def GFFGetDuplicateSGDs(gff_data):
"""Return GFF data with duplicate SGD names
Returns an OrderedDictionary where the keys are the duplicate
SGD names, e.g.
dups = GFFGetDuplicateSGDs(gff)
for sgd in dups.keys():
... loops over SGD names ...
for data in dups[sgd]:
... loops over duplicates ...
Note that duplicates are determined purely by SGD name; no
account is taken of chromosome or strand.
Arguments:
gff_data: a GFFFile object containing the GFF file data
Returns:
OrderedDictionary with SGDs as keys for lists of the
duplicate TabDataLines corresponding to the SGD
"""
# Use ordered dictionary to sort info on duplicates
duplicates = OrderedDictionary()
# Process data line-by-line
for data in gff_data:
attributes = data['attributes']
if 'SGD' in attributes:
# Store data
sgd = attributes['SGD']
if sgd != '':
# Check for duplicates
if sgd in duplicates:
duplicates[sgd].append(data)
else:
duplicates[sgd] = [data]
# Filter out true duplicates i.e. SGDs with at least two
# GFF data lines
for sgd in duplicates.keys():
if len(duplicates[sgd]) < 2:
del(duplicates[sgd])
# Finished
return duplicates
def main():
"""Main program
"""
# Set up logging format
logging.basicConfig(format='%(levelname)s: %(message)s')
p = optparse.OptionParser(usage="%prog [options] <file>.gff",
version="%prog "+__version__,
description=
"Utility to perform various 'cleaning' operations on a GFF file.")
p.add_option('-o',action='store',dest='output_gff',default=None,
help="Name of output GFF file (default is '<file>_clean.gff')")
p.add_option('--prepend',action='store',dest='prepend_str',default=None,
help="String to prepend to seqname in first column")
p.add_option('--clean',action='store_true',dest='do_clean',
help="Perform all the 'cleaning' manipulations on the input data (equivalent "
"to specifying all of --clean-score, --clean-replace-attributes, "
"--clean-exclude-attributes and --clean-group-sgds)")
p.add_option('--clean-score',action='store_true',dest='do_clean_score',
help="Replace 'Anc_*' and blanks in 'score' field with zeroes")
p.add_option('--clean-replace-attributes',action='store_true',
dest='do_clean_replace_attributes',
help="Replace 'ID', 'Gene', 'Parent' and 'Name' attributes with the value "
"of the SGD attribute, if present")
p.add_option('--clean-exclude-attributes',action='store_true',
dest='do_clean_exclude_attributes',
help="Remove the 'kaks', 'kaks2' and 'ncbi' attributes (to remove "
"arbitrary attributes, see the --remove-attribute=... option)")
p.add_option('--clean-group-sgds',action='store_true',dest='do_clean_group_sgds',
help="Group features with the same SGD by adding unique numbers to the 'ID' "
"attributes; IDs will have the form 'CDS:<SGD>:<n>' (where n is a unique "
"number for a given SGD)")
p.add_option('--report-duplicates',action='store_true',dest='report_duplicates',
help="Report duplicate SGD names and write list to <file>_duplicates.gff "
"with line numbers, chromosome, start coordinate and strand.")
p.add_option('--resolve-duplicates',action='store',dest='mapping_file',default=None,
help="Resolve duplicate SGDs by matching against 'best' genes in the supplied "
"mapping file; other non-matching genes are discarded and written to "
"<file>_discarded.gff.")
p.add_option('--discard-unresolved',action='store_true',dest='discard_unresolved',
help="Discard any unresolved duplicates, which are written to "
"<file>_unresolved.gff.")
p.add_option('--insert-missing',action='store',dest='gene_file',default=None,
help="Insert genes from gene file with SGD names that don't appear in the "
"input GFF. If GENE_FILE is blank ('='s must still be present) then the mapping "
"file supplied with the --resolve-duplicates option will be used instead.")
p.add_option('--add-exon-ids',action='store_true',dest='add_exon_ids',default=False,
help="For exon features without an ID attribute, construct and insert an "
"ID of the form 'exon:<Parent>:<n>' (where n is a unique number).")
p.add_option('--add-missing-ids',action='store_true',dest='add_missing_ids',default=False,
help="For features without an ID attribute, construct and insert a "
"generated ID of the form '<feature>:<Parent>:<n>' (where n is a unique "
"number).")
p.add_option('--no-percent-encoding',action='store_true',dest='no_encoding',default=False,
help="Convert encoded attributes to the correct characters in "
"the output GFF. WARNING this may result in a non-cannonical GFF that can't "
"be read correctly by this or other programs.")
p.add_option('--remove-attribute',action='append',dest='rm_attr',
help="Remove attribute RM_ATTR from the list of attributes for all records "
"in the GFF file (can be specified multiple times)")
p.add_option('--strict-attributes',action='store_true',dest='strict_attributes',
help="Remove attributes that don't conform to the KEY=VALUE format")
p.add_option('--debug',action='store_true',dest='debug',
help="Print debugging information")
# Process the command line
options,arguments = p.parse_args()
# Check for debugging
if options.debug:
# Turn on debugging output
logging.getLogger().setLevel(logging.DEBUG)
# Input files
if len(arguments) != 1:
p.error("input GFF file required")
else:
infile = arguments[0]
if not os.path.exists(infile):
p.error("Input file '%s' not found" % infile)
# Report version
p.print_version()
# Set flags based on command line
# String to prepend to first column
prepend_str = options.prepend_str
# Cleaning options
if options.do_clean:
# Update values in the "score" column
clean_score = True
# Clean up the "attributes" column
clean_replace_attributes = True
clean_exclude_attributes = True
# Set ID field in "attributes" to group lines with matching SGDs
group_SGDs = True
else:
# Set options based on user input
clean_score = options.do_clean_score
clean_replace_attributes = options.do_clean_replace_attributes
clean_exclude_attributes = options.do_clean_exclude_attributes
group_SGDs = options.do_clean_group_sgds
# Report duplicate names
report_duplicates = options.report_duplicates
# Resolve duplicated genes using CDS file
if options.mapping_file is not None:
resolve_duplicates = True
cdsfile = options.mapping_file
else:
resolve_duplicates = False
cdsfile = None
# Discard unresolved duplicates
discard_unresolved = options.discard_unresolved
# Insert missing genes
if options.gene_file is not None:
insert_missing = True
if options.gene_file:
genefile = options.gene_file
else:
genefile = cdsfile
else:
insert_missing = False
genefile = None
# Add an artificial exon ID attribute
add_exon_ids = options.add_exon_ids
# Add generated ID attributes
add_missing_ids = options.add_missing_ids
# Suppress encoding of attributes on output
no_attribute_encoding = options.no_encoding
# Remove attributes that don't conform to KEY=VALUE format
strict_attributes = options.strict_attributes
# Name for output files
##outext = os.path.splitext(os.path.basename(infile))[1]
if not options.output_gff:
outbase = os.path.splitext(os.path.basename(infile))[0]
outfile = outbase+'_clean.gff'
else:
outbase = os.path.splitext(os.path.basename(options.output_gff))[0]
outfile = options.output_gff
print "Input : %s" % infile
print "Output: %s" % outfile
dupfile = outbase+'_duplicates.txt'
delfile = outbase+'_discarded.gff'
unresfile = outbase+'_unresolved.gff'
# Read in data from file
gff_data = GFFFile(infile)
# Prepend string to seqname column
if prepend_str is not None:
print "Prepending '%s' to values in 'seqname' column" % prepend_str
for data in gff_data:
data['seqname'] = prepend_str+str(data['seqname'])
# Check/clean score column values
if clean_score:
print "Replacing 'Anc_*' and blanks with '0's in 'score' column"
score_unexpected_values = []
for data in gff_data:
try:
# Numerical value
score = float(data['score'])
if score != 0:
score_unexpected_values.append(data['score'])
except ValueError:
# String value
if data['score'].strip() != '' and not data['score'].startswith('Anc_'):
score_unexpected_values.append(data['score'])
# Replace "Anc_*" or blank values in "score" column with zero
if data['score'].startswith('Anc_') or data['score'].strip() == '':
data['score'] = '0'
# Report unexpected values
n = len(score_unexpected_values)
if n > 0:
logging.warning("%d 'score' values that are not '', 0 or 'Anc_*'" % n)
logging.warning("Other values: %s" % score_unexpected_values)
# Clean up the data in "attributes" column: replace keys
if clean_replace_attributes:
# Initialise mapping of keys from input to output in "attributes" column
# where new values are required etc
attributes_key_map = OrderedDictionary()
attributes_key_map['ID'] = 'SGD'
attributes_key_map['Gene'] = 'SGD'
attributes_key_map['Parent'] = 'SGD'
attributes_key_map['Name'] = 'SGD'
attributes_dont_replace_with_empty_data = True
print "Cleaning up attributes: replacing keys:"
for key in attributes_key_map.keys():
print "\t%s -> %s" % (key,attributes_key_map[key])
if attributes_dont_replace_with_empty_data:
print "(Replacement will be skipped if new data is missing/blank)"
GFFUpdateAttributes(gff_data,attributes_key_map,[],
attributes_dont_replace_with_empty_data)
# Clean up the data in "attributes" column: exclude keys
if clean_exclude_attributes:
# List of keys to exclude
attributes_exclude_keys = ['kaks','kaks2','ncbi']
print "Excluding keys:"
for key in attributes_exclude_keys:
print "\t%s" % key
GFFUpdateAttributes(gff_data,{},attributes_exclude_keys,True)
# Set the IDs for consecutive lines with matching SGD names, to indicate that
# they're in the same gene
if group_SGDs:
print "Grouping SGDs by setting ID's for consecutive lines with the same SGD values"
GFFGroupSGDs(gff_data)
# Find duplicates in input file
if report_duplicates or resolve_duplicates:
duplicate_sgds = GFFGetDuplicateSGDs(gff_data)
if report_duplicates:
# Write to duplicates file
print "Writing duplicate SGD names to %s" % dupfile
fd = open(dupfile,'w')
ndup = 0
ngroups = 0
for sgd in duplicate_sgds.keys():
assert(len(duplicate_sgds[sgd]) > 1)
ndup += 1
fd.write("%s\t" % sgd)
for data in duplicate_sgds[sgd]:
# Write the line number, chromosome, start and strand data
line = ';'.join(('L'+str(data.lineno()),
str(data['seqname']),str(data['start']),str(data['end'])))
fd.write("\t%s" % line)
fd.write("\n")
logging.debug("%s\t%s" % (sgd,duplicate_sgds[sgd]))
for group in GroupGeneSubsets(duplicate_sgds[sgd]):
if len(group) > 1: ngroups += 1
if ndup == 0:
fd.write("No duplicate SGDs\n")
fd.close()
print "%d duplicates found (of which %d are trivial)" % (ndup,ngroups)
if resolve_duplicates:
print "Resolving duplicate SGDs using data from %s" % cdsfile
print "Discarded genes will be written to %s" % delfile
# Get data on best gene mappings from CDS file
# Format is tab-delimited, each line has:
# orf chr start end strand
mapping = TabFile(cdsfile,column_names=('name','chr','start','end','strand'))
# Overlap margin
overlap_margin = 1000
# Perform resolution
result = GFFResolveDuplicateSGDs(gff_data,mapping,duplicate_sgds,overlap_margin)
#
# Report the results
#
# Convenience variables for lists of unresolved, discarded etc duplicates
resolved_sgds = result['resolved_sgds']
unresolved_sgds_no_mapping_genes = result['unresolved_sgds_no_mapping_genes']
unresolved_sgds_no_mapping_genes_after_filter = \
result['unresolved_sgds_no_mapping_genes_after_filter']
unresolved_sgds_no_overlaps = result['unresolved_sgds_no_overlaps']
unresolved_sgds_multiple_matches = result['unresolved_sgds_multiple_matches']
discard = result['discard']
# Remaining unresolved cases
if len(unresolved_sgds_no_mapping_genes) > 0:
print "No mapping genes with same SGDs found in %s:" % cdsfile
for sgd in unresolved_sgds_no_mapping_genes:
print "\t%s" % sgd
print
if len(unresolved_sgds_no_mapping_genes_after_filter) > 0:
print "No mapping genes with same chromosome and/or strand:"
for sgd in unresolved_sgds_no_mapping_genes_after_filter:
print "\t%s" % sgd
print
if len(unresolved_sgds_no_overlaps) > 0:
print "No mapping genes with overlaps:"
for sgd in unresolved_sgds_no_overlaps:
print "\t%s" % sgd
print
if len(unresolved_sgds_multiple_matches) > 0:
print "Multiple matching mapping genes:"
for sgd in unresolved_sgds_multiple_matches:
print "\t%s" % sgd
print
# Summary counts for each case
print "Total number of duplicated indexes : %d" % len(duplicate_sgds.keys())
print "Number of resolved duplicate SGDs : %d" % len(resolved_sgds)
print "Unresolved duplicates:"
print "* No mapping genes with same SGD : %d" % len(unresolved_sgds_no_mapping_genes)
print "* No mapping genes with same chr/str : %d" % len(unresolved_sgds_no_mapping_genes_after_filter)
print "* No mapping genes with overlap : %d" % len(unresolved_sgds_no_overlaps)
print "* Multiple mapping genes match : %d" % len(unresolved_sgds_multiple_matches)
# Remove discarded duplicates from the data
print "Removing discarded duplicates and writing to %s" % delfile
fd = open(delfile,'w')
for discard_data in discard:
try:
ip = gff_data.indexByLineNumber(discard_data.lineno())
del(gff_data[ip])
fd.write("%s\n" % discard_data)
except IndexError:
logging.warning("Failed to delete line %d: not found" % discard_data.lineno())
fd.close()
# Remove unresolved duplicates if requested
if discard_unresolved:
print "Removing unresolved duplicates and writing to %s" % unresfile
# Get list of unresolved SGDs
all_unresolved = result['unresolved_sgds']
# Get list of unresolved duplicates
unresolved = []
for data in gff_data:
attributes = data['attributes']
if 'SGD' in attributes:
if attributes['SGD'] in all_unresolved:
unresolved.append(data)
# Discard them
fu = open(unresfile,'w')
for discard in unresolved:
try:
ip = gff_data.indexByLineNumber(discard.lineno())
del(gff_data[ip])
fu.write("%s\n" % discard)
except IndexError:
logging.warning("Failed to delete line %d: not found" % discard.lineno())
fu.close()
# Look for "missing" genes in mapping file
if insert_missing:
# Get name for file with gene list
if genefile is None:
genefile = cdsfile
print "Inserting unmatched genes from %s" % genefile
# Get gene data from CDS file
# Format is tab-delimited, each line has:
# orf chr start end strand
mapping = TabFile(genefile,column_names=('name','chr','start','end','strand'))
n_genes_before_insert = len(gff_data)
gff_data = GFFInsertMissingGenes(gff_data,mapping)
print "Inserted %d missing genes" % (len(gff_data) - n_genes_before_insert)
# Construct and insert ID for exons
if add_exon_ids:
print "Inserting artificial IDs for exon records"
gff_data = GFFAddExonIDs(gff_data)
# Construct and insert missing ID attributes
if add_missing_ids:
print "Inserting generated IDs for records where IDs are missing"
gff_data = GFFAddIDAttributes(gff_data)
# Strip attributes requested for removal
if options.rm_attr:
print "Removing the following attributes from all records:"
for attr in options.rm_attr:
print "\t* %s" % attr
GFFUpdateAttributes(gff_data,exclude_keys=options.rm_attr)
# Remove attributes that don't conform to KEY=VALUE format
if strict_attributes:
print "Removing attributes that don't conform to KEY=VALUE format"
GFFUpdateAttributes(gff_data,exclude_nokeys=True)
# Suppress percent encoding of attributes
if no_attribute_encoding:
print "Converting encoded special characters in attribute data to non-encoded form"
logging.warning("!!! Special characters will not be correctly encoded in the output !!!")
logging.warning("!!! The resulting GFF may not be readable by this or other programs !!!")
gff_data = GFFDecodeAttributes(gff_data)
# Write to output file
print "Writing output file %s" % outfile
gff_data.write(outfile)
