def generateHicListofpattern(filename,sizeList,resolution):
namedump=filename+"ListedList.dump" #remember: finelame include the chr
if op.exists(namedump):
print("===> Set deja existants")
setlist=pickle.load(open(namedump,"rb"))
else:
filein=open(filename,"r")
setlist=generateemptylistoflist(sizeList)
l=filein.readline()
l=filein.readline()
i=0
while l:
ls=l.split()
begin=(np.float(ls[4])/resolution)
end=(np.float(ls[5])/resolution)
i=0
#print(TElist[0],TElist[1])
if ls[3]!="Simple_repeat":
while i<np.ceil(end-begin):
#print(begin,i,end,ls[3])
#print((end-begin),ls[1],ls[0],ls[4],ls[5],TElist.index(ls[1]),begin+i,chrbegin)
setlist[int(begin+i)].append(ls[3]) #cast is only for list indexation #1 name, 3 family_name
i+=1
else:
while i<np.ceil(end-begin):
falsefamilyname=simplerepeattype(ls[1])
setlist[int(begin+i)].append(falsefamilyname)
i+=1
l=filein.readline()
filein.close()
print("====>Dumping de la liste des sets")
pickle.dump(setlist,open(namedump,"wb"))
print("====> Sauvegarge de la liste des sets en texte")
utils.savelistofset(setlist,filename+"SetList.txt")
return setlist
def generatejumplistofset(filename,windowsize):
namedump=filename+"SetList.dump"
namedumpsizelist=filename+"Sizelist.dump"
if op.exists(namedump):
print("===> Set deja existants")
setlist=pickle.load(open(namedump,"rb"))
sizelist=pickle.load(open(namedumpsizelist,"rb"))
else:
filein=open(filename,"r")
setlist=list()
sizelist=list()
l=filein.readline()
l=filein.readline()
i=0
theset=set() #init
while l:
te=l.split()[1]
Ti=re.match("[A-Z]+-rich",te)
Tii=re.match("[A-Z]+_rich",te)
if not Ti and not Tii:
if i>=windowsize:
i=0
setlist.append(theset)
sizelist.append(len(theset))
theset=set()
theset.add(te)
i+=1
l=filein.readline()
filein.close()
print("====>Dumping de la liste des sets")
pickle.dump(setlist,open(namedump,"wb"))
pickle.dump(setlist,open(namedumpsizelist,"wb"))
print("====> Sauvegarge de la liste des sets en texte")
utils.savelistofset(setlist,filename+"SetList.txt")
return setlist,sizelist
def generatedoublelistofset(filename,windowsize):
namedump=filename+"ListedList.dump" #remember: finelame include the chr
namedumpsizelist=filename+"ListedSizeList.dump"
namedistancehist=filename+"DistanceofTElist.txt"
if op.exists(namedump):
print("===> Set deja existants")
setlist=pickle.load(open(namedump,"rb"))
sizelist=pickle.load(open(namedumpsizelist,"rb"))
else:
filein=open(filename,"r")
setlist=list()
sizelist=list()
distlist=list()
l=filein.readline()
l=filein.readline()
i=0
tampolist=list() #init
while l:
te=l.split()[1] #1 name 3 family
if i>=windowsize: #most of the case
dend=float(l.split()[5])
i=0
#theset=set(tampolist)
setlist.append(tampolist)
sizelist.append(i)
distlist.append(dend-dinit)
tampolist=list() #back to no one
tampolist.append(te)
if i==0:
dinit=float(l.split()[4])
i+=1
l=filein.readline()
filein.close()
print("====>Dumping de la liste des sets")
pickle.dump(setlist,open(namedump,"wb"))
pickle.dump(sizelist,open(namedumpsizelist,"wb"))
utils.savealist(distlist,namedistancehist)
print("====> Sauvegarge de la liste des sets en texte")
utils.savelistofset(setlist,filename+"SetList.txt")
return setlist,sizelist