def _loadNodes(self, dmpDir):
self.nodes = dict()
self.parentNodeRefs = MultiDict()
fh = dmpFileParser(dmpDir + "/nodes.dmp")
for row in fh:
node = Node(row)
self.nodes[node.taxId] = node
self.parentNodeRefs.add(node.parentTaxId, node)
def _loadNames(self, dmpDir):
self.names = MultiDict()
self.sciNames = dict()
sciName = "scientific name"
fh = dmpFileParser(dmpDir + "/names.dmp")
for row in fh:
name = Name(row)
self.names[name.taxId] = name
if name.nameClass == sciName:
self.sciNames[name.nameTxt] = name.taxId
def __buildIdx(self):
from pycbio.sys.MultiDict import MultiDict
self.names = MultiDict()
for row in self:
self.names.add(row.name, row)
def _mkNameIdx(self):
self.nameMap = MultiDict()
for bed in self:
self.nameMap.add(bed.name, bed)
def __mkTNameIdx(self):
self.tNameMap = MultiDict()
for psl in self:
self.tNameMap.add(psl.tName, psl)
