Heatmap

From ExpressionPlot
Jump to: navigation, search

A reliable way to compare transcriptional profiles from disparate samples and platforms is to focus on changes rather than absolute levels, because each comparison is then well controlled. This tool allows you to visualize many such comparisons at once, represented as a grid of colored squares. The x-axis corresponds to a project, each column associated with a different comparison of that project, and the y-axis corresponds to a (possibly different) project, each row associated with a different comparison of that project. The columns of the grid correspond to a numerical scale (a "heatmap") indicating the extent of the correlation or anti-correlation of the two change profiles. Another way to think of the heatmap is that each square summarizes a 4-way plot. Several different methods are provided to calculate a numerical value for the associations:

header
row 1, cell 1 row 1, cell 2 row 1, cell 3
row 2, cell 1 row 2, cell 2 row 2, cell 3


Each of these groups of genes can be further explored in the table browser by clicking on the adjacent action buttons.

The options for 4way are as follows:

x-axis, y-axis Three drop-down menus, which must be selected in order:
  1. Project
  2. Comparison (named comparison from comps.txt) for that project
  3. Event type: for example "genes" or "skipped exons".
P (for each axis) P-value cutoff for significance. Usually 1e-4 is used, as this roughly corresponds to 1-3 expected false positives. For skipped exons on the exon array platform, this field is instead used as a flag to decide whether or not to apply hybridization filters. P = 1 ignores these filters, 0 < P < 1 applies them (recommended).
FC (for each axis) Fold-change cutoff for significance. Default is 1.5.
Limits (x-min x-max y-max y-max) Leave these fields blank for the default behavior, which is to use a plotting window that contains all the points. Alternatively, you can specify a different boundary for any of the four sides. Use negative numbers for decreases and positive for increases. For example, x-min=-2 and x-max=5 means the plotting window will show genes/events decreased in the x-axis comparison up to 2-fold and increased in the x-axis comparison up to 5-fold.
Limit by ID Limit by ID allows you to restrict which points are plotted. Default is plot all points for which there are data, but you can supply a list

of IDs instead that you want to see, and only plot those points. These IDs might be, for example, the first column of a table browser from another comparison. In this way you can examine a gene set of interest within your data. The IDs can be uploaded or pasted (choose from drop-down menu). Pasted IDs can be separated by any whitespace. For uploaded IDs, the first line of the file is checked. If any (tab-separated) field matches /^cluster?id$/i (for example ID or clusterID), then that field is selected from the rest of the lines of the (tab-separated) file. Otherwise the file is parsed as whitespace-separated IDs.

You can also specify, using the drop-down menu, whether you want the limit applied to the x-axis or the y-axis. This only makes a difference if the two axes are from different genomes or event types.

width Width of the image in pixels
height Height of the image in pixels
Names Check the box to add the names of the genes to the plot. Often this doesn't look good, since the names tend to crowd each other out. In that case you can get the name lists by clicking the action buttons.
y=x Add the diagonal line y=x to the plot.
y=-x Add the anti-diagonal line y=-x to the plot.