Adjust default coalescing params based on HLA-DRB1 stress test

esm/components/data_layer/scatter.js

@@ -19,13 +19,13 @@ const default_layout = {
         max_points: 800, // Many plots are 800-2400 px wide, so, more than 1 datum per pixel of average region width
         // Define the "region of interest", like "bottom half of plot"; any points outside this region are taken as is
         //   Values are expressed in terms of data value and will be converted to pixels internally.
-        x_min: '-Infinity',
+        x_min: '-Infinity',  // JSON doesn't handle some valid JS numbers. Kids, don't get a career in computers.
         x_max: 'Infinity',
         y_min: 0,
-        y_max: 2.0,
-        // Expressed in units of px apart. For circular points, point area 40 = radius ~3.5.
-        x_gap: 4,
-        y_gap: 4,
+        y_max: 3.0,
+        // Expressed in units of px apart. For circles, area 40 = radius ~3.5; aim for ~1 diameter distance.
+        x_gap: 7,
+        y_gap: 7,
     },
     fill_opacity: 1,
     y_axis: {

esm/helpers/render.js

@@ -6,6 +6,10 @@
 /**
  * A very simple function aimed at scatter plots: attempts to coalesce "low-significance" SNPs that are too close to
  *  visually distinguish, thus creating a dataset with fewer points that can be rendered more quickly.
+ *
+ *  This depends on the strong and explicit assumption that points are ordered (typically in x position), so that
+ *    nearby points can be grouped by iterating over the data in sequence.
+ *
  * @param {Object[]} data Plot data, annotated with calculated `xc` and `yc` symbols for x and y coordinates (in px).
  * @param {Number} x_min The smallest x value of an "insignificant region" rectangle
  * @param {Number} x_max The largest x value of an "insignificant region" rectangle