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'
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]
return Tax + RankName + ":" + Name[2:] + ";"
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)
fasta.ReadSeqsOnSeq(InputFileName, OnSeq)