def OnSeq(Label, Seq):
Acc = Label.split(";")[0]
Tax = utax2.GetTaxFromLabel(Label)
if KeepTaxAnnot:
Acc = Label
Chars, Names = utax2.TaxToVecs(Tax)
N = len(Chars)
assert len(Names) == N
N = len(Chars)
OutLine = Acc + "\t"
for i in range(0, N):
Char = Chars[i]
Name = Names[i]
OutLine += Char + "__" + Name
if i + 1 < N:
OutLine += "; "
print >> fTax, OutLine
if KeepTaxAnnot:
fasta.WriteSeq(fFa, Seq, Label)
else:
fasta.WriteSeq(fFa, Seq, Acc)
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 OnSeq(Label, Seq):
global MissingRankCount
global EmptyTaxCount
global BadLevelCount
global SeqCount
SeqCount += 1
Acc = utax2.GetAccFromLabel(Label)
Acc = Acc.replace("_", "")
TaxStr = utax2.GetTaxFromLabel(Label)
Fields = TaxStr.split(',')
if len(Fields) != RequiredLevelCount:
BadLevelCount += 1
return
MotTaxStr = ""
LastRankIndex = -1
LastLevelChar = ""
for Field in Fields:
assert Field[1] == ':'
LevelChar = Field[0]
RankIndex = utax2.LevelCharToRankIndex(LevelChar)
if LastRankIndex != -1 and LastRankIndex > 1 and not RankIndex == LastRankIndex + 1:
MissingRankCount += 1
# print >> sys.stderr, "Missing rank: %c,%d %c,%d %s" % (LastLevelChar, LastRankIndex, LevelChar, RankIndex, TaxStr)
break
LastRankIndex = RankIndex
LastLevelChar = LevelChar
if MotTaxStr != "":
MotTaxStr += ';'
MotTaxStr += LevelChar + ":" + Field[2:]
if MotTaxStr == "":
EmptyTaxCount += 1
return
MotTaxStr += ";"
fasta.WriteSeq(fFa, Seq, Acc)
print >> fTax, Acc + "\t" + MotTaxStr
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)