def ReadSeqs3(FileName, OnSeq, ShowProgress=True):
File = open(FileName)
if ShowProgress:
progress.FileInit(File, FileName)
Label = ""
Seq = ""
n = 0
while 1:
Line = File.readline()
if len(Line) == 0:
if Seq != "":
n += 1
OnSeq(Label, Seq)
if ShowProgress:
progress.FileDone()
return
Line = Line.strip()
if len(Line) == 0:
continue
if Line[0] == ">":
if Seq != "":
if ShowProgress and n % 100 == 0:
progress.FileStep()
if TRUNC_LABELS:
Label = Label.split()[0]
OnSeq(Label, Seq)
Label = Line[1:]
Seq = ""
else:
Seq += Line
def ReadSeqs2(FileName, ShowProgress=True):
Seqs = []
Labels = []
File = open(FileName)
if ShowProgress:
progress.FileInit(File, FileName)
while 1:
if len(Labels) % 100 == 0:
progress.FileStep()
Line = File.readline()
if len(Line) == 0:
if ShowProgress:
progress.FileDone()
return Labels, Seqs
Line = Line.strip()
if len(Line) == 0:
continue
if Line[0] == ">":
Id = Line[1:]
if TRUNC_LABELS:
Id = Id.split()[0]
Labels.append(Id)
Seqs.append("")
else:
i = len(Seqs) - 1
Seqs[i] = Seqs[i] + Line
PctIdToCount = {}
PctIdToLCRCount = {}
def Add(LCR, PctId):
sortdict.IncCount(LCRToCount, LCR)
sortdict.IncCount(PctIdToCount, PctId)
sortdict.IncCount2(PctIdToLCRCount, PctId, LCR)
MinPctId = 100
File = open(FileName)
progress.FileInit(File, FileName)
while 1:
Line = File.readline()
progress.FileStep()
if len(Line) == 0:
progress.FileDone()
break
Fields = Line.strip().split('\t')
assert len(Fields) == 3
Label1 = Fields[0]
Label2 = Fields[1]
Dist = float(Fields[2])
assert Dist >= 0.0 and Dist <= 1.0
if Label1 == Label2:
continue
LCR = utax2.GetLCRFromLabels(Label1, Label2)
PctId = int(100.0 * (1.0 - Dist))
if PctId < MinPctId: