Next: Time
Axis and Up: DESIGN
Previous: Variables
The new AlignmentViewer uses three spatial axes and one temporal axis. Any of the above variables can be mapped to any of the four axes using the dialog box in Figure 1. All variables appear on the right side of the box. The four columns of radio buttons indicate which variables are mapped to the X, Y, Z, and time axis, respectively.
Figure 1: Choosing the axis mapping
When the position variable is mapped onto a spatial axis, an alignment appears as a comb-like glyph. The position of the comb in the 4D space is specified by the combination of the four mapped variables, with the comb extending along the position axis. We can imagine that the point opens up into a comb. An example is shown in Figure 2. The beginning, end, and relative length of the comb correspond to the beginning, end, and the length of the alignment. The comb teeth represent the matching vector. Each residue pair score in the matching vector determines the length of the corresponding tooth on the comb---the stronger the residue pair score, the longer the tooth. For example, line A in Figure 2 has a residue pair score of 17, and line B of -4. The tooth color has a dual purpose, encoding both frame number and the sign of the residue pair score. The frame number is encoded by assigning each frame with two colors. The +1 frame alignments use red (positive score---match likely) and blue (negative---match unlikely), +2 alignments green (positive) and yellow (negative), and +3 alignments magenta (positive) and cyan (negative).
If the variable position is not mapped to a spatial axis, an alignment appears as a single point with the matching vector glyph not shown. The result is therefore a 3D scatter plot. Since the matching vector specifies the alignment's composition, the glyph is important in identifying specific regions or trends in the alignment. Therefore, biologists often map the position variable to a spatial axis, resulting in the glyph representation.
Figure 2: Several alignments represented in AV
The glyph concept in the system is a significant departure from other multivariate analysis systems that handle only point sets. This departure is caused by the fundamental difference in the underlying dataset --- our data has an additional matching vector. A question for further consideration is whether we can use other high dimensional presentation techniques profitably to represent still more variables, or whether the current glyph/scatterplot representation is more useful for the biologists.
The visualization created by the old AV [6] can be recreated by mapping position, score, and frame to the X, Y and Z axes, respectively. For example, Figure 2 shows this mapping, and the alignment labeled B has a score of 95.
In using the original AV, biologists noticed that certain useful views were impossible to construct because the three spatial axes need to be scaled independently. For example, the user might want to zoom in on the position axis, but show the entire score axis. Thus, an additional enhancement is independent scaling of the X, Y and Z axes.