def mean(inList, null = "NA"):
"""Calculates mean"""
cList = mData.floatList(inList)
if len(cList) == 0:
mean = null
else:
mean = sum(cList)/len(cList)
return (mean)
def mean(inList, null="NA"):
"""Calculates mean"""
cList = mData.floatList(inList)
if len(cList) == 0:
mean = null
else:
mean = sum(cList) / len(cList)
return (mean)
def quartiles(inList):
"""Returns the 25/50/75 quartiles"""
cList = mData.floatList(inList)
cList.sort()
if len(cList) < 2:
boundaries = ["NA", "NA", "NA"]
else:
boundaries = [median(cList[:len(cList)/2]), median(cList), median(cList[len(cList)/2:])]
return(boundaries)
def median(inList):
"""Calculates median"""
cList = mData.floatList(inList)
cList.sort()
if len(cList) == 0:
median = "NA"
else:
if len(cList)%2 == 1:
median = cList[len(cList)/2]
else:
median = (cList[len(cList)/2]+cList[(len(cList)/2)-1])/2.0
return(median)
def median(inList):
"""Calculates median"""
cList = mData.floatList(inList)
cList.sort()
if len(cList) == 0:
median = "NA"
else:
if len(cList) % 2 == 1:
median = cList[len(cList) / 2]
else:
median = (cList[len(cList) / 2] +
cList[(len(cList) / 2) - 1]) / 2.0
return (median)
def quartiles(inList):
"""Returns the 25/50/75 quartiles"""
cList = mData.floatList(inList)
cList.sort()
if len(cList) < 2:
boundaries = ["NA", "NA", "NA"]
else:
boundaries = [
median(cList[:len(cList) / 2]),
median(cList),
median(cList[len(cList) / 2:])
]
return (boundaries)
def mean_std(inList, sample=True):
"""Calculates mean and std"""
cList = mData.floatList(inList)
if len(cList) == 0:
mean = "NA"
std = "NA"
else:
mean = sum(cList) / float(len(cList))
std = 0.0
for i in cList:
std += (i - mean)**2
if len(cList) > 1:
if sample:
std = math.sqrt(std / (len(cList) - 1))
else:
std = math.sqrt(std / len(cList))
else:
std = 0.0
return (mean, std)
def mean_std(inList, sample = True):
"""Calculates mean and std"""
cList = mData.floatList(inList)
if len(cList) == 0:
mean = "NA"
std = "NA"
else:
mean = sum(cList)/float(len(cList))
std = 0.0
for i in cList:
std += (i-mean)**2
if len(cList) > 1:
if sample:
std = math.sqrt(std/(len(cList)-1))
else:
std = math.sqrt(std/len(cList))
else:
std = 0.0
return(mean, std)
def getCohortMinMaxValues(featureList, sampleList, circleData): """Get the minVal and maxVal of sample scores among the specified featureList for the ring/dataset.""" minValList = [] maxValList = [] for ring in xrange(len(circleData)): ringVals = [] # get ring values in effort to find min/max values for each *ring* for sample in sampleList: if sample in circleData[ring]: for feature in featureList: if feature in circleData[ring][sample]: ringVals.append(circleData[ring][sample][feature]) elif "*" in circleData[ring][sample]: ringVals.append(circleData[ring][sample]["*"]) # find the min & max sample scores for this ring in this feature floatList = mData.floatList(ringVals) minValList.append(min([-0.01] + floatList)) maxValList.append(max([0.01] + floatList)) return (minValList, maxValList)
def getCohortMinMaxValues(featureList, sampleList, circleData):
"""Get the minVal and maxVal of sample scores among the specified featureList for the ring/dataset."""
minValList = []
maxValList = []
for ring in xrange(len(circleData)):
ringVals = []
# get ring values in effort to find min/max values for each *ring*
for sample in sampleList:
if sample in circleData[ring]:
for feature in featureList:
if feature in circleData[ring][sample]:
ringVals.append(circleData[ring][sample][feature])
elif "*" in circleData[ring][sample]:
ringVals.append(circleData[ring][sample]["*"])
# find the min & max sample scores for this ring in this feature
floatList = mData.floatList(ringVals)
minValList.append(min([-0.01] + floatList))
maxValList.append(max([0.01] + floatList))
return (minValList, maxValList)
## cohort png
if len(orderFiles) > 0:
imgFile = "%s/Cohort.png" % (outputDir)
label = "Cohort"
centerCol = rgb(255, 255, 255).tohex()
circleCols = []
for i in range(len(orderData)):
ringCols = []
ringVals = []
for sample in samples:
if sample in orderData[i]:
if orderFeature in orderData[i][sample]:
ringVals.append(orderData[i][sample][orderFeature])
elif "*" in orderData[i][sample]:
ringVals.append(orderData[i][sample]["*"])
minVal = min([-0.01]+mData.floatList(ringVals))
maxVal = max([0.01]+mData.floatList(ringVals))
for sample in samples:
if sample in orderData[i]:
if orderFeature in orderData[i][sample]:
ringCols.append(getColor(orderData[i][sample][orderFeature], minVal, maxVal, minColor = orderColors[i][0], zeroColor = orderColors[i][1], maxColor = orderColors[i][2]))
elif "*" in orderData[i][sample]:
ringCols.append(getColor(orderData[i][sample]["*"], minVal, maxVal, minColor = orderColors[i][0], zeroColor = orderColors[i][1], maxColor = orderColors[i][2]))
else:
ringCols.append(rgb(200, 200, 200).tohex())
else:
ringCols.append(rgb(200, 200, 200).tohex())
circleCols.append(ringCols)
plotCircle(imgFile, label = label, centerCol = centerCol, circleCols = circleCols, innerRadTotal=0.2, outerRadTotal=0.5, width = 5)
## plot images
def cli_routine(outputDir, circleFiles, orderFiles, sampleFile, featureFile, orderFeature, centerFile, colorscaleFile, printLabel, verbose, cohortMinMax=False, purpleHack = True): """Routine for program execution via command-line.""" # I've tried not to touch this method as much as possible. # I don't want to break the way it was working for Sam Ng. # chrisw ## execute samples = [] features = [] if sampleFile != None: samples = mData.rList(sampleFile) if featureFile != None: features = mData.rList(featureFile) # end section for getting lists of samples and features ## read circleFiles # circleData is a list of dict[col][row]=score from each circleFile circleData = [] # circleColorsPalette is a list of (minColor),(zeroColor),(maxColor) circleColorsPalette = [] ## read colorscaleFile # the format is as follows - header compulsory: # min/max color coding color1 color2 color 3 # -2,2 rgb 155,155,155 255,255,255 0,0,0, # - rgb 155,0,155 255,0,255 0,0,0, # the "color format" is intended to support more color format, as I have # seen the html-colors in the code. # Michael ([email protected]) colorscaleData = None if colorscaleFile != None: if cohortMinMax: log("WARNING: The -k option overrides -m") colorscaleData = mData.retRows(colorscaleFile,aslist=True) line=1 for cs in colorscaleData: line = line + 1 if len(cs) != 5: log("ERROR: color scale needs five fields: datapoints, colorcoding(rgb) and three colors\n", die = True) try: cs[0] = [float(x) for x in cs[0].split(",")] except ValueError: pass if len(cs[0]) != 2 and cs[0] != "-": print cs[0] log("ERROR: Two data points or dash needed for color scale\n", die = True) if cs[1].lower() == "rgb": try: cs[2] = rgb(*[float(x) for x in cs[2].split(",")]) cs[3] = rgb(*[float(x) for x in cs[3].split(",")]) cs[4] = rgb(*[float(x) for x in cs[4].split(",")]) except TypeError: log("ERROR: RGB needs three values on line " + str(line) + "\n", die = True) except ValueError: log("ERROR: RGB color not correctly defined on line " + str(line) + "\n", die=True) else: log("ERROR: Unknown color coding on line " + str(line) + ": " + str(cs[1]) + "\n", die=True) for i in xrange(len(circleFiles)): # get data, samples, and features from each circleFile # data is a dict[col][row]=score # cols is a list of sample names # features is a list of feature names (data, cols, rows) = mData.rCRSData(circleFiles[i], retFeatures=True) circleData.append(data) minCol = lightBlueRGB zerCol = whiteRGB maxCol = redRGB if colorscaleFile != None and i<len(colorscaleData): #get colors from specified colorscaleFile minCol = colorscaleData[i][2] zerCol = colorscaleData[i][3] maxCol = colorscaleData[i][4] # special cases for -meth and -mut # if circleFiles[i].endswith("meth"): # maxCol = blueRGB # minCol = redRGB # log("Color: meth\n") # elif circleFiles[i].endswith("mut"): # maxCol = blackRGB # minCol = whiteRGB # log("Color: mut\n") circleColorsPalette.append((minCol, zerCol, maxCol)) # if no sampleFile/featureFile, default to using samples/features from circleFiles if sampleFile == None: samples = list(set(cols) | set(samples)) if featureFile == None: features = list(set(rows) | set(features)) # end section for reading circleFiles ## read centerFile centerData = None if centerFile != None: centerData = mData.r2Col(centerFile, header=True) ## sort if orderFeature != None: if len(orderFiles) > 0: orderData = [] orderColors = [] for i in xrange(len(orderFiles)): orderData.append(mData.rCRSData(orderFiles[i])) minCol = whiteRGB zerCol = whiteRGB maxCol = blackRGB orderColors.append((minCol, zerCol, maxCol)) else: orderData = circleData # sort samples based on sample score in orderData # priority of sorting determined by orderFiles parameter samples.sort(lambda x, y: scmp(x, y, orderFeature, orderData)) ## cohort png # cgi will probably not use orderFiles if len(orderFiles) > 0: imgFile = "%s/Cohort.png" % (outputDir) label = "Cohort" centerCol = whiteRGB.tohex() cohortCircleCols = [] for i in xrange(len(orderData)): ringCols = [] ringVals = [] for sample in samples: if sample in orderData[i]: if orderFeature in orderData[i][sample]: ringVals.append(orderData[i][sample][orderFeature]) elif "*" in orderData[i][sample]: ringVals.append(orderData[i][sample]["*"]) minVal = min([-0.01] + mData.floatList(ringVals)) maxVal = max([0.01] + mData.floatList(ringVals)) for sample in samples: if sample in orderData[i]: if orderFeature in orderData[i][sample]: ringCols.append(getColor(orderData[i][sample][orderFeature], minVal, maxVal, minColor=orderColors[i][0], zeroColor=orderColors[i][1], maxColor=orderColors[i][2])) elif "*" in orderData[i][sample]: ringCols.append(getColor(orderData[i][sample]["*"], minVal, maxVal, minColor=orderColors[i][0], zeroColor=orderColors[i][1], maxColor=orderColors[i][2])) else: ringCols.append(greyRGB.tohex()) else: ringCols.append(greyRGB.tohex()) cohortCircleCols.append(ringCols) plotCircle(imgFile, label=label, centerCol=centerCol, circleCols=cohortCircleCols, innerRadTotal=0.2, outerRadTotal=0.5, width=5) # end section for sample ordering ## plot images if centerData != None: centerDataFloatList = mData.floatList(centerData.values()) centerDataMinVal = min([-0.01] + centerDataFloatList) centerDataMaxVal = max([0.01] + centerDataFloatList) # get min/max values for datasets if cohortMinMax: (minValList, maxValList) = getCohortMinMaxValues(features, samples, circleData) else: (minValList, maxValList) = (None, None) if colorscaleData != None: (minValList, maxValList) = getColorScaleMinMaxValues(minValList, maxValList, len(circleData), colorscaleData) for feature in features: log("Drawing %s\n" % (feature)) centerColHex = None if centerData != None: if feature in centerData: centerColHex = getColor(centerData[feature], centerDataMinVal, centerDataMaxVal, minColor=lightBlueRGB, zeroColor=whiteRGB, purple0Hack=purpleHack) imgFile = "%s/%s.png" % (outputDir, re.sub("[/:]", "_", feature)) label = "" if printLabel: label = feature image_width = 5.0 drawCircleImageForFeature(feature, samples, label, imgFile, circleData, circleColorsPalette, width=image_width, centerColHex=centerColHex, minValList=minValList, maxValList=maxValList, purple0Hack=purpleHack) for sample in samples: log("ordered samples: %s\n" % (sample))
def drawCircleImageForFeature(feature, samples, label, imgFile, circleData, circleColors, centerColHex=None, width=5, minValList=None, maxValList=None, purple0Hack=False): """Draw a circle map image and write it to a file.""" # feature - feature to draw image for. This is some kind of concept: for example, a gene. # samples - sample names of data # label - label to use in image # imgFile - file object to which image will be written # circleData - data struct containing sample data for features. It is a list of dict[col][row]=score . # circleColors - a list of (minColor),(zeroColor),(maxColor) # centerColHex - hex code for center color fill. If none, then make transparent center. # centerCol is the color of the center of the circleImage # centerCol = whiteRGB.tohex() # circleCols is a list. Each member of the list represents a list of colors in a ring. circleCols = [] # iterate through rings of data for ring in xrange(len(circleData)): ringCols = [] # get minVal and maxVal minVal = None maxVal = None if minValList == None or maxValList == None or minValList[ring] == None or maxValList[ring] == None: ringVals = [] # get ring values in effort to find min/max values for each *ring* for sample in samples: if sample in circleData[ring]: if feature in circleData[ring][sample]: ringVals.append(circleData[ring][sample][feature]) elif "*" in circleData[ring][sample]: ringVals.append(circleData[ring][sample]["*"]) # find the min & max sample scores for this ring in this feature floatList = mData.floatList(ringVals) minVal = min([-0.01] + floatList) maxVal = max([0.01] + floatList) else: minVal = minValList[ring] maxVal = maxValList[ring] # convert scores into colors for sample in samples: if sample in circleData[ring]: if feature in circleData[ring][sample]: ringCols.append(getColor(circleData[ring][sample][feature], minVal, maxVal, minColor=circleColors[ring][0], zeroColor=circleColors[ring][1], maxColor=circleColors[ring][2], purple0Hack=purple0Hack)) elif "*" in circleData[ring][sample]: ringCols.append(getColor(circleData[ring][sample]["*"], minVal, maxVal, minColor=circleColors[ring][0], zeroColor=circleColors[ring][1], maxColor=circleColors[ring][2], purple0Hack=purple0Hack)) else: # sample exists, but no score for the feature ringCols.append(greyRGB.tohex()) else: # this sample not found in the sample data ringCols.append(greyRGB.tohex()) # add the ring circleCols.append(ringCols) # plot the image plotCircle(imgFile, label=label, centerColHex=centerColHex, circleCols=circleCols, innerRadTotal=0.2, outerRadTotal=0.5, width=width)
def cli_routine(outputDir,
circleFiles,
orderFiles,
sampleFile,
featureFile,
orderFeature,
centerFile,
colorscaleFile,
printLabel,
verbose,
cohortMinMax=False,
purpleHack=True):
"""Routine for program execution via command-line."""
# I've tried not to touch this method as much as possible.
# I don't want to break the way it was working for Sam Ng.
# chrisw
## execute
samples = []
features = []
if sampleFile != None:
samples = mData.rList(sampleFile)
if featureFile != None:
features = mData.rList(featureFile)
# end section for getting lists of samples and features
## read circleFiles
# circleData is a list of dict[col][row]=score from each circleFile
circleData = []
# circleColorsPalette is a list of (minColor),(zeroColor),(maxColor)
circleColorsPalette = []
## read colorscaleFile
# the format is as follows - header compulsory:
# min/max color coding color1 color2 color 3
# -2,2 rgb 155,155,155 255,255,255 0,0,0,
# - rgb 155,0,155 255,0,255 0,0,0,
# the "color format" is intended to support more color format, as I have
# seen the html-colors in the code.
# Michael ([email protected])
colorscaleData = None
if colorscaleFile != None:
if cohortMinMax:
log("WARNING: The -k option overrides -m")
colorscaleData = mData.retRows(colorscaleFile, aslist=True)
line = 1
for cs in colorscaleData:
line = line + 1
if len(cs) != 5:
log("ERROR: color scale needs five fields: datapoints, colorcoding(rgb) and three colors\n",
die=True)
try:
cs[0] = [float(x) for x in cs[0].split(",")]
except ValueError:
pass
if len(cs[0]) != 2 and cs[0] != "-":
print cs[0]
log("ERROR: Two data points or dash needed for color scale\n",
die=True)
if cs[1].lower() == "rgb":
try:
cs[2] = rgb(*[float(x) for x in cs[2].split(",")])
cs[3] = rgb(*[float(x) for x in cs[3].split(",")])
cs[4] = rgb(*[float(x) for x in cs[4].split(",")])
except TypeError:
log("ERROR: RGB needs three values on line " + str(line) +
"\n",
die=True)
except ValueError:
log("ERROR: RGB color not correctly defined on line " +
str(line) + "\n",
die=True)
else:
log("ERROR: Unknown color coding on line " + str(line) + ": " +
str(cs[1]) + "\n",
die=True)
for i in xrange(len(circleFiles)):
# get data, samples, and features from each circleFile
# data is a dict[col][row]=score
# cols is a list of sample names
# features is a list of feature names
(data, cols, rows) = mData.rCRSData(circleFiles[i], retFeatures=True)
circleData.append(data)
minCol = lightBlueRGB
zerCol = whiteRGB
maxCol = redRGB
if colorscaleFile != None and i < len(colorscaleData):
#get colors from specified colorscaleFile
minCol = colorscaleData[i][2]
zerCol = colorscaleData[i][3]
maxCol = colorscaleData[i][4]
# special cases for -meth and -mut
# if circleFiles[i].endswith("meth"):
# maxCol = blueRGB
# minCol = redRGB
# log("Color: meth\n")
# elif circleFiles[i].endswith("mut"):
# maxCol = blackRGB
# minCol = whiteRGB
# log("Color: mut\n")
circleColorsPalette.append((minCol, zerCol, maxCol))
# if no sampleFile/featureFile, default to using samples/features from circleFiles
if sampleFile == None:
samples = list(set(cols) | set(samples))
if featureFile == None:
features = list(set(rows) | set(features))
# end section for reading circleFiles
## read centerFile
centerData = None
if centerFile != None:
centerData = mData.r2Col(centerFile, header=True)
## sort
if orderFeature != None:
if len(orderFiles) > 0:
orderData = []
orderColors = []
for i in xrange(len(orderFiles)):
orderData.append(mData.rCRSData(orderFiles[i]))
minCol = whiteRGB
zerCol = whiteRGB
maxCol = blackRGB
orderColors.append((minCol, zerCol, maxCol))
else:
orderData = circleData
# sort samples based on sample score in orderData
# priority of sorting determined by orderFiles parameter
samples.sort(lambda x, y: scmp(x, y, orderFeature, orderData))
## cohort png
# cgi will probably not use orderFiles
if len(orderFiles) > 0:
imgFile = "%s/Cohort.png" % (outputDir)
label = "Cohort"
centerCol = whiteRGB.tohex()
cohortCircleCols = []
for i in xrange(len(orderData)):
ringCols = []
ringVals = []
for sample in samples:
if sample in orderData[i]:
if orderFeature in orderData[i][sample]:
ringVals.append(orderData[i][sample][orderFeature])
elif "*" in orderData[i][sample]:
ringVals.append(orderData[i][sample]["*"])
minVal = min([-0.01] + mData.floatList(ringVals))
maxVal = max([0.01] + mData.floatList(ringVals))
for sample in samples:
if sample in orderData[i]:
if orderFeature in orderData[i][sample]:
ringCols.append(
getColor(orderData[i][sample][orderFeature],
minVal,
maxVal,
minColor=orderColors[i][0],
zeroColor=orderColors[i][1],
maxColor=orderColors[i][2]))
elif "*" in orderData[i][sample]:
ringCols.append(
getColor(orderData[i][sample]["*"],
minVal,
maxVal,
minColor=orderColors[i][0],
zeroColor=orderColors[i][1],
maxColor=orderColors[i][2]))
else:
ringCols.append(greyRGB.tohex())
else:
ringCols.append(greyRGB.tohex())
cohortCircleCols.append(ringCols)
plotCircle(imgFile,
label=label,
centerCol=centerCol,
circleCols=cohortCircleCols,
innerRadTotal=0.2,
outerRadTotal=0.5,
width=5)
# end section for sample ordering
## plot images
if centerData != None:
centerDataFloatList = mData.floatList(centerData.values())
centerDataMinVal = min([-0.01] + centerDataFloatList)
centerDataMaxVal = max([0.01] + centerDataFloatList)
# get min/max values for datasets
if cohortMinMax:
(minValList,
maxValList) = getCohortMinMaxValues(features, samples, circleData)
else:
(minValList, maxValList) = (None, None)
if colorscaleData != None:
(minValList,
maxValList) = getColorScaleMinMaxValues(minValList, maxValList,
len(circleData),
colorscaleData)
for feature in features:
log("Drawing %s\n" % (feature))
centerColHex = None
if centerData != None:
if feature in centerData:
centerColHex = getColor(centerData[feature],
centerDataMinVal,
centerDataMaxVal,
minColor=lightBlueRGB,
zeroColor=whiteRGB,
purple0Hack=purpleHack)
imgFile = "%s/%s.png" % (outputDir, re.sub("[/:]", "_", feature))
label = ""
if printLabel:
label = feature
image_width = 5.0
drawCircleImageForFeature(feature,
samples,
label,
imgFile,
circleData,
circleColorsPalette,
width=image_width,
centerColHex=centerColHex,
minValList=minValList,
maxValList=maxValList,
purple0Hack=purpleHack)
for sample in samples:
log("ordered samples: %s\n" % (sample))
def drawCircleImageForFeature(feature,
samples,
label,
imgFile,
circleData,
circleColors,
centerColHex=None,
width=5,
minValList=None,
maxValList=None,
purple0Hack=False):
"""Draw a circle map image and write it to a file."""
# feature - feature to draw image for. This is some kind of concept: for example, a gene.
# samples - sample names of data
# label - label to use in image
# imgFile - file object to which image will be written
# circleData - data struct containing sample data for features. It is a list of dict[col][row]=score .
# circleColors - a list of (minColor),(zeroColor),(maxColor)
# centerColHex - hex code for center color fill. If none, then make transparent center.
# centerCol is the color of the center of the circleImage
# centerCol = whiteRGB.tohex()
# circleCols is a list. Each member of the list represents a list of colors in a ring.
circleCols = []
# iterate through rings of data
for ring in xrange(len(circleData)):
ringCols = []
# get minVal and maxVal
minVal = None
maxVal = None
if minValList == None or maxValList == None or minValList[
ring] == None or maxValList[ring] == None:
ringVals = []
# get ring values in effort to find min/max values for each *ring*
for sample in samples:
if sample in circleData[ring]:
if feature in circleData[ring][sample]:
ringVals.append(circleData[ring][sample][feature])
elif "*" in circleData[ring][sample]:
ringVals.append(circleData[ring][sample]["*"])
# find the min & max sample scores for this ring in this feature
floatList = mData.floatList(ringVals)
minVal = min([-0.01] + floatList)
maxVal = max([0.01] + floatList)
else:
minVal = minValList[ring]
maxVal = maxValList[ring]
# convert scores into colors
for sample in samples:
if sample in circleData[ring]:
if feature in circleData[ring][sample]:
ringCols.append(
getColor(circleData[ring][sample][feature],
minVal,
maxVal,
minColor=circleColors[ring][0],
zeroColor=circleColors[ring][1],
maxColor=circleColors[ring][2],
purple0Hack=purple0Hack))
elif "*" in circleData[ring][sample]:
ringCols.append(
getColor(circleData[ring][sample]["*"],
minVal,
maxVal,
minColor=circleColors[ring][0],
zeroColor=circleColors[ring][1],
maxColor=circleColors[ring][2],
purple0Hack=purple0Hack))
else:
# sample exists, but no score for the feature
ringCols.append(greyRGB.tohex())
else:
# this sample not found in the sample data
ringCols.append(greyRGB.tohex())
# add the ring
circleCols.append(ringCols)
# plot the image
plotCircle(imgFile,
label=label,
centerColHex=centerColHex,
circleCols=circleCols,
innerRadTotal=0.2,
outerRadTotal=0.5,
width=width)