def OnSeq(Label, Seq):
global Missing
Acc = utax2.GetAccFromLabel(Label)
Family = utax2.GetNameFromLabel(Label, 'f')
Genus = utax2.GetNameFromLabel(Label, 'g')
Species = utax2.GetNameFromLabel(Label, 's')
if Genus == "" or Species == "":
Missing += 1
return
if Family == "":
Family = "NA"
Family = Family.replace("f:", "")
Genus = Genus.replace("g:", "")
Species = Species.replace("s:", "")
NewLabel = Acc + "\t" + Species + "\t" + Genus + "\t" + Family
fasta.WriteSeq(sys.stdout, Seq, NewLabel)
def DoPredFile(FileName, NameToCountA, NameToCountB):
f = open(FileName)
TP = 0
TN = 0
FN = 0
OC = 0
MC = 0
NC = 0
Known = 0
Novel = 0
while 1:
Line = f.readline()
if len(Line) == 0:
return Known, Novel, TP, TN, FN, OC, MC, NC
Fields = Line[:-1].split('\t')
assert len(Fields) == 2
QueryLabel = Fields[0]
Pred = Fields[1]
if Pred.endswith(';'):
Pred = Pred[:-1]
QueryName = utax2.GetNameFromLabel(QueryLabel, Rank)
assert QueryName != ""
if Pred.find("tax=") >= 0:
Pred = Pred.split("tax=")[1]
if Pred == "*":
PredName = ""
else:
PredName = utax2.GetNameFromTaxStr(Pred, Rank)
if PredName != "":
NC += 1
Count = sortdict.GetCount(NameToCountB, QueryName)
IsKnown = (Count > 0)
if IsKnown:
Known += 1
else:
Novel += 1
if PredName == QueryName and not IsKnown:
die.Die("QueryName=%s, PredName=%s >%s" %
(QueryName, PredName, QueryLabel))
if PredName == QueryName:
XX = "TP"
TP += 1
elif PredName == "":
if Count == 0:
XX = "TN"
TN += 1
else:
XX = "FN"
FN += 1
else:
if Count == 0:
XX = "OC"
OC += 1
else:
XX = "MC"
MC += 1
if REPORT:
Acc = fasta.GetAccFromLabel(QueryLabel)
if IsKnown:
k = "known"
else:
k = "novel"
PredNameStr = "-"
if PredName != "":
PredNameStr = PredName
s = Acc
s += "\t" + XX
s += "\t" + k
s += "\t" + QueryName
s += "\t" + PredNameStr
s += "\t" + str(Count)
print s