Trying to get the stats to work

1 files changed, 69 insertions, 40 deletions

diff --git a/wqflask/basicStatistics/BasicStatisticsFunctions.py b/wqflask/basicStatistics/BasicStatisticsFunctions.py
index 285addae..d2584bdd 100755
--- a/wqflask/basicStatistics/BasicStatisticsFunctions.py
+++ b/wqflask/basicStatistics/BasicStatisticsFunctions.py
@@ -1,9 +1,15 @@
+from __future__ import print_function
+
 #import string
 from math import *
 #import piddle as pid
 #import os
 
-#import reaper
+from pprint import pformat as pf
+
+from corestats import Stats
+
+import reaper
 from htmlgen import HTMLgen2 as HT
 
 #from utility import Plot
@@ -12,58 +18,81 @@ from base import webqtlConfig
 from dbFunction import webqtlDatabaseFunction
 
 def basicStatsTable(vals, trait_type=None, cellid=None, heritability=None):
-
+    st = {}  # This is the dictionary where we'll put everything for the template
     valsOnly = []
     dataXZ = vals[:]
     for i in range(len(dataXZ)):
         valsOnly.append(dataXZ[i][1])
 
-    traitmean, traitmedian, traitvar, traitstdev, traitsem, N = reaper.anova(valsOnly) #ZS: Should convert this from reaper to R in the future
+    (st['traitmean'],
+     st['traitmedian'],
+     st['traitvar'],
+     st['traitstdev'],
+     st['traitsem'],
+     st['N']) = reaper.anova(valsOnly) #ZS: Should convert this from reaper to R in the future
+
+    #tbl = HT.TableLite(cellpadding=20, cellspacing=0)
+    #dataXZ = vals[:]
+    dataXZ = sorted(vals, webqtlUtil.cmpOrder)
+
+    print("data for stats is:", pf(dataXZ))
+    for num, item in enumerate(dataXZ):
+        print(" %i - %s" % (num, item))
+    print("  length:", len(dataXZ))
+
+    st['min'] = dataXZ[0][1]
+    st['max'] = dataXZ[-1][1]
+
+    numbers = [x[1] for x in dataXZ]
+    stats = Stats(numbers)
+
+    at75 = stats.percentile(75)
+    at25 = stats.percentile(25)
+    print("Interquartile:", at75 - at25)
+
+    #tbl.append(HT.TR(HT.TD("Statistic",align="left", Class="fs14 fwb ffl b1 cw cbrb", width = 180),
+    #                HT.TD("Value", align="right", Class="fs14 fwb ffl b1 cw cbrb", width = 60)))
+    #tbl.append(HT.TR(HT.TD("N of Samples",align="left", Class="fs13 b1 cbw c222"),
+    #                HT.TD(N,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    #tbl.append(HT.TR(HT.TD("Mean",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%2.3f" % traitmean,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    #tbl.append(HT.TR(HT.TD("Median",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%2.3f" % traitmedian,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    ##tbl.append(HT.TR(HT.TD("Variance",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    ##               HT.TD("%2.3f" % traitvar,nowrap="yes",align="left", Class="fs13 b1 cbw c222")))
+    #tbl.append(HT.TR(HT.TD("Standard Error (SE)",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%2.3f" % traitsem,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    #tbl.append(HT.TR(HT.TD("Standard Deviation (SD)", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%2.3f" % traitstdev,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    #tbl.append(HT.TR(HT.TD("Minimum", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%s" % dataXZ[0][1],nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+    #tbl.append(HT.TR(HT.TD("Maximum", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+    #                HT.TD("%s" % dataXZ[-1][1],nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+
+
 
-    tbl = HT.TableLite(cellpadding=20, cellspacing=0)
-    dataXZ = vals[:]
-    dataXZ.sort(webqtlUtil.cmpOrder)
-    tbl.append(HT.TR(HT.TD("Statistic",align="left", Class="fs14 fwb ffl b1 cw cbrb", width = 180),
-                    HT.TD("Value", align="right", Class="fs14 fwb ffl b1 cw cbrb", width = 60)))
-    tbl.append(HT.TR(HT.TD("N of Samples",align="left", Class="fs13 b1 cbw c222"),
-                    HT.TD(N,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    tbl.append(HT.TR(HT.TD("Mean",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%2.3f" % traitmean,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    tbl.append(HT.TR(HT.TD("Median",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%2.3f" % traitmedian,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    #tbl.append(HT.TR(HT.TD("Variance",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-    #               HT.TD("%2.3f" % traitvar,nowrap="yes",align="left", Class="fs13 b1 cbw c222")))
-    tbl.append(HT.TR(HT.TD("Standard Error (SE)",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%2.3f" % traitsem,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    tbl.append(HT.TR(HT.TD("Standard Deviation (SD)", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%2.3f" % traitstdev,nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    tbl.append(HT.TR(HT.TD("Minimum", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%s" % dataXZ[0][1],nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-    tbl.append(HT.TR(HT.TD("Maximum", align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                    HT.TD("%s" % dataXZ[-1][1],nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
     if (trait_type != None and trait_type == 'ProbeSet'):
-        #IRQuest = HT.Href(text="Interquartile Range", url=webqtlConfig.glossaryfile +"#Interquartile",target="_blank", Class="fs14")
-        #IRQuest.append(HT.BR())
-        #IRQuest.append(" (fold difference)")
-        tbl.append(HT.TR(HT.TD("Range (log2)",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                HT.TD("%2.3f" % (dataXZ[-1][1]-dataXZ[0][1]),nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-        tbl.append(HT.TR(HT.TD(HT.Span("Range (fold)"),align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                HT.TD("%2.2f" % pow(2.0,(dataXZ[-1][1]-dataXZ[0][1])), nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
-        tbl.append(HT.TR(HT.TD(HT.Span(HT.Href(url="/glossary.html#Interquartile", target="_blank", text="Interquartile Range", Class="non_bold")), align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
-                HT.TD("%2.2f" % pow(2.0,(dataXZ[int((N-1)*3.0/4.0)][1]-dataXZ[int((N-1)/4.0)][1])), nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+        #tbl.append(HT.TR(HT.TD("Range (log2)",align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+        #        HT.TD("%2.3f" % (dataXZ[-1][1]-dataXZ[0][1]),nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+        #tbl.append(HT.TR(HT.TD(HT.Span("Range (fold)"),align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+        #        HT.TD("%2.2f" % pow(2.0,(dataXZ[-1][1]-dataXZ[0][1])), nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+        #tbl.append(HT.TR(HT.TD(HT.Span(HT.Href(url="/glossary.html#Interquartile", target="_blank", text="Interquartile Range", Class="non_bold")), align="left", Class="fs13 b1 cbw c222",nowrap="yes"),
+        #        HT.TD("%2.2f" % pow(2.0,(dataXZ[int((N-1)*3.0/4.0)][1]-dataXZ[int((N-1)/4.0)][1])), nowrap="yes", Class="fs13 b1 cbw c222"), align="right"))
+        st['range_log2'] = dataXZ[-1][1]-dataXZ[0][1]
+        st['range_fold'] = pow(2.0, (dataXZ[-1][1]-dataXZ[0][1]))
+        st['interquartile'] = pow(2.0, (dataXZ[int((st['N']-1)*3.0/4.0)][1]-dataXZ[int((st['N']-1)/4.0)][1]))
 
         #XZ, 04/01/2009: don't try to get H2 value for probe.
-        if cellid:
-            pass
-        else:
+        if not cellid:
             if heritability:
-                tbl.append(HT.TR(HT.TD(HT.Span("Heritability"),align="center", Class="fs13 b1 cbw c222",nowrap="yes"),HT.TD("%s" % heritability, nowrap="yes",align="center", Class="fs13 b1 cbw c222")))
-            else:
-                pass
+                # This field needs to still be put into the Jinja2 template
+                st['heritability'] = heritability
+                #tbl.append(HT.TR(HT.TD(HT.Span("Heritability"),align="center", Class="fs13 b1 cbw c222",nowrap="yes"),HT.TD("%s" % heritability, nowrap="yes",align="center", Class="fs13 b1 cbw c222")))
+
         # Lei Yan
         # 2008/12/19
 
-    return tbl
+    return st
 
 def plotNormalProbability(vals=None, RISet='', title=None, showstrains=0, specialStrains=[None], size=(750,500)):