boundary is thicker than a secondary tile boundary. In the following example the colorization is by residue type (default) with a palette of color assignment provided at the bottom of the window and the secondary tiling is by atom type (no coloring).
Two Sides
By default only one side of the docking interface is visualized. The following example shows both sides of a docking interface visualized next to each other (how to achieve this view will be discussed in a subsequent section). Note that the title bar gives an indication as to front verses back by placing the chain identifier for the backside within brackets.
Merged Sides
The Merged view allows for both sides of the interface to be visualized simultaneously. When merged, the “back” side remains essentially unchanged while the “front” side is reduced to frames that outline the tiles from the “front” side. These frames have the black boundary-edge plus a colored inner frame that corresponds to the tile coloring that would be used in non-merge mode. On the “back” side, primary tile boundaries will be rendered as dashed lines.
Interface Embedding Menu
The visualization options for the embedded interface are accessed with a popup menu – this is invoked with a mouse right-click executed anywhere within the embedding visualization panel.
Reversing Visualizations
The docking interface is two sided by nature and by default only one perspective is visualized at a time. The Reverse option will cause the “other side” of the interface to become visible. Note that the title bar at the top of the drawing provides annotation as to what protein chain is on the “front verses back side” by a labeling convention of front [back], e.g. A [B] or B [A] in the following example.
Duplicating Visualizations
The Duplicate Panel option creates a copy (floating dialog) of the embedded interface, which may then be manipulated independently of the original visualization, i.e. it has its own popup menu control.
Primary Tiling
On the submenu of the Primary Tiling option are listed the eight tiling schemes that are currently supported.
Secondary Tiling
The submenu beneath Secondary Tiling is essentially the same as primary. To reduce the visual complexity of the rendering it may sometimes be desirable to set the secondary and primary tiling to the same value – this effectively removes the secondary tile boundaries from the drawing, as illustrated in the following image.
Border By Residue
It may be desirable to maintain some visualization feedback in regards to residue characteristics even when primary tiling characteristics have been set to something different than residue. This is accomplished by forcing (may be toggled on/off) the residue boundaries to always be rendered as a thick black line while at the same time not emphasizing the primary tile boundary lines. To get an idea of what this means, compare the images where the primary tiling as been set to backbone verses side-chain (image on right is bordering residue boundaries).
Mouse Interaction
Mouse-over Tool-tip
Pausing the mouse-pointer above the flattened interface will invoke the appearance of a tool-tip giving the characteristic values of the tile beneath. Also, this will always include information as to residue and atom beneath the pointer. In the un-merged mode the tool-tip appears as follows:
In merge mode, the tool-tip will also reflect information from the backside of the interface. In keeping with the convention used elsewhere the backside will be annotated within brackets.
Selection Mark
A left-mouse click above the interface embedding will select the secondary tile beneath the pointer (pressing the Ctrl key while doing this will cause the primary tile to be selected). The selected region will also be noted in the 3D representation.
In merge mode, mouse selection applies to the backside unless the Alt key is depressed in which case it applies to the primary tile on the front side as illustrated in the following.
Selected tiles are reflected in all views that are currently open. Multiple tile selection is also possible.
Interface Retraction
Interface retraction is a deletion process defined over the Delaunay triangulation. The process uses the triangle-tetrahedron persistence pairing to identify, quantify, and the remove features (also known as domes) in the Delaunay triangulation. Features are removed iteratively until there are no tetrahedra remaining. The removal of a feature creates a new retracted complex, which is given an index called the rank. The interface surface is constructed for each retracted complex. The initial retraction creates the initial retracted complex (rank 0) by starting from the boundary of the Delaunay triangulation (the convex hull) and iteratively removing tetrahedra with persistence zero until no more removals can be made. Subsequent retractions are performed by first identifying the triangle-tetrahedron pair with the largest seal value (as defined by the seal function) whose triangle exists on the boundary of the complex, and then removing the feature defined by the pair. The feature is removed by first collapsing the tetrahedron of the pairing and subsequently tunneling outwards until the triangle on the boundary is reached. The visual effect of this operation is that the retracted portion of the docking interface disappears from both the 3D and embedded visualizations.
Multiple Manifolds
The docking interface surface exists as a collection of manifolds with boundary glued at their boundaries when more than two chains are used in constructing the surface. Since the current embedding technique only considers a single manifold with boundary as input, interface surfaces constructed with more than two chains must be split into individual manifolds with boundary prior to embedding. When a docking interface has more than one manifold they are rendered separately, each on their own tabbed panel. Note: a gap in the manifold numbering implies that that portion of the interface surface did not meet certain topological constraints in order to be embedded – in this situation the manifold is skipped.
