Skip to content

Latest commit

 

History

History
196 lines (160 loc) · 7.57 KB

README.md

File metadata and controls

196 lines (160 loc) · 7.57 KB

Eye diagram visualization using D3.js

Note: This is "under-development" (to the extent I have time). While usable, the experience might not be very smooth with regard to configuration, documentation, etc.

Eye diagram visualization using the D3.js javascript library.

See http://www.tmpl.fi/d3js_eye_diagram/ for an example diagram.

See Probabilistic retrieval and visualization of biologically relevant microarray experiments by Caldas et al. for the original visualization technique. Their code is available at https://github.com/ouzor/eyediagram.

See eye_diagram_data.R for an example how to generate eye diagram datafile from CCAGFA objects.

Usage on a web page

(1) In the head of the html page, include D3.js, the diagram code (which exports a single javascript function draw_diagram), and some css (here #dg is the id of the svg element of the visualization).

<script src="https://cdnjs.cloudflare.com/ajax/libs/d3/3.5.6/d3.min.js" charset="utf-8"></script>
<script src="eye_diagram_browser.js" charset="utf-8"></script>
<style>
#dg {
  background-color: #fff;
  font-size: 4px;
}
#dg .line {
  fill: none;
}
#dg .nodes {
  stroke: #000;
  stroke-width: 0.5;
}
</style>

(2) In the body of the html page, include the svg element, set some options, load data (here, from data.json; see below), and draw the diagram.

<svg id="dg"></svg>

<script type="text/javascript">
// options, see the opts structure below for details
var height = 600;
var width = 700;
var tension = 1;
var ctrl_point_offset_coeff = 0.3;
var Z_y_margin_coeff = 0.25;
var radius = 400;
var nodesize = 3;
var Z_labels_dy = 8;
var Z_labels_dx = 0;
var Y_labels_dy = 2;
var Y_labels_dx = 5;
var Y_left_angle = 3.1415 / 2;
var Y_right_angle = 3.1415 / 2;

var opts = { height: height,
             width: width,
             // parameter of edge behaviour (d3 "bundle" spline parameter)
             tension: tension,
             // parameter of edge behaviour (determines where spline control
             // points lie between the connected nodes)
             ctrl_point_offset_coeff: ctrl_point_offset_coeff,
             // Z are the nodes in the middle of the diagram positioned on a
             // vertical line
             Z: {
               // nodes will lie in a vertical line between y0 and y1 at equal
               // spacing
               y0: Z_y_margin_coeff * height,
               y1: (1 - Z_y_margin_coeff) * height,
               // x position of the nodes
               x: width / 2,
               nodesize: nodesize,
               draw_labels: true,
               // label text horizontal offset
               labels_dx: Z_labels_dx,
               // label text vertical offset
               labels_dy: Z_labels_dy,
               // label text anchor
               labels_anchor: "middle"
             },
             // Y_left and Y_right are side nodes positioned along a part of
             //  the perimeter of a circle according to the options
             Y_left: {
               // how large part of the perimeter of the cirle is used
               // (angle_at_node runs from -angle/2 to angle/2 with equal
               // spacing)
               angle: Y_left_angle,
               // vertical radius of the cirle
               y_radius: radius,
               // vertical center of the circle
               // [node_y_pos = y_offset + y_radius * sin(angle_at_node)]
               y_offset: height / 2,
               // horizontal radius of the circle (negative for left size nodes)
               x_radius: -radius,
               // horizontal center of the cirle
               // [node_x_pos = x_offset + x_radius * cos(angle_at_node)]
               x_offset: radius + 0.05 * width,
               // 1 for left side nodes, -1 for right side (if 
               // x_radius != y_radius, edges will not be exactly normal to the
               // tangent of the circle)
               tan_dir: 1,
               nodesize: nodesize,
               draw_labels: true,
               labels_dx: -Y_labels_dx,
               labels_dy: Y_labels_dy,
               labels_anchor: "end"
             },
             Y_right: {
               angle: Y_right_angle,
               y_radius: radius,
               y_offset: height / 2,
               x_radius: radius,
               x_offset: 0.95 * width - radius,
               tan_dir: -1,
               nodesize: nodesize,
               draw_labels: true,
               labels_dx: Y_labels_dx,
               labels_dy: Y_labels_dy,
               labels_anchor: "start"
             }
           };

// load data and draw
d3.json("data.json", function(error, json) {
  if (error) return console.warn(error);
  draw_diagram("#dg", json.Y_left, json.Y_right, json.Z, json.Y_left_to_Z, json.Y_right_to_Z, opts);
});
</script>

Data format

The draw_diagram-function takes 5 parameters describing the nodes and edges of the diagram (in addition to the id and options).

The first three (json.Y_left, json.Y_right, json.Z above) describe the nodes of the diagram (in the order left, right, middle). They should be javascript arrays with elements of form { "label": "Y1", "c": "#000000" }, where "label" gives the label text and "c" gives the fill color of the node.

The latter two (json.Y_left_to_Z and json.Y_right_to_Z above) describe the edges of the diagram (from the left and right nodes to the middle nodes). They should be javascript arrays with elements of form {"Y": 0, "Z": 0, "w": 1, "o": 0.6, "c": "#990000" }, where "Y" gives the index of the side node in the corresponding node array, "Z" the index of the middle node, "w" the width of the edge, "o" the opacity, and "c" the color.

See data.json for an example file.

See eye_diagram_data.R for an example of generating the json-file in R code.

Rendering to svg-file

Rendering directly to an svg-file is useful, for example, for post-processing (adding further annotations etc.) or saving to pdf using CairoSVG (see below) or Inkscape.

To export the svg diagram (if data filename is omitted, data.json is used by default; if configuration file is omitted, the default configuration in the script is used; currently if given, the configuration file needs to be complete):

node export_eye_diagram.js data.json config.json > test.svg

This requires nodejs with d3 and jsdom, which can be installed using npm:

npm install d3
npm install jsdom

One can use python based CairoSVG to convert svg to pdf (see http://stackoverflow.com/questions/14903233/converting-svg-with-embedded-css-to-pdf-in-python). Install it and dependencies via pip:

pip3 install cairosvg lxml tinycss cssselect

Then run:

cairosvg test.svg -o test.pdf

TODO

  • Make nicer looking example using real data.
  • Make better automatic configuration tailored according to the data file.
  • Allow specifying partial configuration and use defaults elsewhere.
  • Code is now duplicated between browser and nodejs versions; make them use same core code.
  • Make the R code data generation from CCAGFA objects more flexible (try optimizing variable and component ordering for best visual effect?).

Contact

License

The MIT License. See LICENSE file for the full license text.