def prepareConverter(self, fileIn):
FileUtility.isValid(fileIn)
sys.stderr.write("counting lines to prepare converter\n")
numOfLines = FileUtility.countLines(fileIn)
sys.stderr.write(str(numOfLines)+" to read\n")
readed = 0
for line in open(fileIn):
genome = int(line.split()[0])
taxon = int(line.split()[1])
self.Converter[genome] = taxon
readed += 1
if readed%100000 == 0:
sys.stderr.write(str(readed)+" lines readed out of "+str(numOfLines)+"\n")
def populate(self, file_name1, taxon_name):
# Create table...
curs = self.cursor
curs.execute("create table GENOME_TO_TAXON (GENOMES integer, TAXON integer, PRIMARY KEY (GENOMES))")
self.converter.commit()
# Check is file is ok
FileUtility.isValid(file_name1)
sys.stderr.write("counting lines to prepare converter\n")
n_lines = FileUtility.countLines(file_name1)
sys.stderr.write(str(n_lines)+" to read\n")
readed = 0
# populate the DB
for line in open(file_name1):
genome = int(line.split()[0])
taxon = int(line.split()[1])
curs.execute('insert into GENOME_TO_TAXON (GENOMES, TAXON) values (?, ?)', (genome, taxon))
readed += 1
if readed % 10000 == 0:
self.converter.commit()
sys.stderr.write(str(readed)+" lines readed out of "+str(n_lines)+"\n")
FileUtility.isValid(taxon_name)
sys.stderr.write("counting lines to prepare converter\n")
n_lines = FileUtility.countLines(taxon_name)
sys.stderr.write(str(n_lines)+" to read\n")
readed = 0
curs.execute("create table TAXON_NAMES (TAXON integer, NAME VARCHAR(100), RANK VARCHAR(30), "
"PRIMARY KEY (TAXON))")
for line in open(taxon_name):
taxon = int(line.split("\t")[0])
name = str(line.split("\t")[1])
rank = str(line.split("\t")[2]).rstrip("\n")
curs.execute('insert into TAXON_NAMES (TAXON, NAME, RANK) values (?, ?, ?)', (taxon, name, rank))
readed += 1
if readed % 10000 == 0:
self.converter.commit()
sys.stderr.write(str(readed)+" lines readed out of "+str(n_lines)+"\n")
def prepareConverter(self, fileIn):
FileUtility.isValid(fileIn)
sys.stderr.write("counting lines to prepare converter\n")
numOfLines = FileUtility.countLines(fileIn)
sys.stderr.write(str(numOfLines) + " to read\n")
readed = 0
for line in open(fileIn):
genome = int(line.split()[0])
taxon = int(line.split()[1])
self.Converter[genome] = taxon
readed += 1
if readed % 100000 == 0:
sys.stderr.write(
str(readed) + " lines readed out of " + str(numOfLines) +
"\n")
def read_tree_of_life(self, file_in):
FileUtility.isValid(file_in)
to_read = FileUtility.countLines(file_in)
sys.stderr.write(str(to_read)+" lines to read to construct tree\n")
num_line = 0
for line in open(file_in):
parent = int(line.split()[0])
kid = int(line.split()[1])
if kid == parent:
sys.stderr.write("Warning: I can't create a link from %s to %s (line %s)\n" % (parent, kid, num_line))
continue
if self.node_exist(parent):
if self.node_exist(kid):
self.get_node(parent).add_kid(kid)
self.get_node(kid).set_parent(parent)
else:
new_node = Taxon(kid)
self.create_node(new_node)
self.get_node(parent).add_kid(kid)
self.get_node(kid).set_parent(parent)
else:
if self.node_exist(kid):
new_node = Taxon(parent)
self.create_node(new_node)
self.get_node(kid).set_parent(parent)
self.get_node(parent).add_kid(kid)
else:
new_kid = Taxon(kid)
new_parent = Taxon(parent)
new_kid.set_parent(parent)
new_parent.add_kid(kid)
self.create_node(new_kid)
self.create_node(new_parent)
num_line += 1
if num_line % 100000 == 0:
sys.stderr.write(str(num_line)+" lines read\n")
sys.stderr.write("Tree base constructed\n")
