def OnSeq(Label, Seq):
Acc = fasta.GetAccFromLabel(Label)
TaxStr = utax2.GetTaxFromLabel(Label)
k = 'k'
if TaxStr.find("d:") >= 0:
k = 'd'
# NR_117221.1 species:Mycobacterium arosiense;genus:Mycobacterium;family:Mycobacteriaceae;order:Corynebacteriales;class:Actinobacteria;phylum:Actinobacteria;superkingdom:Bacteria;
Tax = ""
if TaxStr.find(",s:") > 0:
Tax = AppendRank(Tax, TaxStr, 's')
Tax = AppendRank(Tax, TaxStr, 'g')
Tax = AppendRank(Tax, TaxStr, 'f')
Tax = AppendRank(Tax, TaxStr, 'o')
Tax = AppendRank(Tax, TaxStr, 'c')
Tax = AppendRank(Tax, TaxStr, 'p')
Tax = AppendRank(Tax, TaxStr, k)
Acc = Acc.upper()
NewLabel = "REF_" + Acc
fasta.WriteSeq(fFa, Seq, NewLabel)
print >> fTax, "%s\t%s" % (NewLabel, Tax)
def FixQueryLabel(Label):
Acc = fasta.GetAccFromLabel(Label)
Tax = fasta.GetTaxFromLabel(Label)
s = Acc + ";tax="
Fields = Tax.split(',')
n = len(Fields)
for i in range(0, n):
Field = Fields[i]
if i > 0:
s += ","
s += Field[0] + ":" + Field[2:]
s += ";"
return s
def OnSeq(Label, Seq):
global AccToTax
global DomainRank
Acc = fasta.GetAccFromLabel(Label)
Tax = fasta.GetTaxFromLabel(Label)
Acc = Acc.upper()
if Acc.startswith("REF_"):
Acc = Acc.replace("REF_", "")
AccToTax[Acc] = Tax
if DomainRank == '?':
DomainRank = Tax[0]
def OnSeq(Label, Seq):
global AccToTax
global DomainRank
global Ranks
Acc = fasta.GetAccFromLabel(Label)
Tax = fasta.GetTaxFromLabel(Label)
Acc = Acc.upper()
AccToTax[Acc] = Tax
if DomainRank == '?':
DomainRank = Tax[0]
Ranks = DomainRank + Ranks[1:]
assert Ranks[1] == 'p'
def OnSeq(Label, Seq):
Acc = fasta.GetAccFromLabel(Label)
TaxStr = utax2.GetTaxFromLabel(Label)
k = 'k'
if TaxStr.find("d:") >= 0:
k = 'd'
# DQ200983.1.1404.B Bacteria;Actinobacteria;Actinobacteria;Frankiales;Geodermatophilaceae;Blastococcus;Blastococcus jejuensis;
Tax = ""
Tax = AppendRank(Tax, TaxStr, k)
Tax = AppendRank(Tax, TaxStr, 'p')
Tax = AppendRank(Tax, TaxStr, 'c')
Tax = AppendRank(Tax, TaxStr, 'o')
Tax = AppendRank(Tax, TaxStr, 'f')
Tax = AppendRank(Tax, TaxStr, 'g')
Tax = AppendRank(Tax, TaxStr, 's')
Acc = Acc.upper()
fasta.WriteSeq(fFa, Seq, Acc)
print >> fTax, "%s\t%s" % (Acc, Tax)
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
assert Ranks[1] == 'p'
fasta.ReadSeqsOnSeq(FastaFileName, OnSeq)
f = open(FileName)
while 1:
Line = f.readline()
if len(Line) == 0:
break
# gi_1018196556 Bacteria;Proteobacteria;Gammaproteobacteria;Oceanospirillales;Halomonadaceae; 96.58 292 91.03
Fields = Line[:-1].split('\t')
Label = Fields[0]
Acc = fasta.GetAccFromLabel(Label)
Acc = Acc.upper()
QTax = AccToTax[Acc]
Tax = Fields[1]
Fields2 = Tax.split(";")
n = len(Fields2) - 1
assert n <= len(Ranks)
Pred = ""
for i in range(0, n):
if i > 0:
Pred += ","
Rank = Ranks[i]
Pred += Rank + ":" + Fields2[i]
