Unifrac tested (#123)

* Added unifrac calculation and small tests * Added large test from biocore/unifrac * Added large test from biocore/unifrac * fix travis error * Default to 0 for unspecified branch length * Fix style * Update empress/support_files/js/summary-helper.js Co-Authored-By: Yoshiki Vázquez Baeza <[email protected]> * Ported all unweighted unifrac tests from scikit-bio * minor comments error * Remove maxDiff, add comments, fix test names
biocore · Nov 14, 2019 · 98e1765 · 98e1765
1 parent d97e4bc
commit 98e1765
Show file tree

Hide file tree

Showing 4 changed files with 444 additions and 7 deletions.
diff --git a/empress/support_files/js/biom-table.js b/empress/support_files/js/biom-table.js
@@ -35,6 +35,25 @@ define([], function() {
         this._samp = samp;
     }
 
+    /**
+     * Returns a list of observations in the sample
+     * Example: countArray is [1,0], obsIDs is ['OTU1', 'OTU2'] => returns ['OTU1']
+     *
+     * @param {Array} countArray - Array of counts for different tips
+     * @param {Array} obsIDs - Array of observation IDs for each index of the countArray
+     *
+     * @return {Array}
+     */
+    BIOMTable.convertToObs = function(countArray, obsIDs) {
+        var obs = [];
+        for (var i = 0; i < countArray.length; i++) {
+            if (countArray[i] != 0) {
+                obs.push(obsIDs[i]);
+            }
+        }
+        return obs;
+    };
+
     /**
      * Returns a list of observations in the samples
      *
@@ -51,7 +70,7 @@ define([], function() {
             var obs = this._obs[sIds[i]];
             obs.forEach(addToResult);
         }
-        return Array(result);
+        return Array.from(result);
     };
 
     /**

diff --git a/empress/support_files/js/summary-helper.js b/empress/support_files/js/summary-helper.js
@@ -0,0 +1,67 @@
+define([], function() {
+    /**
+     * @class SummaryHelper
+     * Class functions for generating summary used in empress
+     */
+    function SummaryHelper() {}
+
+    /**
+     * Returns unifrac value for 2 sets of sample IDs given a biom table
+     * @param {BiomTable} biomTable - biom table that contains info about samples
+     * @param {BPTree} tree - bp tree
+     * @param {Array} sIds1 - First array of sample IDs to calculate unifrac for
+     * @param {Array} sIds2 - Second array of sample IDs to calculate unifrac for
+     *
+     * @return{Number}
+     */
+    SummaryHelper.unifrac = function(biomTable, tree, sIds1, sIds2) {
+        var uniq = 0;
+        var total = 0;
+        var uniqObs1 = biomTable.getObjservationUnionForSamples(sIds1);
+        var uniqObs2 = biomTable.getObjservationUnionForSamples(sIds2);
+
+        // Elements are in postorder with first element at index 0
+        // To keep track of if the node is in the sample1
+        var count1 = new Uint8Array(tree.names_.length);
+        // To keep track of if the node is in the sample2
+        var count2 = new Uint8Array(tree.names_.length);
+
+        // Based on the info in count1 and count2, calculate branch length
+        // Some optimization can be done, maybe perform the operation on the arrays
+        for (var i = 1; i < tree.size; i++) {
+            var treeIndex = tree.postorderselect(i);
+            var node = tree.name(treeIndex);
+
+            // If the node is leaf, check if it is in the union for sample
+            if (tree.isleaf(treeIndex)) {
+                count1[i - 1] = uniqObs1.includes(node);
+                count2[i - 1] = uniqObs2.includes(node);
+            }
+
+            // Update parent status
+            var parentPostOrder = tree.postorder(tree.parent(treeIndex));
+            count1[parentPostOrder - 1] |= count1[i - 1];
+            count2[parentPostOrder - 1] |= count2[i - 1];
+
+            var branchLength = tree.length(treeIndex);
+            if (branchLength) {
+                // Unique branch
+                if (count1[i - 1] ^ count2[i - 1]) {
+                    uniq += branchLength;
+                }
+                // The branch belongs to either or both samples
+                if (count1[i - 1] || count2[i - 1]) {
+                    total += branchLength;
+                }
+            }
+        }
+
+        console.log(count1);
+        console.log(count2);
+        console.log(uniq);
+        console.log(total);
+        return total == 0 ? 0 : uniq / total;
+    };
+
+    return SummaryHelper;
+});
diff --git a/tests/index.html b/tests/index.html
@@ -36,27 +36,31 @@
           'BPTree' : './support_files/js/bp-tree',
           'Camera' : './support_files/js/camera',
           'Colorer' : './support_files/js/colorer',
+          'BiomTable': './support_files/js/biom-table',
+          'SummaryHelper': './support_files/js/summary-helper',
 
           /* test paths */
           'testBPTree' : './../tests/test-bp-tree',
           'testByteTree' : './../tests/test-byte-array',
-          'testCamera' : './../tests/test-camera'
+          'testCamera' : './../tests/test-camera',
+          'testSummaryHelper': './../tests/test-summary-helper'
         }
     });
 
     // load tests
     require(
-        ['jquery', 'glMatrix', 'ByteArray', 'BPTree', 'Camera', 'testBPTree',
-         'testByteTree', 'testCamera'],
+        ['jquery', 'glMatrix', 'ByteArray', 'BPTree', 'Camera',
+         'BiomTable', 'SummaryHelper', 'testBPTree', 'testByteTree', 'testCamera',
+         'testSummaryHelper'],
 
         // start tests
-        function ($, gl, ByteArray, BPTree, Camera,testBPTree, testByteTree,
-                  testCamera) {
+        function ($, gl, ByteArray, BPTree, Camera, BiomTable, SummaryHelper,
+          testBPTree, testByteTree, testCamera, testSummaryHelper) {
             $(document).ready(function() {
                 QUnit.start();
             });
 
     });
   </script>
 </body>
-</html>
+</html>