class SBOLOutput(SuperGSLFunction):
"""Generate SBOL document containing the assemblies."""
name = 'sbol'
def get_arguments(self):
return [('filename', str), ('assemblies', list)]
def get_return_type(self):
return type(None)
def sanitize_identifier(self, identifier):
"""Sanitize SuperGSL Identifiers to conform to identifiers in the SBOL spec."""
bad_chars = '[]~:'
for c in bad_chars:
identifier = identifier.replace(c, '_')
return identifier
def handle_assembly(self, assembly):
"""Add each assembly to the SBOL Document."""
self.assembly_count += 1
label = node.label
if not label:
label = 'Assembly%05d' % self.assembly_count
assembly = ComponentDefinition(label)
self.sbol_doc.addComponentDefinition(assembly)
part_components = []
for part_node in node.parts:
part = part_node.part
sanitized_ident = self.sanitize_identifier(part.identifier)
part_component = ComponentDefinition(sanitized_ident)
part_component.roles = part.roles
part_component.sequence = Sequence(sanitized_ident,
str(part.get_sequence().seq))
part_components.append(part_component)
assembly.assemblePrimaryStructure(part_components)
assembly.compile()
def execute(self, params: dict):
"""Initialize the SBOL Document."""
setHomespace('http://sbols.org/SuperGSL_Example/')
Config.setOption('sbol_compliant_uris', True)
Config.setOption('sbol_typed_uris', True)
self.sbol_doc = Document()
self.assembly_count = 0
assembly_list: AssemblyResultSet = params[0]
for assembly in assembly_list:
self.handle_assembly(assembly)
self.sbol_doc.write('output_sbol.xml')
def write_sbol(filled_library,
filled_library_metadata,
filled_description,
molecule_type=BIOPAX_DNA,
part_column="Part Name",
sequence_column="Sequence",
description_column="Description (Optional)",
role_column="Role",
length_column="length (bp)"):
"""
the function reads an excel spreadsheet and converts it into valid sbol
Parameters
----------
filled_library : DATAFRAME
Dataframe containing the library part data
filled_metadata : DATAFRAME
Dataframe containing the metadata
filled_description : DATAFRAME
Dataframe containing the description data
molecule_type : STRING, default = BIOPAX_DNA
Molecule type
part_column : STRING, default = "Part Name"
Name of the dataframe column that contains the part names
sequence_column : STRING, default = "Sequence"
Name of the dataframe column that contains the sequence
description_column : STRING, default = "Description (Optional)"
Name of the dataframe column that contains the library description
role_column : STRING, default = "Role"
Name of the dataframe column that contains the Role
length_column : STRING, default = "length (bp)"
Name of the dataframe column that contains the sequence length
Returns
-------
doc: SBOL Document
Document containing all components and sequences
Example
-------
Doc = write_sbol(filled_library, filled_library_metadata, filled_description)
"""
doc = Document()
Config.setOption('sbol_typed_uris', False)
#Define SBOL object and components
#Parts Library
molecule_type = BIOPAX_DNA #Change later
part_column = "Part Name"
sequence_column = "Sequence"
description_column = "Description (Optional)"
role_column = "Role"
length_column = "length (bp)"
for index, row in filled_library.iterrows():
component = ComponentDefinition(row[part_column], molecule_type)
component.roles = ontology[row[role_column]]
component.name = row[part_column]
if not (pd.isnull(row[description_column])):
component.description = row[description_column]
doc.addComponentDefinition(component)
row[sequence_column] = "".join(row[sequence_column].split())
row[sequence_column] = row[sequence_column].replace(u"\ufeff", "")
row[sequence_column] = row[sequence_column].lower()
if len(row[sequence_column]) != row[length_column]:
logging.warning(
f"The length of the sequence {row[part_column]} does not coincide with the length in the corresponding column 'length (bp)'"
)
sequence = Sequence(f"{row[part_column]}_sequence",
row[sequence_column], SBOL_ENCODING_IUPAC)
doc.addSequence(sequence)
component.sequences = sequence
#Metadata
doc.description = str(filled_description.values)
doc.name = filled_library_metadata.iloc[0, 1]
doc.write('SBOL_testcollection.xml')
return (doc)