Add argument for performing leaveOneOutTest without averaging

Is not quite the original leaveOneOut test as it was done before.
The original test takes extremely long to run because it
must average up the family after each exact rate rule.
In this method, we remove the original rate rule, and see if
using an average that (should incorporate the original) does a
good job of estimating that rate rule.

The option to use the original version of the algorithm can still be
used by setting averaging=True

Author: connie

testing/evaluateKinetics.py

@@ -75,7 +75,7 @@ def getKineticsDepository(FullDatabase, family, depositoryLabel):
 
     return exactKinetics, approxKinetics
 
-def getKineticsLeaveOneOut(family):
+def getKineticsLeaveOneOut(family, averaging=True):
     """
     Performs the leave one out test on a family. It returns a dictionary of 
     the original exact nodes and a dictionary of the new averaged nodes. 
@@ -91,13 +91,35 @@ def getKineticsLeaveOneOut(family):
 
     for entryKey in family.rules.entries.keys():
         template = family.retrieveTemplate(entryKey.split(';'))
-        exactKinetics[entryKey], exactKineticsEntry=family.rules.estimateKinetics(template)
+        exactKineticsData, exactKineticsEntry=family.rules.estimateKinetics(template)
+        if exactKineticsData.comment:
+            # This means it was an averaged node so skip
+            continue
+
+        exactKinetics[entryKey] = exactKineticsData
+
+        if averaging:
+            # In this scheme, we remove the data fully and try to pretend the database 
+            # wants this kinetic value when we know nothing about it
+            familyCopy=copy.deepcopy(family)
+            familyCopy.rules.entries.pop(entryKey)
+            familyCopy.fillKineticsRulesByAveragingUp()
+            approxKinetics[entryKey], approxKineticsEntry=familyCopy.rules.estimateKinetics(template)
+        else:
+            # In this scheme, do not re-average the tree, just try to see if the nearest
+            # best node provides a good estimate for the original kinetics.  
+            # This takes significanty less time to run, but is not a true validation of the data, 
+            # it is just testing our kinetics selection algorithm
+
+            originalEntry = family.rules.entries(entryKey)
+            family.rules.entries.pop(entryKey)
+            approxKinetics[entryKey], approxKineticsEntry=family.rules.estimateKinetics(template)
+            # Re-add the data back into the original family
+            family.rules.entries[entryKey] = originalEntry
+
+
 
-        familyCopy=copy.deepcopy(family)
-        familyCopy.rules.entries.pop(entryKey)
-        familyCopy.fillKineticsRulesByAveragingUp()
 
-        approxKinetics[entryKey], approxKineticsEntry=familyCopy.rules.estimateKinetics(template)
 
     return exactKinetics, approxKinetics
 
@@ -289,7 +311,7 @@ def compareNIST(FullDatabase, trialDir):
 
 
 
-def leaveOneOut(FullDatabase, trialDir):
+def leaveOneOut(FullDatabase, trialDir, averaging=True):
     """
     Performs leave one out analysis on all the kinetics families.
     The algorithm deletes a single entry in the family, and then re-averages the tree
@@ -315,7 +337,10 @@ def leaveOneOut(FullDatabase, trialDir):
         if len(family.rules.entries) < 2:
             print '    Skipping', familyName, ': only has one rate rule...'
         else:
-            exactKinetics, approxKinetics = getKineticsLeaveOneOut(family)
+            if not averaging:
+                # Pre-average the family if averaging is not turned on
+                family.fillKineticsRulesByAveragingUp()
+            exactKinetics, approxKinetics = getKineticsLeaveOneOut(family, averaging)
             parityData=analyzeForParity(exactKinetics, approxKinetics, cutoff=8.0)
 
             if len(parityData)<2:
@@ -370,7 +395,7 @@ def leaveOneOut(FullDatabase, trialDir):
 
     print '--------------------------------------------'
     print 'Performing the leave on out test on the kinetics families...'
-    leaveOneOut(FullDatabase, trialDir)
+    leaveOneOut(FullDatabase, trialDir, averaging=False)
 
     print '--------------------------------------------'
     print 'Filling up the family rate rules by averaging... Expect larger number of rate rules in subsequent tests'