aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorzsloan2018-04-09 16:51:54 +0000
committerzsloan2018-04-09 16:51:54 +0000
commitc9b0ab18457929bd7ca458f7207e50fe14099d6a (patch)
tree5cbc39ef322f6246c90914eb0b8f74fa05ba630e
parentfb62420ddbbf0189c9b0fb6d227121836fc377d8 (diff)
downloadgenenetwork2-c9b0ab18457929bd7ca458f7207e50fe14099d6a.tar.gz
Removed the basicStatistics directory/contents because nothing there was being used except corestats, which I moved to utility
Removed box plot code from Plot.py since we no longer use it either
-rw-r--r--wqflask/basicStatistics/BasicStatisticsFunctions.py207
-rw-r--r--wqflask/basicStatistics/__init__.py0
-rw-r--r--wqflask/utility/Plot.py134
-rw-r--r--wqflask/utility/corestats.py (renamed from wqflask/basicStatistics/corestats.py)8
-rw-r--r--wqflask/wqflask/show_trait/show_trait.py1
5 files changed, 3 insertions, 347 deletions
diff --git a/wqflask/basicStatistics/BasicStatisticsFunctions.py b/wqflask/basicStatistics/BasicStatisticsFunctions.py
deleted file mode 100644
index 1e5646a1..00000000
--- a/wqflask/basicStatistics/BasicStatisticsFunctions.py
+++ /dev/null
@@ -1,207 +0,0 @@
-from __future__ import print_function
-
-#import string
-from math import *
-#import piddle as pid
-#import os
-import traceback
-
-from pprint import pformat as pf
-
-from corestats import Stats
-
-import reaper
-from htmlgen import HTMLgen2 as HT
-
-#from utility import Plot
-from utility import webqtlUtil
-from base import webqtlConfig
-from db import webqtlDatabaseFunction
-
-def basicStatsTable(vals, trait_type=None, cellid=None, heritability=None):
- print("basicStatsTable called - len of vals", len(vals))
- 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])
-
- (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("should get a stack")
- traceback.print_stack()
- 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"))
-
-
-
- if (trait_type != None and trait_type == 'ProbeSet'):
- #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 not cellid:
- if heritability:
- # 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 st
-
-def plotNormalProbability(vals=None, RISet='', title=None, showstrains=0, specialStrains=[None], size=(750,500)):
-
- dataXZ = vals[:]
- dataXZ.sort(webqtlUtil.cmpOrder)
- dataLabel = []
- dataX = map(lambda X: X[1], dataXZ)
-
- showLabel = showstrains
- if len(dataXZ) > 50:
- showLabel = 0
- for item in dataXZ:
- strainName = webqtlUtil.genShortStrainName(RISet=RISet, input_strainName=item[0])
- dataLabel.append(strainName)
-
- dataY=Plot.U(len(dataX))
- dataZ=map(Plot.inverseCumul,dataY)
- c = pid.PILCanvas(size=(750,500))
- Plot.plotXY(c, dataZ, dataX, dataLabel = dataLabel, XLabel='Expected Z score', connectdot=0, YLabel='Trait value', title=title, specialCases=specialStrains, showLabel = showLabel)
-
- filename= webqtlUtil.genRandStr("nP_")
- c.save(webqtlConfig.GENERATED_IMAGE_DIR+filename, format='gif')
-
- img=HT.Image('/image/'+filename+'.gif',border=0)
-
- return img
-
-def plotBoxPlot(vals):
-
- valsOnly = []
- dataXZ = vals[:]
- for i in range(len(dataXZ)):
- valsOnly.append(dataXZ[i][1])
-
- plotHeight = 320
- plotWidth = 220
- xLeftOffset = 60
- xRightOffset = 40
- yTopOffset = 40
- yBottomOffset = 60
-
- canvasHeight = plotHeight + yTopOffset + yBottomOffset
- canvasWidth = plotWidth + xLeftOffset + xRightOffset
- canvas = pid.PILCanvas(size=(canvasWidth,canvasHeight))
- XXX = [('', valsOnly[:])]
-
- Plot.plotBoxPlot(canvas, XXX, offset=(xLeftOffset, xRightOffset, yTopOffset, yBottomOffset), XLabel= "Trait")
- filename= webqtlUtil.genRandStr("Box_")
- canvas.save(webqtlConfig.GENERATED_IMAGE_DIR+filename, format='gif')
- img=HT.Image('/image/'+filename+'.gif',border=0)
-
- plotLink = HT.Span("More about ", HT.Href(text="Box Plots", url="http://davidmlane.com/hyperstat/A37797.html", target="_blank", Class="fs13"))
-
- return img, plotLink
-
-def plotBarGraph(identification='', RISet='', vals=None, type="name"):
-
- this_identification = "unnamed trait"
- if identification:
- this_identification = identification
-
- if type=="rank":
- dataXZ = vals[:]
- dataXZ.sort(webqtlUtil.cmpOrder)
- title='%s' % this_identification
- else:
- dataXZ = vals[:]
- title='%s' % this_identification
-
- tvals = []
- tnames = []
- tvars = []
- for i in range(len(dataXZ)):
- tvals.append(dataXZ[i][1])
- tnames.append(webqtlUtil.genShortStrainName(RISet=RISet, input_strainName=dataXZ[i][0]))
- tvars.append(dataXZ[i][2])
- nnStrain = len(tnames)
-
- sLabel = 1
-
- ###determine bar width and space width
- if nnStrain < 20:
- sw = 4
- elif nnStrain < 40:
- sw = 3
- else:
- sw = 2
-
- ### 700 is the default plot width minus Xoffsets for 40 strains
- defaultWidth = 650
- if nnStrain > 40:
- defaultWidth += (nnStrain-40)*10
- defaultOffset = 100
- bw = int(0.5+(defaultWidth - (nnStrain-1.0)*sw)/nnStrain)
- if bw < 10:
- bw = 10
-
- plotWidth = (nnStrain-1)*sw + nnStrain*bw + defaultOffset
- plotHeight = 500
- #print [plotWidth, plotHeight, bw, sw, nnStrain]
- c = pid.PILCanvas(size=(plotWidth,plotHeight))
- Plot.plotBarText(c, tvals, tnames, variance=tvars, YLabel='Value', title=title, sLabel = sLabel, barSpace = sw)
-
- filename= webqtlUtil.genRandStr("Bar_")
- c.save(webqtlConfig.GENERATED_IMAGE_DIR+filename, format='gif')
- img=HT.Image('/image/'+filename+'.gif',border=0)
-
- return img
diff --git a/wqflask/basicStatistics/__init__.py b/wqflask/basicStatistics/__init__.py
deleted file mode 100644
index e69de29b..00000000
--- a/wqflask/basicStatistics/__init__.py
+++ /dev/null
diff --git a/wqflask/utility/Plot.py b/wqflask/utility/Plot.py
index d60e2bb2..3a8b8dd5 100644
--- a/wqflask/utility/Plot.py
+++ b/wqflask/utility/Plot.py
@@ -36,11 +36,10 @@ from numarray import linear_algebra as la
from numarray import ones, array, dot, swapaxes
import reaper
-# sys.path.append("..") Never in a running webserver
-from basicStatistics import corestats
import svg
import webqtlUtil
+import corestats
from base import webqtlConfig
import utility.logger
@@ -315,137 +314,6 @@ def find_outliers(vals):
logger.debug(pf(locals()))
return upper_bound, lower_bound
-
-def plotBoxPlot(canvas, data, offset= (40, 40, 40, 40), XLabel="Category", YLabel="Value"):
- xLeftOffset, xRightOffset, yTopOffset, yBottomOffset = offset
- plotWidth = canvas.size[0] - xLeftOffset - xRightOffset
- plotHeight = canvas.size[1] - yTopOffset - yBottomOffset
- iValues = []
- for item in data:
- for item2 in item[1]:
- try:
- iValues.append(item2[1])
- except:
- iValues.append(item2)
-
- #draw frame
- max_Y = max(iValues)
- min_Y = min(iValues)
- scaleY = detScale(min_Y, max_Y)
- Yll = scaleY[0]
- Yur = scaleY[1]
- nStep = scaleY[2]
- stepY = (Yur - Yll)/nStep
- stepYPixel = plotHeight/(nStep)
- canvas.drawRect(plotWidth+xLeftOffset, plotHeight + yTopOffset, xLeftOffset, yTopOffset)
-
- ##draw Y Scale
- YYY = Yll
- YCoord = plotHeight + yTopOffset
- scaleFont=pid.Font(ttf="cour",size=11,bold=1)
- for i in range(nStep+1):
- strY = cformat(d=YYY, rank=0)
- YCoord = max(YCoord, yTopOffset)
- canvas.drawLine(xLeftOffset,YCoord,xLeftOffset-5,YCoord)
- canvas.drawString(strY, xLeftOffset -30,YCoord +5,font=scaleFont)
- YYY += stepY
- YCoord -= stepYPixel
-
- ##draw X Scale
- stepX = plotWidth/len(data)
- XCoord = xLeftOffset + 0.5*stepX
- YCoord = plotHeight + yTopOffset
- scaleFont = pid.Font(ttf="tahoma",size=12,bold=0)
- labelFont = pid.Font(ttf="tahoma",size=13,bold=0)
- for item in data:
- itemname, itemvalue = item
- canvas.drawLine(XCoord, YCoord,XCoord, YCoord+5, color=pid.black)
- canvas.drawString(itemname, XCoord - canvas.stringWidth(itemname,font=labelFont)/2.0,\
- YCoord +20,font=labelFont)
-
- nValue = len(itemvalue)
- catValue = []
- for item2 in itemvalue:
- try:
- tstrain, tvalue = item2
- except:
- tvalue = item2
- if nValue <= 4:
- canvas.drawCross(XCoord, plotHeight + yTopOffset - (tvalue-Yll)*plotHeight/(Yur - Yll), color=pid.red,size=5)
- else:
- catValue.append(tvalue)
- if catValue != []:
- catMean = gmean(catValue)
- catMedian = gmedian(catValue)
- lowHinge = gpercentile(catValue, 25)
- upHinge = gpercentile(catValue, 75)
- Hstep = 1.5*(upHinge - lowHinge)
-
- outlier = []
- extrem = []
-
- upperAdj = None
- for item in catValue:
- if item >= upHinge + 2*Hstep:
- extrem.append(item)
- elif item >= upHinge + Hstep:
- outlier.append(item)
- elif item > upHinge and item < upHinge + Hstep:
- if upperAdj == None or item > upperAdj:
- upperAdj = item
- else:
- pass
- lowerAdj = None
- for item in catValue:
- if item <= lowHinge - 2*Hstep:
- extrem.append(item)
- elif item <= lowHinge - Hstep:
- outlier.append(item)
- if item < lowHinge and item > lowHinge - Hstep:
- if lowerAdj == None or item < lowerAdj:
- lowerAdj = item
- else:
- pass
- canvas.drawRect(XCoord-20, plotHeight + yTopOffset - (lowHinge-Yll)*plotHeight/(Yur - Yll), \
- XCoord+20, plotHeight + yTopOffset - (upHinge-Yll)*plotHeight/(Yur - Yll))
- canvas.drawLine(XCoord-20, plotHeight + yTopOffset - (catMedian-Yll)*plotHeight/(Yur - Yll), \
- XCoord+20, plotHeight + yTopOffset - (catMedian-Yll)*plotHeight/(Yur - Yll))
- if upperAdj != None:
- canvas.drawLine(XCoord, plotHeight + yTopOffset - (upHinge-Yll)*plotHeight/(Yur - Yll), \
- XCoord, plotHeight + yTopOffset - (upperAdj-Yll)*plotHeight/(Yur - Yll))
- canvas.drawLine(XCoord-20, plotHeight + yTopOffset - (upperAdj-Yll)*plotHeight/(Yur - Yll), \
- XCoord+20, plotHeight + yTopOffset - (upperAdj-Yll)*plotHeight/(Yur - Yll))
- if lowerAdj != None:
- canvas.drawLine(XCoord, plotHeight + yTopOffset - (lowHinge-Yll)*plotHeight/(Yur - Yll), \
- XCoord, plotHeight + yTopOffset - (lowerAdj-Yll)*plotHeight/(Yur - Yll))
- canvas.drawLine(XCoord-20, plotHeight + yTopOffset - (lowerAdj-Yll)*plotHeight/(Yur - Yll), \
- XCoord+20, plotHeight + yTopOffset - (lowerAdj-Yll)*plotHeight/(Yur - Yll))
-
- outlierFont = pid.Font(ttf="cour",size=12,bold=0)
- if outlier != []:
- for item in outlier:
- yc = plotHeight + yTopOffset - (item-Yll)*plotHeight/(Yur - Yll)
- #canvas.drawEllipse(XCoord-3, yc-3, XCoord+3, yc+3)
- canvas.drawString('o', XCoord-3, yc+5, font=outlierFont, color=pid.orange)
- if extrem != []:
- for item in extrem:
- yc = plotHeight + yTopOffset - (item-Yll)*plotHeight/(Yur - Yll)
- #canvas.drawEllipse(XCoord-3, yc-3, XCoord+3, yc+3)
- canvas.drawString('*', XCoord-3, yc+6, font=outlierFont, color=pid.red)
-
- canvas.drawCross(XCoord, plotHeight + yTopOffset - (catMean-Yll)*plotHeight/(Yur - Yll), \
- color=pid.blue,size=3)
- #print(catMean, catMedian, cat25per, cat75per)
- pass
-
- XCoord += stepX
-
- labelFont=pid.Font(ttf="verdana",size=18,bold=0)
- canvas.drawString(XLabel, xLeftOffset + (plotWidth -canvas.stringWidth(XLabel,font=labelFont))/2.0, \
- YCoord +40, font=labelFont)
- canvas.drawString(YLabel,xLeftOffset-40, YCoord-(plotHeight -canvas.stringWidth(YLabel,font=labelFont))/2.0,\
- font=labelFont, angle =90)
-
def plotSecurity(canvas, text="12345"):
if not text:
return
diff --git a/wqflask/basicStatistics/corestats.py b/wqflask/utility/corestats.py
index eba84c52..c48183ed 100644
--- a/wqflask/basicStatistics/corestats.py
+++ b/wqflask/utility/corestats.py
@@ -13,11 +13,9 @@
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
-
-
import sys
-
+#ZS: Should switch to using some third party library for this; maybe scipy has an equivalent
class Stats:
def __init__(self, sequence):
@@ -89,8 +87,6 @@ class Stats:
return value
-
-
# Sample script using this class:
# -------------------------------------------
# #!/usr/bin/env python
@@ -100,4 +96,4 @@ class Stats:
# stats = corestats.Stats(sequence)
# print stats.avg()
# print stats.percentile(90)
-# -------------------------------------------
+# ------------------------------------------- \ No newline at end of file
diff --git a/wqflask/wqflask/show_trait/show_trait.py b/wqflask/wqflask/show_trait/show_trait.py
index 6de5cd5f..1f000564 100644
--- a/wqflask/wqflask/show_trait/show_trait.py
+++ b/wqflask/wqflask/show_trait/show_trait.py
@@ -24,7 +24,6 @@ from utility import webqtlUtil, Plot, Bunch, helper_functions
from base.trait import GeneralTrait
from base import data_set
from db import webqtlDatabaseFunction
-from basicStatistics import BasicStatisticsFunctions
from pprint import pformat as pf