))
worker.daemon = True
worker.start()
'''start background Thread that will run a loop to check run statistics and print
We use thread, because I think this is necessary for a process that watches global variables like linesTested'''
worker = Thread(target=checkStats)
worker.daemon = True
worker.start()
##########################################################
#get windows and analyse
if windType == "coordinate":
windowGenerator = genomics.slidingCoordWindows(genoFile,
windSize,
stepSize,
sampleData.indNames,
include=scafsToInclude,
exclude=scafsToExclude)
elif windType == "sites":
windowGenerator = genomics.slidingSitesWindows(genoFile,
windSize,
overlap,
maxDist,
minSites,
sampleData.indNames,
include=scafsToInclude,
exclude=scafsToExclude)
else:
windowGenerator = genomics.predefinedCoordWindows(genoFile, windCoords,
sampleData.indNames)
if include:
scafsFile = open(include, "rU")
scafsToInclude = [line.rstrip() for line in scafsFile.readlines()]
print >> sys.stderr, len(scafsToInclude), "scaffolds will be analysed."
scafsFile.close()
else:
scafsToInclude = None
##########################################################################################################
#get windows and analyse
if windType == "coordinate":
windowGenerator = genomics.slidingCoordWindows(inFile,
windSize,
stepSize,
headerLine=args.header,
names=args.columns,
include=scafsToInclude,
exclude=scafsToExclude)
elif windType == "sites":
windowGenerator = genomics.slidingSitesWindows(inFile,
windSize,
overlap,
maxDist,
minSites,
headerLine=args.header,
names=args.columns,
include=scafsToInclude,
exclude=scafsToExclude)
else:
windowGenerator = genomics.predefinedCoordWindows(inFile,
if include:
scafsFile = open(include, "rU")
scafsToInclude = [line.rstrip() for line in scafsFile.readlines()]
print >> sys.stderr, len(scafsToInclude), "scaffolds will be analysed."
scafsFile.close()
else:
scafsToInclude = None
##########################################################################################################
#get windows and analyse
if windType == "coordinate":
windowGenerator = genomics.slidingCoordWindows(inFile,
windSize,
stepSize,
columns,
include=scafsToInclude,
exclude=scafsToExclude)
elif windType == "sites":
windowGenerator = genomics.slidingSitesWindows(inFile,
windSize,
overlap,
maxDist,
minSites,
columns,
include=scafsToInclude,
exclude=scafsToExclude)
else:
windowGenerator = genomics.predefinedCoordWindows(inFile, windCoords,
columns)
##########################################################
if args.windType == "cat":
window = genomics.parseGenoFile(genoFile, headerLine = args.header, names=sampleData.indNames)
windowQueue.put((windowsQueued,window))
windowsQueued += 1
else:
#get windows and analyse
if args.windType == "coordinate": windowGenerator = genomics.slidingCoordWindows(genoFile, windSize, stepSize,
headerLine = args.header,
names = sampleData.indNames,
include = scafsToInclude,
exclude = scafsToExclude)
elif args.windType == "sites": windowGenerator = genomics.slidingSitesWindows(genoFile, windSize, overlap,
maxDist, minSites,
headerLine = args.header,
names = sampleData.indNames,
include = scafsToInclude,
exclude = scafsToExclude)
else: windowGenerator = genomics.predefinedCoordWindows(genoFile, windCoords,
headerLine = args.header,
names = sampleData.indNames)
for window in windowGenerator:
windowQueue.put((windowsQueued,window))
windowsQueued += 1
windSize = args.windSize
minSites = args.minSites
stepSize = args.stepSize
overlap = args.overlap
maxDist = args.maxDist
else:
#to get contigs, we just use very lare non-overlapping coordinate windows
windType = "coordinate"
windSize = 1e7
minSites = 1
stepSize = 1e7
#initialise window generator
if windType == "coordinate":
windowGenerator = genomics.slidingCoordWindows(
genoFile,
windSize,
stepSize,
args.samples,
splitPhased=args.splitPhased)
else:
windowGenerator = genomics.slidingSitesWindows(
genoFile,
windSize,
overlap,
maxDist,
minSites,
args.samples,
splitPhased=args.splitPhased)
for window in windowGenerator:
posString = str(window.firstPos()) + "_" + str(window.lastPos())
scafsToExclude = None
if include:
scafsFile = open(include, "rU")
scafsToInclude = [line.rstrip() for line in scafsFile.readlines()]
print >> sys.stderr, len(scafsToInclude), "scaffolds will be analysed."
scafsFile.close()
else:
scafsToInclude = None
##########################################################################################################
#get windows and analyse
if windType == "coordinate": windowGenerator = genomics.slidingCoordWindows(inFile, windSize, stepSize,
args.columns,
include = scafsToInclude,
exclude = scafsToExclude)
elif windType == "sites": windowGenerator = genomics.slidingSitesWindows(inFile, windSize, overlap,
maxDist, minSites, args.columns,
include = scafsToInclude,
exclude = scafsToExclude)
else: windowGenerator = genomics.predefinedCoordWindows(inFile, windCoords, args.columns)
n=0
for window in windowGenerator:
#if its the first window, get the headings and write
if n == 0:
for name in window.names:
outFile.write(","+",".join([name + "_mean",name + "_median", name + "_max", name + "_min", name + "_sd", name + "_sum"]))
