def makegcmat(fastafile,sizef,step,achr,savename): gcdict=gccontentforonechr(fastafile,sizef,step) gcvec=gcdict[achr] L=len(gcvec) mat=np.zeros((L,L)) i=0 j=0 while i<L: while j<L: val=(float(gcvec[i])+float(gcvec[j]))/2 mat[i,j]=val mat[j,i]=val j+=1 i+=1 j=i i=0 vec=[0]*L while i<L: vec[i]+=np.mean(np.diag(mat,k=i)) i+=1 utils.savealist(vec,savename+"gcfd")
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