class GenbankToGenome:
def __init__(self, config):
self.cfg = config
self.gi = GenomeInterface(config)
self.dfu = DataFileUtil(config.callbackURL)
self.aUtil = AssemblyUtil(config.callbackURL)
self.ws = Workspace(config.workspaceURL)
self._messages = []
self.time_string = str(
datetime.datetime.fromtimestamp(
time.time()).strftime('%Y_%m_%d_%H_%M_%S'))
yml_text = open('/kb/module/kbase.yml').read()
self.version = re.search("module-version:\n\W+(.+)\n",
yml_text).group(1)
self.generate_parents = False
self.generate_ids = False
self.genes = OrderedDict()
self.mrnas = OrderedDict()
self.cdss = OrderedDict()
self.noncoding = []
self.ontologies_present = defaultdict(dict)
self.ontology_events = list()
self.skiped_features = Counter()
self.feature_counts = Counter()
self.orphan_types = Counter()
self.contig_seq = {}
self.circ_contigs = set()
self.features_spaning_zero = set()
self.genome_warnings = []
self.genome_suspect = False
self.defects = Counter()
self.spoofed_genes = 0
self.excluded_features = ('source', 'exon', 'fasta_record')
self.ont_mappings = load_ontology_mappings('/kb/module/data')
self.code_table = 11
self.re_api_url = config.re_api_url
# dict with feature 'id's that have been used more than once.
self.used_twice_identifiers = {}
self.default_params = {
'source':
'Genbank',
'taxon_wsname':
self.cfg.raw['taxon-workspace-name'],
'taxon_lookup_obj_name':
self.cfg.raw['taxon-lookup-object-name'],
'ontology_wsname':
self.cfg.raw['ontology-workspace-name'],
'ontology_GO_obj_name':
self.cfg.raw['ontology-gene-ontology-obj-name'],
'ontology_PO_obj_name':
self.cfg.raw['ontology-plant-ontology-obj-name'],
'release':
None,
'genetic_code':
11,
'generate_ids_if_needed':
0,
'metadata': {}
}
@property
def messages(self):
return "\n".join(self._messages)
def refactored_import(self, ctx, params):
# 1) validate parameters and extract defaults
self.validate_params(params)
# 2) construct the input directory staging area
input_directory = self.stage_input(params)
# 3) update default params
self.default_params.update(params)
params = self.default_params
self.generate_parents = params.get('generate_missing_genes')
self.generate_ids = params.get('generate_ids_if_needed')
if params.get('genetic_code'):
self.code_table = params['genetic_code']
# 4) Do the upload
files = self._find_input_files(input_directory)
consolidated_file = self._join_files_skip_empty_lines(files)
genome = self.parse_genbank(consolidated_file, params)
if params.get('genetic_code'):
genome["genetic_code"] = params['genetic_code']
result = self.gi.save_one_genome({
'workspace': params['workspace_name'],
'name': params['genome_name'],
'data': genome,
"meta": params['metadata'],
})
ref = f"{result['info'][6]}/{result['info'][0]}/{result['info'][4]}"
logging.info(f"Genome saved to {ref}")
# 5) clear the temp directory
shutil.rmtree(input_directory)
# 6) return the result
info = result['info']
details = {'genome_ref': ref, 'genome_info': info}
return details
@staticmethod
def validate_params(params):
if 'workspace_name' not in params:
raise ValueError('required "workspace_name" field was not defined')
if 'genome_name' not in params:
raise ValueError('required "genome_name" field was not defined')
if 'file' not in params:
raise ValueError('required "file" field was not defined')
# one and only one of 'path', 'shock_id', or 'ftp_url' is required
file = params['file']
if not isinstance(file, dict):
raise ValueError('required "file" field must be a map/dict')
sources = ('path', 'shock_id', 'ftp_url')
n_valid_fields = sum(1 for f in sources if file.get(f))
if n_valid_fields < 1:
raise ValueError(f'required "file" field must include one source: '
f'{", ".join(sources)}')
if n_valid_fields > 1:
raise ValueError(
f'required "file" field has too many sources specified: '
f'{", ".join(file.keys())}')
if params.get('genetic_code'):
if not (isinstance(params['genetic_code'], int)
and 0 < params['genetic_code'] < 32):
raise ValueError(f"Invalid genetic code specified: {params}")
def stage_input(self, params):
""" Setup the input_directory by fetching the files and uncompressing if needed. """
# construct the input directory where we stage files
input_directory = os.path.join(
self.cfg.sharedFolder, f'genome-upload-staging-{uuid.uuid4()}')
os.makedirs(input_directory)
# at this point, the 'file' input is validated, so we don't have to catch any special cases
# we expect one and only one of path, shock_id, or ftp_url
# determine how to get the file: if it is from shock, download it. If it
# is just sitting there, then use it. Move the file to the staging input directory
file = params['file']
genbank_file_path = None
if file.get('path') is not None:
# copy the local file to the input staging directory
# (NOTE: could just move it, but then this method would have the side effect of moving your
# file which another SDK module might have an open handle on)
local_file_path = file['path']
genbank_file_path = os.path.join(input_directory,
os.path.basename(local_file_path))
shutil.copy2(local_file_path, genbank_file_path)
if 'shock_id' in file and file['shock_id'] is not None:
# handle shock file
logging.info(
f'Downloading file from SHOCK node: {self.cfg.shockURL} - {file["shock_id"]}'
)
sys.stdout.flush()
file_name = self.dfu.shock_to_file({
'file_path': input_directory,
'shock_id': file['shock_id']
})['node_file_name']
genbank_file_path = os.path.join(input_directory, file_name)
if 'ftp_url' in file and file['ftp_url'] is not None:
logging.info('Downloading file from: ' + str(file['ftp_url']))
local_file_path = self.dfu.download_web_file({
'file_url':
file['ftp_url'],
'download_type':
'FTP'
})['copy_file_path']
genbank_file_path = os.path.join(input_directory,
os.path.basename(local_file_path))
shutil.copy2(local_file_path, genbank_file_path)
# extract the file if it is compressed
if genbank_file_path is not None:
logging.info("staged input file =" + genbank_file_path)
self.dfu.unpack_file({'file_path': genbank_file_path})
else:
raise ValueError(
'No valid files could be extracted based on the input')
return input_directory
def parse_genbank(self, file_path, params):
logging.info("Saving original file to shock")
shock_res = self.dfu.file_to_shock({
'file_path': file_path,
'make_handle': 1,
'pack': 'gzip',
})
# Write and save assembly file
assembly_ref = self._save_assembly(file_path, params)
assembly_data = self.dfu.get_objects({
'object_refs': [assembly_ref],
'ignore_errors': 0
})['data'][0]['data']
genome = {
"id": params['genome_name'],
"original_source_file_name": os.path.basename(file_path),
"assembly_ref": assembly_ref,
"gc_content": assembly_data['gc_content'],
"dna_size": assembly_data['dna_size'],
"md5": assembly_data['md5'],
"genbank_handle_ref": shock_res['handle']['hid'],
"publications": set(),
"contig_ids": [],
"contig_lengths": [],
}
genome['source'], genome['genome_tiers'] = self.gi.determine_tier(
params['source'])
if params.get('genome_type'):
genome['genome_type'] = params['genome_type']
# Set taxonomy-related fields in the genome
# Also validates the given taxon ID
if params.get('taxon_id'):
set_taxon_data(int(params['taxon_id']), self.re_api_url, genome)
else:
set_default_taxon_data(genome)
dates = []
# Parse data from genbank file
contigs = Bio.SeqIO.parse(file_path, "genbank")
for record in contigs:
r_annot = record.annotations
logging.info("parsing contig: " + record.id)
try:
dates.append(time.strptime(r_annot.get('date'), "%d-%b-%Y"))
except (TypeError, ValueError):
pass
genome['contig_ids'].append(record.id)
genome['contig_lengths'].append(len(record))
genome["publications"] |= self._get_pubs(r_annot)
# only do the following once(on the first contig)
if "source_id" not in genome:
genome["source_id"] = record.id.split('.')[0]
organism = r_annot.get('organism', 'Unknown Organism')
if params.get('scientific_name'):
genome['scientific_name'] = params['scientific_name']
else:
genome['scientific_name'] = organism
self.code_table = genome['genetic_code']
genome["molecule_type"] = r_annot.get('molecule_type', 'DNA')
genome['notes'] = r_annot.get('comment',
"").replace('\\n', '\n')
self._parse_features(record, genome['source'])
genome.update(self.get_feature_lists())
genome['num_contigs'] = len(genome['contig_ids'])
# add dates
dates.sort()
if dates:
genome['external_source_origination_date'] = time.strftime(
"%d-%b-%Y", dates[0])
if dates[0] != dates[-1]:
genome['external_source_origination_date'] += " _ " + \
time.strftime("%d-%b-%Y", dates[-1])
if self.ontologies_present:
genome['ontologies_present'] = dict(self.ontologies_present)
genome["ontology_events"] = self.ontology_events
genome['feature_counts'] = dict(self.feature_counts)
# can't serialize a set
genome['publications'] = list(genome['publications'])
if len(genome['cdss']) and (self.defects['cds_seq_not_matching'] /
float(len(genome['cdss'])) > 0.02):
self.genome_warnings.append(
warnings["genome_inc_translation"].format(
self.defects['cds_seq_not_matching'], len(genome['cdss'])))
self.genome_suspect = 1
if self.defects['bad_parent_loc']:
self.genome_warnings.append(
f"There were {self.defects['bad_parent_loc']} parent/child "
"relationships that were not able to be determined. Some of "
"these may have splice variants that may be valid relationships."
)
if self.defects['spoofed_genes']:
self.genome_warnings.append(warnings['spoofed_genome'].format(
self.defects['spoofed_genes']))
genome['suspect'] = 1
if self.defects['not_trans_spliced']:
self.genome_warnings.append(
warnings['genome_not_trans_spliced'].format(
self.defects['not_trans_spliced']))
genome['suspect'] = 1
if self.genome_warnings:
genome['warnings'] = self.genome_warnings
if self.genome_suspect:
genome['suspect'] = 1
logging.info(f"Feature Counts: {genome['feature_counts']}")
return genome
def _save_assembly(self, genbank_file, params):
"""Convert genbank file to fasta and sve as assembly"""
contigs = Bio.SeqIO.parse(genbank_file, "genbank")
assembly_id = f"{params['genome_name']}_assembly"
fasta_file = f"{self.cfg.sharedFolder}/{params['genome_name']}_assembly.fasta"
out_contigs = []
extra_info = defaultdict(dict)
for in_contig in contigs:
if in_contig.annotations.get('topology', "") == 'circular':
extra_info[in_contig.id]['is_circ'] = 1
self.circ_contigs.add(in_contig.id)
elif in_contig.annotations.get('topology', "") == 'linear':
extra_info[in_contig.id]['is_circ'] = 0
out_contigs.append(in_contig)
self.contig_seq[in_contig.id] = in_contig.seq.upper()
assembly_ref = params.get("use_existing_assembly")
if assembly_ref:
if not re.match("\d+\/\d+\/\d+", assembly_ref):
raise ValueError(
f"Assembly ref: {assembly_ref} is not a valid format. Must"
f" be in numerical <ws>/<object>/<version> format.")
ret = self.dfu.get_objects({'object_refs':
[assembly_ref]})['data'][0]
if "KBaseGenomeAnnotations.Assembly" not in ret['info'][2]:
raise ValueError(
f"{assembly_ref} is not a reference to an assembly")
unmatched_ids = list()
unmatched_ids_md5s = list()
for current_contig in self.contig_seq.keys():
current_contig_md5 = hashlib.md5(
str(self.contig_seq[current_contig]).encode(
'utf8')).hexdigest()
if current_contig in ret['data']['contigs']:
if current_contig_md5 != ret['data']['contigs'][
current_contig]['md5']:
unmatched_ids_md5s.append(current_contig)
else:
unmatched_ids.append(current_contig)
if len(unmatched_ids) > 0:
raise ValueError(warnings['assembly_ref_extra_contigs'].format(
", ".join(unmatched_ids)))
if len(unmatched_ids_md5s) > 0:
raise ValueError(warnings["assembly_ref_diff_seq"].format(
", ".join(unmatched_ids_md5s)))
logging.info(f"Using supplied assembly: {assembly_ref}")
return assembly_ref
logging.info("Saving sequence as Assembly object")
Bio.SeqIO.write(out_contigs, fasta_file, "fasta")
assembly_ref = self.aUtil.save_assembly_from_fasta({
'file': {
'path': fasta_file
},
'workspace_name':
params['workspace_name'],
'assembly_name':
assembly_id,
'type':
params.get('genome_type', 'isolate'),
'contig_info':
extra_info
})
logging.info(f"Assembly saved to {assembly_ref}")
return assembly_ref
def _find_input_files(self, input_directory):
logging.info("Scanning for Genbank Format files.")
valid_extensions = [".gbff", ".gbk", ".gb", ".genbank", ".dat", ".gbf"]
files = os.listdir(os.path.abspath(input_directory))
logging.info("Genbank Files : " + ", ".join(files))
genbank_files = [
x for x in files
if os.path.splitext(x)[-1].lower() in valid_extensions
]
if len(genbank_files) == 0:
raise Exception(
f"The input directory does not have any files with one of the "
f"following extensions {','.join(valid_extensions)}.")
logging.info(f"Found {len(genbank_files)} genbank files")
input_files = []
for genbank_file in genbank_files:
input_files.append(os.path.join(input_directory, genbank_file))
return input_files
def _join_files_skip_empty_lines(self, input_files):
""" Applies strip to each line of each input file.
Args:
input_files: Paths to input files in Genbank format.
Returns:
Path to resulting file (currenly it's the same file as input).
"""
if len(input_files) == 0:
raise ValueError("NO GENBANK FILE")
temp_dir = os.path.join(os.path.dirname(input_files[0]), "combined")
if not os.path.exists(temp_dir):
os.makedirs(temp_dir)
ret_file = os.path.join(temp_dir, os.path.basename(input_files[0]))
# take in Genbank file and remove all empty lines from it.
with open(ret_file, 'w', buffering=2**20) as f_out:
for input_file in input_files:
with open(input_file, 'r') as f_in:
for line in f_in:
line = line.rstrip('\r\n')
if line.strip():
f_out.write(line + '\n')
return ret_file
def _get_pubs(self, r_annotations):
"""Get a contig's publications"""
pub_list = []
for in_pub in r_annotations.get('references', []):
# don't add blank pubs
if not in_pub.authors:
continue
out_pub = [
0, # pmid
"", # source
in_pub.title,
"", # web address
"", # date
in_pub.authors,
in_pub.journal,
]
date_match = re.match("\((\d{4})\)", in_pub.journal)
if date_match:
out_pub[4] = date_match.group(1)
if in_pub.pubmed_id:
out_pub[0:4] = [
int(in_pub.pubmed_id), "PubMed", in_pub.title,
f"http://www.ncbi.nlm.nih.gov/pubmed/{in_pub.pubmed_id}"
]
pub_list.append(tuple(out_pub))
logging.info(f"Parsed {len(pub_list)} publication records")
return set(pub_list)
def _get_id(self, feat, tags=None):
"""Assign a id to a feature based on the first tag that exists"""
_id = ""
if not tags:
tags = ['locus_tag', 'kbase_id']
for t in tags:
_id = feat.qualifiers.get(t, [""])[0]
if _id:
break
if not _id:
if feat.type == 'gene':
if not self.generate_ids:
raise ValueError(
f"Unable to find a valid id for gene "
f"among these tags: {', '.join(tags)}. Correct the "
f"file or rerun with generate_ids\n {feat}")
self.orphan_types['gene'] += 1
_id = f"gene_{self.orphan_types['gene']}"
if 'rna' in feat.type.lower() or feat.type in {
'CDS', 'sig_peptide', 'five_prime_UTR', 'three_prime_UTR'
}:
_id = f"gene_{self.orphan_types['gene']}"
return _id
def _parse_features(self, record, source):
def _location(feat):
"""Convert to KBase style location objects"""
strand_trans = ("", "+", "-")
loc = []
for part in feat.location.parts:
contig_id = part.ref if part.ref else record.id
if part.strand >= 0:
begin = int(part.start) + 1
else:
begin = int(part.end)
loc.append(
(contig_id, begin, strand_trans[part.strand], len(part)))
return loc
def _warn(message):
if message not in out_feat.get('warnings', []):
out_feat['warnings'] = out_feat.get('warnings', []) + [message]
def _check_suspect_location(parent=None):
if 'trans_splicing' in out_feat.get('flags', []):
return
if out_feat['location'] == sorted(
out_feat['location'],
reverse=(in_feature.location.strand == -1)):
return
if record.id in self.circ_contigs and \
in_feature.location.start == 0 \
and in_feature.location.end == len(record):
self.features_spaning_zero.add(out_feat['id'])
return
if parent and parent['id'] in self.features_spaning_zero:
return
_warn(warnings['not_trans_spliced'])
self.defects['not_trans_spliced'] += 1
for in_feature in record.features:
if in_feature.type in self.excluded_features:
self.skiped_features[in_feature.type] += 1
continue
feat_seq = self._get_seq(in_feature, record.id)
if source == "Ensembl":
_id = self._get_id(in_feature, ['gene', 'locus_tag'])
else:
_id = self._get_id(in_feature)
# The following is common to all the feature types
out_feat = {
"id": "_".join([_id, in_feature.type]),
"location": _location(in_feature),
"dna_sequence": str(feat_seq),
"dna_sequence_length": len(feat_seq),
"md5": hashlib.md5(str(feat_seq).encode('utf8')).hexdigest(),
}
if not _id:
out_feat['id'] = in_feature.type
# validate input feature
# note that end is the larger number regardless of strand
if int(in_feature.location.end) > len(record):
self.genome_warnings.append(
warnings["coordinates_off_end"].format(out_feat['id']))
self.genome_suspect = 1
continue
for piece in in_feature.location.parts:
if not isinstance(piece.start, ExactPosition) \
or not isinstance(piece.end, ExactPosition):
_warn(warnings["non_exact_coordinates"])
self.feature_counts[in_feature.type] += 1
# add optional fields
if 'note' in in_feature.qualifiers:
out_feat['note'] = in_feature.qualifiers["note"][0]
out_feat.update(self._get_aliases_flags_functions(in_feature))
ont, db_xrefs = self._get_ontology_db_xrefs(in_feature)
if ont:
out_feat['ontology_terms'] = ont
if db_xrefs:
out_feat['db_xrefs'] = db_xrefs
if 'inference' in in_feature.qualifiers:
out_feat['inference_data'] = parse_inferences(
in_feature.qualifiers['inference'])
_check_suspect_location(self.genes.get(_id))
# add type specific features
if in_feature.type == 'CDS':
self.process_cds(_id, feat_seq, in_feature, out_feat)
elif in_feature.type == 'gene':
self.process_gene(_id, out_feat)
elif in_feature.type == 'mRNA':
self.process_mrna(_id, out_feat)
else:
self.noncoding.append(
self.process_noncoding(_id, in_feature.type, out_feat))
def get_feature_lists(self):
"""sort genes into their final arrays"""
coding = []
for g in self.genes.values():
if len(g['cdss']):
if g['mrnas'] and len(g['mrnas']) != len(g['cdss']):
msg = "The length of the mrna and cdss arrays are not equal"
g['warnings'] = g.get('warnings', []) + [msg]
# remove duplicates that may arise from CDS info propagation
for key in ('functions', 'aliases', 'db_xrefs'):
if key in g:
g[key] = list(set(g[key]))
if not g['mrnas']:
del g['mrnas']
del g['type']
coding.append(g)
self.feature_counts["protein_encoding_gene"] += 1
else:
del g['mrnas'], g['cdss']
self.noncoding.append(g)
self.feature_counts["non_coding_genes"] += 1
self.feature_counts["non_coding_features"] = len(self.noncoding)
return {
'features': coding,
'non_coding_features': self.noncoding,
'cdss': list(self.cdss.values()),
'mrnas': list(self.mrnas.values())
}
def _get_seq(self, feat, contig):
"""Extract the DNA sequence for a feature"""
seq = []
for part in feat.location.parts:
strand = part.strand
# handle trans-splicing across contigs
if part.ref:
part_contig = part.ref
else:
part_contig = contig
if strand >= 0:
seq.append(
str(self.contig_seq[part_contig][part.start:part.end]))
else:
seq.append(
str(self.contig_seq[part_contig]
[part.start:part.end].reverse_complement()))
return "".join(seq)
def _create_ontology_event(self, ontology_type):
"""Creates the ontology_event if necessary
Returns the index of the ontology event back."""
if ontology_type not in self.ont_mappings:
raise ValueError(f"{ontology_type} is not a supported ontology")
if "event_index" not in self.ont_mappings[ontology_type]:
self.ont_mappings[ontology_type]['event_index'] = len(
self.ontology_events)
if ontology_type == "GO":
ontology_ref = "KBaseOntology/gene_ontology"
elif ontology_type == "PO":
ontology_ref = "KBaseOntology/plant_ontology"
else:
ontology_ref = f"KBaseOntology/{ontology_type.lower()}_ontology"
self.ontology_events.append({
"method": "GenomeFileUtils Genbank uploader from annotations",
"method_version": self.version,
"timestamp": self.time_string,
"id": ontology_type,
"ontology_ref": ontology_ref
})
return self.ont_mappings[ontology_type]['event_index']
def _get_ontology_db_xrefs(self, feature):
"""Splits the ontology info from the other db_xrefs"""
ontology = defaultdict(dict)
db_xrefs = []
for key in ("GO_process", "GO_function", "GO_component"):
ontology_event_index = self._create_ontology_event("GO")
for term in feature.qualifiers.get(key, []):
sp = term.split(" - ")
ontology['GO'][sp[0]] = [ontology_event_index]
self.ontologies_present['GO'][
sp[0]] = self.ont_mappings['GO'].get(sp[0], '')
for ref in feature.qualifiers.get('db_xref', []):
if ref.startswith('GO:'):
ontology['GO'][ref] = [self._create_ontology_event("GO")]
self.ontologies_present['GO'][ref] = self.ont_mappings[
'GO'].get(ref, '')
elif ref.startswith('PO:'):
ontology['PO'][ref] = [self._create_ontology_event("PO")]
self.ontologies_present['PO'][ref] = self.ont_mappings[
'PO'].get(ref, '')
elif ref.startswith('KO:'):
ontology['KO'][ref] = [self._create_ontology_event("KO")]
self.ontologies_present['KO'][ref] = self.ont_mappings[
'KO'].get(ref, '')
elif ref.startswith('COG'):
ontology['COG'][ref] = [self._create_ontology_event("COG")]
self.ontologies_present['COG'][ref] = self.ont_mappings[
'COG'].get(ref, '')
elif ref.startswith('PF'):
ontology['PFAM'][ref] = [self._create_ontology_event("PFAM")]
self.ontologies_present['PFAM'][ref] = self.ont_mappings[
'PFAM'].get(ref, '')
elif ref.startswith('TIGR'):
ontology['TIGRFAM'][ref] = [
self._create_ontology_event("TIGRFAM")
]
self.ontologies_present['TIGRFAM'][ref] = self.ont_mappings[
'TIGRFAM'].get(ref, '')
elif ":" not in ref:
db_xrefs.append(tuple(["Unknown_Source", ref]))
else:
db_xrefs.append(tuple(ref.split(":", 1)))
return dict(ontology), sorted(db_xrefs)
@staticmethod
def _get_aliases_flags_functions(feat):
"""Get the values for aliases flags and features from qualifiers"""
alias_keys = {
'locus_tag', 'old_locus_tag', 'protein_id', 'transcript_id',
'gene', 'EC_number', 'gene_synonym'
}
result = defaultdict(list)
for key, val_list in feat.qualifiers.items():
if key in alias_keys:
result['aliases'].extend([(key, val) for val in val_list])
# flags have no other information associated with them
if val_list == ['']:
result['flags'].append(key)
if key == 'function':
result['functional_descriptions'].extend(
val_list[0].split('; '))
if key == 'product':
result['functions'] = val_list
return result
def _find_parent_gene(self, potential_id, feature):
"""Unfortunately, Genbank files don't have a parent ID and the features can be out of
order at times. To account for this, the this function works backwards from the end of
list of IDs and stops when if finds a parent with valid coordinates or it hits the maximum
number of tries"""
if potential_id in self.genes:
lookup_attempts = 0
while lookup_attempts < MAX_PARENT_LOOKUPS:
if is_parent(self.genes[potential_id], feature):
return potential_id
lookup_attempts += 1
try:
potential_id = list(
self.genes.keys())[-(lookup_attempts + 1)]
except IndexError:
break # no more genes that could match exist
self.defects['bad_parent_loc'] += 1
return None
def assign_new_id(self, _id):
"""given a feature id that has already been used, add a unique modifier to it"""
_id_modifier = self.used_twice_identifiers.get(_id, 1)
self.used_twice_identifiers[_id] = _id_modifier + 1
return _id + "." + str(_id_modifier)
def process_gene(self, _id, out_feat):
out_feat.update({
"id": _id,
"type": 'gene',
"mrnas": [],
'cdss': [],
})
if _id in self.genes:
_id = self.assign_new_id(_id)
out_feat.update({"id": _id})
# raise ValueError(f"Duplicate gene ID: {_id}")
self.genes[_id] = out_feat
def process_noncoding(self, gene_id, feat_type, out_feat):
out_feat["type"] = feat_type
# this prevents big misc_features from blowing up the genome size
if out_feat['dna_sequence_length'] > MAX_MISC_FEATURE_SIZE:
del out_feat['dna_sequence']
gene_id = self._find_parent_gene(gene_id, out_feat)
if gene_id:
if 'children' not in self.genes[gene_id]:
self.genes[gene_id]['children'] = []
out_feat['id'] += "_" + str(
len(self.genes[gene_id]['children']) + 1)
self.genes[gene_id]['children'].append(out_feat['id'])
out_feat['parent_gene'] = gene_id
else:
self.orphan_types[feat_type] += 1
out_feat['id'] += "_" + str(self.orphan_types[feat_type])
return out_feat
def process_mrna(self, gene_id, out_feat):
if gene_id not in self.genes and self.generate_parents:
self.process_gene(gene_id, copy.copy(out_feat))
gene_id = self._find_parent_gene(gene_id, out_feat)
if gene_id:
out_feat['id'] = "_".join(
(gene_id, "mRNA", str(len(self.genes[gene_id]['mrnas']) + 1)))
self.genes[gene_id]['mrnas'].append(out_feat['id'])
out_feat['parent_gene'] = gene_id
else:
self.orphan_types['mrna'] += 1
out_feat['id'] = f"mRNA_{self.orphan_types['mrna']}"
out_feat['warnings'] = out_feat.get('warnings', []) + [
'Unable to find parent gene for ' + str(out_feat['id'])
]
self.mrnas[out_feat['id']] = out_feat
def process_cds(self, gene_id, feat_seq, in_feature, out_feat):
# Associate CDS with parents
cds_warnings = out_feat.get('warnings', [])
validated_gene_id = self._find_parent_gene(gene_id, out_feat)
if validated_gene_id:
out_feat['id'] = "_".join(
(validated_gene_id, "CDS",
str(len(self.genes[validated_gene_id]['cdss']) + 1)))
self.genes[validated_gene_id]['cdss'].append(out_feat['id'])
out_feat['parent_gene'] = validated_gene_id
elif self.generate_parents and gene_id not in self.genes:
new_feat = copy.copy(out_feat)
new_feat['id'] = gene_id
new_feat['warnings'] = [warnings['spoofed_gene']]
self.orphan_types['gene'] += 1
self.defects['spoofed_genes'] += 1
self.process_gene(new_feat['id'], new_feat)
out_feat['id'] = "_".join(
(gene_id, "CDS", str(len(self.genes[gene_id]['cdss']) + 1)))
self.genes[gene_id]['cdss'].append(out_feat['id'])
out_feat['parent_gene'] = gene_id
else:
self.orphan_types['cds'] += 1
out_feat['id'] = f"CDS_{self.orphan_types['cds']}"
cds_warnings.append(
f"Unable to find parent gene for {out_feat['id']}")
# there is a 1 to 1 relationship of mRNA to CDS so XXX_mRNA_1 will match XXX_CDS_1
mrna_id = out_feat["id"].replace('CDS', 'mRNA')
if mrna_id in self.mrnas:
if not is_parent(self.mrnas[mrna_id], out_feat):
cds_warnings.append(warnings['cds_mrna_cds'].format(mrna_id))
self.mrnas[mrna_id]['warnings'] = self.mrnas[mrna_id].get(
'warnings', []) + [warnings['cds_mrna_mrna']]
self.defects['bad_parent_loc'] += 1
else:
out_feat['parent_mrna'] = mrna_id
self.mrnas[mrna_id]['cds'] = out_feat['id']
# process protein
prot_seq = in_feature.qualifiers.get("translation", [""])[0]
# allow a little slack to account for frameshift and stop codon
if prot_seq and abs(len(prot_seq) * 3 - len(feat_seq)) > 4:
cds_warnings.append(warnings["inconsistent_CDS_length"].format(
len(feat_seq), len(prot_seq)))
self.genome_warnings.append(
warnings['genome_inc_CDS_length'].format(
out_feat['id'], len(feat_seq), len(prot_seq)))
self.genome_suspect = 1
try:
if prot_seq and prot_seq != Seq.translate(
feat_seq, self.code_table, cds=True).strip("*"):
cds_warnings.append(warnings["inconsistent_translation"])
self.defects['cds_seq_not_matching'] += 1
except TranslationError as e:
cds_warnings.append("Unable to verify protein sequence:" + str(e))
if not prot_seq:
try:
prot_seq = Seq.translate(feat_seq, self.code_table,
cds=True).strip("*")
cds_warnings.append(warnings["no_translation_supplied"])
except TranslationError as e:
cds_warnings.append(warnings["no_translation_supplied"] +
str(e))
out_feat.update({
"protein_translation":
prot_seq,
"protein_md5":
hashlib.md5(prot_seq.encode('utf8')).hexdigest(),
"protein_translation_length":
len(prot_seq),
})
if out_feat.get('parent_gene'):
propagate_cds_props_to_gene(out_feat,
self.genes[out_feat['parent_gene']])
if cds_warnings:
out_feat['warnings'] = cds_warnings
self.cdss[out_feat['id']] = out_feat
def test_user(self):
self.assertEqual(GenomeInterface.determine_tier('RefSeq user'),
('RefSeq', ['ExternalDB', 'User']))
class FastaGFFToGenome:
def __init__(self, config):
self.cfg = config
self.au = AssemblyUtil(config.callbackURL)
self.dfu = DataFileUtil(self.cfg.callbackURL)
self.gi = GenomeInterface(self.cfg)
self.taxon_wsname = self.cfg.raw['taxon-workspace-name']
self.time_string = str(
datetime.datetime.fromtimestamp(
time.time()).strftime('%Y_%m_%d_%H_%M_%S'))
yml_text = open('/kb/module/kbase.yml').read()
self.version = re.search("module-version:\n\W+(.+)\n",
yml_text).group(1)
self.ont_mappings = load_ontology_mappings('/kb/module/data')
self.code_table = 11
self.skip_types = ('exon', 'five_prime_UTR', 'three_prime_UTR',
'start_codon', 'stop_codon', 'region', 'chromosome',
'scaffold')
self.spoof_gene_count = 0
self.is_phytozome = False
self.strict = True
self.generate_genes = False
self.warnings = []
self.feature_dict = collections.OrderedDict()
self.cdss = set()
self.ontologies_present = collections.defaultdict(dict)
self.ontology_events = list()
self.skiped_features = collections.Counter()
self.feature_counts = collections.Counter()
def warn(self, message):
self.warnings.append(message)
def generate_genome_json(self, params):
# 1) validate parameters
self._validate_import_file_params(params)
self.code_table = params.get('genetic_code', 11)
# 2) construct the input directory staging area
input_directory = os.path.join(self.cfg.sharedFolder,
'fast_gff_upload_' + str(uuid.uuid4()))
os.makedirs(input_directory)
file_paths = self._stage_input(params, input_directory)
# 3) extract out the parameters
params = self._set_parsed_params(params)
if params.get('generate_missing_genes'):
self.generate_genes = True
# 4) do the upload
genome = self._gen_genome_json(
input_fasta_file=file_paths["fasta_file"],
input_gff_file=file_paths["gff_file"],
workspace_name=params['workspace_name'],
core_genome_name=params['genome_name'],
scientific_name=params['scientific_name'],
source=params['source'],
source_id=params['source_id'],
release=params['release'],
)
if params.get('genetic_code'):
genome["genetic_code"] = params['genetic_code']
return genome, input_directory
def import_file(self, params):
genome, input_directory = self.generate_genome_json(params)
json.dump(genome,
open(
"{}/{}.json".format(self.cfg.sharedFolder, genome['id']),
'w'),
indent=4)
result = self.gi.save_one_genome({
'workspace': params['workspace_name'],
'name': params['genome_name'],
'data': genome,
"meta": params.get('metadata', {}),
})
report_string = 'A genome with {} contigs and the following feature ' \
'types was imported: {}'\
.format(len(genome['contig_ids']), "\n".join(
[k + ": " + str(v) for k, v in genome['feature_counts'].items()]))
log(report_string)
# 5) clear the temp directory
shutil.rmtree(input_directory)
# 6) return the result
info = result['info']
details = {
'genome_ref':
str(info[6]) + '/' + str(info[0]) + '/' + str(info[4]),
'genome_info': info
}
return details
def _gen_genome_json(self,
input_gff_file=None,
input_fasta_file=None,
workspace_name=None,
core_genome_name=None,
scientific_name="unknown_taxon",
source=None,
source_id=None,
release=None):
# reading in GFF file
features_by_contig = self._retrieve_gff_file(input_gff_file)
contig_ids = set()
# parse feature information
fasta_contigs = Bio.SeqIO.parse(input_fasta_file, "fasta")
for contig in fasta_contigs:
molecule_type = str(contig.seq.alphabet).replace(
'IUPACAmbiguous', '').strip('()')
contig_ids.add(contig.id)
for feature in features_by_contig.get(contig.id, []):
self._transform_feature(contig, feature)
for cid in set(features_by_contig.keys()) - contig_ids:
self.warn("Sequence name {} does not match a sequence id in the "
"FASTA file. {} features will not be imported.".format(
cid, len(features_by_contig[cid])))
if self.strict:
raise ValueError(
"Every feature sequence id must match a fasta sequence id")
self._process_cdss()
# save assembly file
assembly_ref = self.au.save_assembly_from_fasta({
'file': {
'path': input_fasta_file
},
'workspace_name':
workspace_name,
'assembly_name':
core_genome_name + ".assembly"
})
assembly_data = self.dfu.get_objects({
'object_refs': [assembly_ref],
'ignore_errors': 0
})['data'][0]['data']
# generate genome info
genome = self._gen_genome_info(core_genome_name, scientific_name,
assembly_ref, source, source_id,
assembly_data, input_gff_file,
molecule_type)
genome['release'] = release
if self.spoof_gene_count > 0:
genome['warnings'] = genome.get('warnings', []) + \
[warnings['spoofed_genome'].format(self.spoof_gene_count)]
genome['suspect'] = 1
return genome
@staticmethod
def _location(in_feature):
in_feature['strand'] = in_feature['strand'].replace(
"-1", "-").translate(strand_table)
if in_feature['strand'] == '+':
start = in_feature['start']
elif in_feature['strand'] == '-':
start = in_feature['end']
else:
raise ValueError('Invalid feature strand: {}'.format(
in_feature['strand']))
return [
in_feature['contig'], start, in_feature['strand'],
in_feature['end'] - in_feature['start'] + 1
]
@staticmethod
def _validate_import_file_params(params):
"""
validate_import_file_params:
validates params passed to FastaGFFToGenome.import_file method
"""
# check for required parameters
for p in ['workspace_name', 'genome_name', 'fasta_file', 'gff_file']:
if p not in params:
raise ValueError(
'"{}" parameter is required, but missing'.format(p))
# one and only one of 'path', or 'shock_id' is required
for key in ('fasta_file', 'gff_file'):
file = params[key]
if not isinstance(file, dict):
raise ValueError(
'Required "{}" field must be a map/dict'.format(key))
n_valid_fields = 0
if 'path' in file and file['path'] is not None:
n_valid_fields += 1
if 'shock_id' in file and file['shock_id'] is not None:
n_valid_fields += 1
if 'ftp_url' in file and file['ftp_url'] is not None:
n_valid_fields += 1
raise ValueError(
'FTP link is currently not supported for FastaGFFToGenome')
if n_valid_fields < 1:
error_msg = 'Required "{}" field must include one source: '.format(
key)
error_msg += 'path | shock_id'
raise ValueError(error_msg)
if n_valid_fields > 1:
error_msg = 'Required "{}" field has too many sources specified: '.format(
key)
error_msg += str(list(file.keys()))
raise ValueError(error_msg)
if params.get('genetic_code'):
if not (isinstance(params['genetic_code'], int)
and 0 < params['genetic_code'] < 32):
raise ValueError(
"Invalid genetic code specified: {}".format(params))
def _set_parsed_params(self, params):
log('Setting params')
default_params = {
'taxon_wsname': self.cfg.raw['taxon-workspace-name'],
'scientific_name': 'unknown_taxon',
'taxon_reference': None,
'source': 'User',
'release': None,
'metadata': {},
'source_id': 'unknown',
}
default_params.update(params)
log(json.dumps(default_params, indent=1))
return default_params
def _stage_input(self, params, input_directory):
"""
stage_input: Setup the input_directory by fetching the files and uncompressing if needed
"""
file_paths = dict()
for key in ('fasta_file', 'gff_file'):
file = params[key]
file_path = None
if 'path' in file and file['path'] is not None:
local_file_path = file['path']
file_path = os.path.join(input_directory,
os.path.basename(local_file_path))
log('Moving file from {} to {}'.format(local_file_path,
file_path))
shutil.copy2(local_file_path, file_path)
if 'shock_id' in file and file['shock_id'] is not None:
# handle shock file
log('Downloading file from SHOCK node: {}-{}'.format(
self.cfg.sharedFolder, file['shock_id']))
sys.stdout.flush()
file_name = self.dfu.shock_to_file({
'file_path': input_directory,
'shock_id': file['shock_id']
})['node_file_name']
file_path = os.path.join(input_directory, file_name)
# extract the file if it is compressed
if file_path is not None:
log("staged input file =" + file_path)
sys.stdout.flush()
dfUtil_result = self.dfu.unpack_file({'file_path': file_path})
file_paths[key] = dfUtil_result['file_path']
else:
raise ValueError(
'No valid files could be extracted based on the input')
return file_paths
def _retrieve_gff_file(self, input_gff_file):
"""
_retrieve_gff_file: retrieve info from gff_file
"""
log("Reading GFF file")
feature_list = collections.defaultdict(list)
is_patric = 0
gff_file_handle = open(input_gff_file)
current_line = gff_file_handle.readline()
line_count = 0
while (current_line != ''):
current_line = current_line.strip()
if (current_line.isspace() or current_line == ""
or current_line.startswith("#")):
pass
else:
#Split line
(contig_id, source_id, feature_type, start, end, score, strand,
phase, attributes) = current_line.split('\t')
#Checking to see if Phytozome
if "phytozome" in source_id.lower():
self.is_phytozome = True
#Checking to see if Phytozome
if "PATRIC" in source_id:
is_patric = True
#PATRIC prepends their contig ids with some gibberish
if is_patric and "|" in contig_id:
contig_id = contig_id.split("|", 1)[1]
#Populating basic feature object
ftr = {
'contig': contig_id,
'source': source_id,
'type': feature_type,
'start': int(start),
'end': int(end),
'score': score,
'strand': strand,
'phase': phase,
'attributes': collections.defaultdict(list)
}
#Populating with attribute key-value pair
#This is where the feature id is from
for attribute in attributes.split(";"):
attribute = attribute.strip()
#Sometimes empty string
if not attribute:
continue
#Use of 1 to limit split as '=' character can also be made available later
#Sometimes lack of "=", assume spaces instead
if ("=" in attribute):
key, value = attribute.split("=", 1)
ftr['attributes'][key.lower()].append(
parse.unquote(value.strip('"')))
elif (" " in attribute):
key, value = attribute.split(" ", 1)
ftr['attributes'][key.lower()].append(
parse.unquote(value.strip('"')))
else:
pass
#log("Warning: attribute "+attribute+" cannot be separated into key,value pair")
ftr['attributes']['raw'] = attributes
if "id" in ftr['attributes']:
ftr['ID'] = ftr['attributes']['id'][0]
if "parent" in ftr['attributes']:
ftr['Parent'] = ftr['attributes']['parent'][0]
feature_list[contig_id].append(ftr)
current_line = gff_file_handle.readline()
gff_file_handle.close()
#Some GFF/GTF files don't use "ID" so we go through the possibilities
feature_list = self._add_missing_identifiers(feature_list)
#Most bacterial files have only CDSs
#In order to work with prokaryotic and eukaryotic gene structure synonymously
#Here we add feature dictionaries representing the parent gene and mRNAs
#feature_list = self._add_missing_parents(feature_list)
#Phytozome has the annoying habit of editing their identifiers so we fix them
if self.is_phytozome:
self._update_phytozome_features(feature_list)
#All identifiers need to be checked so that they follow the same general rules
#Rules are listed within the function itself
feature_list = self._update_identifiers(feature_list)
return feature_list
def _add_missing_identifiers(self, feature_list):
log("Adding missing identifiers")
#General rule is to iterate through a range of possibilities if "ID" is missing
for contig in feature_list:
for i, feat in enumerate(feature_list[contig]):
if "ID" not in feature_list[contig][i]:
for key in ("transcriptid", "proteinid", "pacid", "parent",
"name", 'transcript_id'):
if key in feature_list[contig][i]['attributes']:
feature_list[contig][i]['ID'] = feature_list[
contig][i]['attributes'][key][0]
break
if feat['type'] not in self.skip_types:
self.feature_counts[feat['type']] += 1
#If the process fails, throw an error
if "ID" not in feature_list[contig][i]:
feat['ID'] = "{}_{}".format(
feat['type'], self.feature_counts[feat['type']])
#log("Warning: Could find unique ID to utilize in GFF attributes: {}. "
# "ID '{}' has been assigned".format(feat['attributes'], feat['ID']))
return feature_list
def _add_missing_parents(self, feature_list):
#General rules is if CDS or RNA missing parent, add them
for contig in feature_list:
ftrs = feature_list[contig]
new_ftrs = []
for i in range(len(ftrs)):
if ftrs[i]["type"] in self.skip_types:
continue
if ("Parent" not in ftrs[i]):
#Assuming parent doesn't exist at all, so create de novo instead of trying to find it
if ("RNA" in ftrs[i]["type"] or "CDS" in ftrs[i]["type"]):
new_gene_ftr = copy.deepcopy(ftrs[i])
new_gene_ftr["type"] = "gene"
ftrs[i]["Parent"] = new_gene_ftr["ID"]
new_ftrs.append(new_gene_ftr)
if ("CDS" in ftrs[i]["type"]):
new_rna_ftr = copy.deepcopy(ftrs[i])
new_rna_ftr["type"] = "mRNA"
new_ftrs.append(new_rna_ftr)
ftrs[i]["Parent"] = new_rna_ftr["ID"]
new_ftrs.append(ftrs[i])
feature_list[contig] = new_ftrs
return feature_list
@staticmethod
def _update_phytozome_features(feature_list):
#General rule is to use the "Name" field where possible
#And update parent attribute correspondingly
for contig in feature_list:
feature_position_dict = {}
for i in range(len(feature_list[contig])):
#Maintain old_id for reference
#Sometimes ID isn't available, so use PACid
old_id = None
for key in ("id", "pacid"):
if (key in feature_list[contig][i]['attributes']):
old_id = feature_list[contig][i]['attributes'][key][0]
break
if (old_id is None):
#This should be an error
#log("Cannot find unique ID, PACid, or pacid in GFF "
# "attributes: " + feature_list[contig][i][contig])
continue
#Retain old_id
feature_position_dict[old_id] = i
# Clip off the increment on CDS IDs so fragments of the same
# CDS share the same ID
if "CDS" in feature_list[contig][i]["ID"]:
feature_list[contig][i]["ID"] = feature_list[contig][i][
"ID"].rsplit('.', 1)[0]
#In Phytozome, gene and mRNA have "Name" field, CDS do not
if ("name" in feature_list[contig][i]['attributes']):
feature_list[contig][i]["ID"] = feature_list[contig][i][
'attributes']['name'][0]
if ("Parent" in feature_list[contig][i]):
#Update Parent to match new ID of parent ftr
feature_list[contig][i]["Parent"] = feature_list[contig][
feature_position_dict[feature_list[contig][i]
["Parent"]]]["ID"]
return feature_list
def _update_identifiers(self, feature_list):
#General rules:
#1) Genes keep identifier
#2) RNAs keep identifier only if its different from gene, otherwise append ".mRNA"
#3) CDS always uses RNA identifier with ".CDS" appended
mRNA_parent_dict = dict()
for contig in feature_list:
for ftr in feature_list[contig]:
if ftr["type"] in self.skip_types:
continue
if ("Parent" in ftr):
#Retain old_id of parents
old_id = ftr["ID"]
if (ftr["ID"] == ftr["Parent"] or "CDS" in ftr["type"]):
ftr["ID"] = ftr["Parent"] + "." + ftr["type"]
#link old to new ids for mRNA to use with CDS
if ("RNA" in ftr["type"]):
mRNA_parent_dict[old_id] = ftr["ID"]
return feature_list
def _check_location_order(self, locations):
"""If order looks good return None.
If out of order return warning
If on multiple strands return warning"""
strand = None
last_start = 0
for location in locations:
if strand == None:
strand = location[2]
elif strand != location[2]:
return warnings["both_strand_coordinates"]
if strand == "-":
locations = reversed(locations)
for location in locations:
if last_start > location[1]:
return warnings["out_of_order"]
else:
last_start = location[1]
return None
def _create_ontology_event(self, ontology_type):
"""Creates the ontology_event if necessary
Returns the index of the ontology event back."""
if ontology_type not in self.ont_mappings:
raise ValueError(
"{} is not a supported ontology".format(ontology_type))
if "event_index" not in self.ont_mappings[ontology_type]:
self.ont_mappings[ontology_type]['event_index'] = len(
self.ontology_events)
if ontology_type == "GO":
ontology_ref = "KBaseOntology/gene_ontology"
elif ontology_type == "PO":
ontology_ref = "KBaseOntology/plant_ontology"
else:
ontology_ref = f"KBaseOntology/{ontology_type.lower()}_ontology"
self.ontology_events.append({
"method": "GenomeFileUtils Genbank uploader from annotations",
"method_version": self.version,
"timestamp": self.time_string,
"id": ontology_type,
"ontology_ref": ontology_ref
})
return self.ont_mappings[ontology_type]['event_index']
def _get_ontology_db_xrefs(self, feature):
"""Splits the ontology info from the other db_xrefs"""
ontology = collections.defaultdict(dict)
db_xrefs = []
# these are keys are formatted strangely and require special parsing
for key in ("go_process", "go_function", "go_component"):
ontology_event_index = self._create_ontology_event("GO")
for term in feature.get(key, []):
sp = term.split(" - ")
ontology['GO'][sp[0]] = [ontology_event_index]
self.ontologies_present['GO'][
sp[0]] = self.ont_mappings['GO'].get(sp[0], '')
# CATH terms are not distinct from EC numbers so myst be found by key
for term in feature.get('cath_funfam', []) + feature.get('cath', []):
for ref in term.split(','):
ontology['CATH'][ref] = [self._create_ontology_event("CATH")]
self.ontologies_present['CATH'][ref] = self.ont_mappings[
'CATH'].get(ref, '')
search_keys = [
'ontology_term', 'db_xref', 'dbxref', 'product_source', 'tigrfam',
'pfam', 'cog', 'go', 'po', 'ko'
]
ont_terms = []
# flatten out into list of values
for key in search_keys:
if key in feature:
ont_terms += [x for y in feature[key] for x in y.split(',')]
for ref in ont_terms:
if ref.startswith('GO:'):
ontology['GO'][ref] = [self._create_ontology_event("GO")]
self.ontologies_present['GO'][ref] = self.ont_mappings[
'GO'].get(ref, '')
elif ref.startswith('PO:'):
ontology['PO'][ref] = [self._create_ontology_event("PO")]
self.ontologies_present['PO'][ref] = self.ont_mappings[
'PO'].get(ref, '')
elif ref.startswith('KO:'):
ontology['KO'][ref] = [self._create_ontology_event("KO")]
self.ontologies_present['KO'][ref] = self.ont_mappings[
'KO'].get(ref, '')
elif ref.startswith('COG'):
ontology['COG'][ref] = [self._create_ontology_event("COG")]
self.ontologies_present['COG'][ref] = self.ont_mappings[
'COG'].get(ref, '')
elif ref.startswith('PF'):
ontology['PFAM'][ref] = [self._create_ontology_event("PFAM")]
self.ontologies_present['PFAM'][ref] = self.ont_mappings[
'PFAM'].get(ref, '')
elif ref.startswith('TIGR'):
ontology['TIGRFAM'][ref] = [
self._create_ontology_event("TIGRFAM")
]
self.ontologies_present['TIGRFAM'][ref] = self.ont_mappings[
'TIGRFAM'].get(ref, '')
else:
db_xrefs.append(tuple(ref.split(":", 1)))
return dict(ontology), db_xrefs
def _transform_feature(self, contig, in_feature):
"""Converts a feature from the gff ftr format into the appropriate
format for a genome object """
def _aliases(feat):
keys = ('locus_tag', 'old_locus_tag', 'protein_id',
'transcript_id', 'gene', 'ec_number', 'gene_synonym')
alias_list = []
for key in keys:
if key in feat['attributes']:
alias_list.extend([(key, val)
for val in feat['attributes'][key]])
return alias_list
if in_feature['start'] < 1 or in_feature['end'] > len(contig):
self.warn("Feature with invalid location for specified "
"contig: " + str(in_feature))
if self.strict:
raise ValueError(
"Features must be completely contained within the Contig in the "
"Fasta file. Feature: " + str(in_feature))
return
feat_seq = contig.seq[in_feature['start'] -
1:in_feature['end']].upper()
if in_feature['strand'] in {'-', '-1'}:
feat_seq = feat_seq.reverse_complement()
# if the feature ID is duplicated (CDS or transpliced gene) we only
# need to update the location and dna_sequence
if in_feature.get('ID') in self.feature_dict:
existing = self.feature_dict[in_feature['ID']]
existing['location'].append(self._location(in_feature))
existing['dna_sequence'] = existing.get('dna_sequence',
'') + str(feat_seq)
existing['dna_sequence_length'] = len(existing['dna_sequence'])
return
# The following is common to all the feature types
out_feat = {
"id": in_feature.get('ID'),
"type": in_feature['type'],
"location": [self._location(in_feature)],
"dna_sequence": str(feat_seq),
"dna_sequence_length": len(feat_seq),
"md5": hashlib.md5(str(feat_seq).encode('utf8')).hexdigest(),
}
# add optional fields
if 'note' in in_feature['attributes']:
out_feat['note'] = in_feature['attributes']["note"][0]
ont, db_xrefs = self._get_ontology_db_xrefs(in_feature['attributes'])
if ont:
out_feat['ontology_terms'] = ont
aliases = _aliases(in_feature)
if aliases:
out_feat['aliases'] = aliases
if db_xrefs:
out_feat['db_xrefs'] = db_xrefs
if 'product' in in_feature['attributes']:
out_feat['functions'] = in_feature['attributes']["product"]
if 'product_name' in in_feature['attributes']:
if "functions" in out_feat:
out_feat['functions'].extend(
in_feature['attributes']["product_name"])
else:
out_feat['functions'] = in_feature['attributes'][
"product_name"]
if 'function' in in_feature['attributes']:
out_feat['functional_descriptions'] = in_feature['attributes'][
"function"]
if 'inference' in in_feature['attributes']:
GenomeUtils.parse_inferences(in_feature['attributes']['inference'])
if 'trans-splicing' in in_feature['attributes'].get('exception', []):
out_feat['flags'] = out_feat.get('flags', []) + ['trans_splicing']
if 'pseudo' in in_feature['attributes'].get('exception', []):
out_feat['flags'] = out_feat.get('flags', []) + ['pseudo']
if 'ribosomal-slippage' in in_feature['attributes'].get(
'exception', []):
out_feat['flags'] = out_feat.get('flags',
[]) + ['ribosomal_slippage']
parent_id = in_feature.get('Parent', '')
if parent_id and parent_id not in self.feature_dict:
raise ValueError(
"Parent ID: {} was not found in feature ID list.".format(
parent_id))
# if the feature is a exon or UTR, it will only be used to update the
# location and sequence of it's parent, we add the info to it parent
# feature but not the feature dict
if in_feature['type'] in self.skip_types:
if parent_id and in_feature['type'] in {
'exon', 'five_prime_UTR', 'three_prime_UTR'
}:
parent = self.feature_dict[parent_id]
if in_feature['type'] not in parent:
parent[in_feature['type']] = []
parent[in_feature['type']].append(out_feat)
return
# add type specific features
elif 'gene' in in_feature['type']:
out_feat['protein_translation_length'] = 0
out_feat['cdss'] = []
elif in_feature['type'] == 'CDS':
if parent_id:
parent = self.feature_dict[parent_id]
if 'cdss' in parent: # parent must be a gene
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings[
"genes_CDS_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"] = out_feat.get('warnings', []) + [
warnings[
"CDS_fail_child_of_gene_coordinate_validation"]
.format(parent_id)
]
parent['cdss'].append(in_feature['ID'])
out_feat['parent_gene'] = parent_id
else: # parent must be mRNA
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings["mRNA_fail_parent_coordinate_validation"].
format(out_feat["id"])
]
out_feat["warnings"] = out_feat.get('warnings', []) + [
warnings[
"CDS_fail_child_of_mRNA_coordinate_validation"]
.format(parent_id)
]
parent['cds'] = in_feature['ID']
out_feat['parent_mrna'] = parent_id
parent_gene = self.feature_dict[parent['parent_gene']]
parent_gene['cdss'].append(in_feature['ID'])
out_feat['parent_gene'] = parent['parent_gene']
# keep track of CDSs for post processing
self.cdss.add(out_feat['id'])
elif in_feature['type'] == 'mRNA':
if parent_id:
parent = self.feature_dict[parent_id]
if 'mrnas' not in parent:
parent['mrnas'] = []
if 'cdss' in parent: # parent must be a gene
parent['mrnas'].append(in_feature['ID'])
out_feat['parent_gene'] = parent_id
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings["genes_mRNA_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"] = out_feat.get('warnings', []) + [
warnings["mRNAs_parent_gene_fails_location_validation"]
.format(parent_id)
]
else:
out_feat["type"] = in_feature['type']
# this prevents big misc_features from blowing up the genome size
if out_feat['dna_sequence_length'] > MAX_MISC_FEATURE_SIZE:
del out_feat['dna_sequence']
if parent_id:
parent = self.feature_dict[parent_id]
if 'children' not in parent:
parent['children'] = []
parent['children'].append(out_feat['id'])
out_feat['parent_gene'] = parent_id
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings[
"generic_parents_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"] = out_feat.get('warnings', []) + [
warnings[
"generic_childs_parent_fails_location_validation"].
format(parent_id)
]
self.feature_dict[out_feat['id']] = out_feat
def _process_cdss(self):
"""Because CDSs can have multiple fragments, it's necessary to go
back over them to calculate a final protein sequence"""
for cds_id in self.cdss:
cds = self.feature_dict[cds_id]
try:
prot_seq = str(
Seq(cds['dna_sequence']).translate(self.code_table,
cds=True).strip("*"))
except TranslationError as e:
cds['warnings'] = cds.get('warnings', []) + [str(e)]
prot_seq = ""
cds.update({
"protein_translation":
prot_seq,
"protein_md5":
hashlib.md5(prot_seq.encode('utf8')).hexdigest(),
"protein_translation_length":
len(prot_seq),
})
if 'parent_gene' in cds:
parent_gene = self.feature_dict[cds['parent_gene']]
# no propigation for now
propagate_cds_props_to_gene(cds, parent_gene)
elif self.generate_genes:
spoof = copy.copy(cds)
spoof['type'] = 'gene'
spoof['id'] = cds['id'] + "_gene"
spoof['cdss'] = [cds['id']]
spoof['warnings'] = [
warnings['spoofed_gene'].format(cds['id'])
]
self.feature_dict[spoof['id']] = spoof
cds['parent_gene'] = spoof['id']
self.spoof_gene_count += 1
else:
raise ValueError(warnings['no_spoof'])
self.feature_dict[cds['id']] = cds
def _update_from_exons(self, feature):
"""This function updates the sequence and location of a feature based
on it's UTRs, CDSs and exon information"""
# note that start and end here are in direction of translation
def start(loc):
return loc[0][1]
def end(loc):
if loc[-1][2] == "+":
return loc[-1][1] + loc[-1][3] + 1
else:
return loc[-1][1] - loc[-1][3] - 1
if 'exon' in feature:
# update the feature with the exon locations and sequences
feature['location'] = [x['location'][0] for x in feature['exon']]
feature['dna_sequence'] = "".join(x['dna_sequence']
for x in feature['exon'])
feature['dna_sequence_length'] = len(feature['dna_sequence'])
# construct feature location from utrs and cdss if present
elif 'cds' in feature:
cds = [copy.deepcopy(self.feature_dict[feature['cds']])]
locs = []
seq = ""
for frag in feature.get('five_prime_UTR', []) + cds + \
feature.get('three_prime_UTR', []):
# merge into last location if adjacent
if locs and abs(end(locs) - start(frag['location'])) == 1:
# extend the location length by the length of the first
# location in the fragment
first = frag['location'].pop(0)
locs[-1][3] += first[3]
locs.extend(frag['location'])
seq += frag['dna_sequence']
feature['location'] = locs
feature['dna_sequence'] = seq
feature['dna_sequence_length'] = len(seq)
# remove these properties as they are no longer needed
for x in ['five_prime_UTR', 'three_prime_UTR', 'exon']:
feature.pop(x, None)
else:
ValueError('Feature {} must contain either exon or cds data to '
'construct an accurate location and sequence'.format(
feature['id']))
def _gen_genome_info(self, core_genome_name, scientific_name, assembly_ref,
source, source_id, assembly, input_gff_file,
molecule_type):
"""
_gen_genome_info: generate genome info
"""
genome = dict()
genome["id"] = core_genome_name
genome["scientific_name"] = scientific_name
genome["assembly_ref"] = assembly_ref
genome['molecule_type'] = molecule_type
genome["features"] = []
genome["cdss"] = []
genome["mrnas"] = []
genome['non_coding_features'] = []
genome["gc_content"] = assembly["gc_content"]
genome["dna_size"] = assembly["dna_size"]
genome['md5'] = assembly['md5']
genome['contig_ids'], genome['contig_lengths'] = zip(
*[(k, v['length']) for k, v in assembly['contigs'].items()])
genome['num_contigs'] = len(assembly['contigs'])
genome['ontologies_present'] = dict(self.ontologies_present)
genome['ontology_events'] = self.ontology_events
genome['taxonomy'], genome['taxon_ref'], genome['domain'], \
genome["genetic_code"] = self.gi.retrieve_taxon(self.taxon_wsname,
genome['scientific_name'])
genome['source'], genome['genome_tiers'] = self.gi.determine_tier(
source)
genome['source_id'] = source_id
# Phytozome gff files are not compatible with the RNASeq Pipeline
# so it's better to build from the object than cache the file
if self.is_phytozome:
gff_file_to_shock = self.dfu.file_to_shock({
'file_path': input_gff_file,
'make_handle': 1,
'pack': "gzip"
})
genome['gff_handle_ref'] = gff_file_to_shock['handle']['hid']
for feature in self.feature_dict.values():
self.feature_counts[feature['type']] += 1
if 'exon' in feature or feature['type'] == 'mRNA':
self._update_from_exons(feature)
# Test if location order is in order.
is_transpliced = "flags" in feature and "trans_splicing" in feature[
"flags"]
if not is_transpliced and len(feature["location"]) > 1:
# Check the order only if not trans_spliced and has more than 1 location.
location_warning = self._check_location_order(
feature["location"])
if location_warning is not None:
feature["warnings"] = feature.get('warnings',
[]) + [location_warning]
contig_len = genome["contig_lengths"][genome["contig_ids"].index(
feature["location"][0][0])]
feature = check_full_contig_length_or_multi_strand_feature(
feature, is_transpliced, contig_len, self.skip_types)
# sort features into their respective arrays
if feature['type'] == 'CDS':
del feature['type']
genome['cdss'].append(feature)
elif feature['type'] == 'mRNA':
del feature['type']
genome['mrnas'].append(feature)
elif feature['type'] == 'gene':
# remove duplicates that may arise from CDS info propagation
for key in ('functions', 'aliases', 'db_xrefs'):
if key in feature:
feature[key] = list(set(feature[key]))
if feature['cdss']:
del feature['type']
self.feature_counts["protein_encoding_gene"] += 1
genome['features'].append(feature)
else:
feature.pop('mrnas', None)
feature.pop('cdss', None)
feature.pop('protein_translation_length', None)
self.feature_counts["non_coding_features"] += 1
genome['non_coding_features'].append(feature)
else:
genome['non_coding_features'].append(feature)
if self.warnings:
genome['warnings'] = self.warnings
genome['feature_counts'] = dict(self.feature_counts)
return genome
class FastaGFFToGenome:
def __init__(self, config):
self.cfg = config
self.au = AssemblyUtil(config.callbackURL)
self.dfu = DataFileUtil(self.cfg.callbackURL)
self.gi = GenomeInterface(self.cfg)
self.taxon_wsname = self.cfg.raw['taxon-workspace-name']
self.time_string = str(
datetime.datetime.fromtimestamp(
time.time()).strftime('%Y_%m_%d_%H_%M_%S'))
yml_text = open('/kb/module/kbase.yml').read()
mod_match = re.search(r'module-version:\n\W+(.+)\n', yml_text)
if mod_match:
self.version = mod_match.group(1)
else:
self.version = None
self.ont_mappings = load_ontology_mappings('/kb/module/data')
self.code_table = 11
self.skip_types = ('exon', 'five_prime_UTR', 'three_prime_UTR',
'start_codon', 'stop_codon', 'region', 'chromosome',
'scaffold')
self.spoof_gene_count = 0
self.is_phytozome = False
self.is_metagenome = False
self.strict = True
self.generate_genes = False
self.warnings = [] # type: list
self.feature_dict = collections.OrderedDict() # type: dict
self.cdss = set() # type: set
self.ontologies_present = collections.defaultdict(dict) # type: dict
self.ontology_events = list() # type: list
self.skiped_features = collections.Counter(
) # type: collections.Counter
self.feature_counts = collections.Counter(
) # type: collections.Counter
self.re_api_url = config.re_api_url
def warn(self, message):
self.warnings.append(message)
def generate_genome_json(self, params):
# 1) validate parameters
self._validate_import_file_params(params)
self.code_table = params.get('genetic_code', 11)
# 2) construct the input directory staging area
input_directory = os.path.join(self.cfg.sharedFolder,
'fast_gff_upload_' + str(uuid.uuid4()))
os.makedirs(input_directory)
file_paths = self._stage_input(params, input_directory)
# 3) extract out the parameters
params = self._set_parsed_params(params)
if params.get('generate_missing_genes'):
self.generate_genes = True
# 4) do the upload
genome = self._gen_genome_json(params, file_paths["gff_file"],
file_paths["fasta_file"])
return genome, input_directory
def import_file(self, params):
self.is_metagenome = params.get('is_metagenome', False)
if self.is_metagenome:
ws_datatype = "KBaseMetagenomes.AnnotatedMetagenomeAssembly"
else:
ws_datatype = "KBaseGenomes.Genome"
genome, input_directory = self.generate_genome_json(params)
json.dump(genome,
open(f"{self.cfg.sharedFolder}/{genome['id']}.json", 'w'),
indent=4)
result = self.gi.save_one_genome({
'workspace': params['workspace_name'],
'name': params['genome_name'],
'data': genome,
"meta": params.get('metadata', {}),
'workspace_datatype': ws_datatype,
})
feature_types = "\n".join(
[f"{k}: {v}" for k, v in genome['feature_counts'].items()])
report_string = (
f"A genome with {len(genome['contig_ids'])} contigs and the following feature "
f"types was imported: \n{feature_types}")
# XXX report_string is unused except for this log
logging.info(report_string)
# 5) clear the temp directory
shutil.rmtree(input_directory)
# 6) return the result
info = result['info']
prefix = ''
if self.is_metagenome:
prefix = 'meta'
details = {
prefix + 'genome_ref': f'{info[6]}/{info[0]}/{info[4]}',
prefix + 'genome_info': info
}
return details
def _gen_genome_json(self, params, input_gff_file, input_fasta_file):
# reading in GFF file
features_by_contig = self._retrieve_gff_file(input_gff_file)
contig_ids = set()
# parse feature information
fasta_contigs = Bio.SeqIO.parse(input_fasta_file, "fasta")
for contig in fasta_contigs:
molecule_type = str(contig.seq.alphabet).replace(
'IUPACAmbiguous', '').strip('()')
contig_ids.add(contig.id)
for feature in features_by_contig.get(contig.id, []):
self._transform_feature(contig, feature)
for cid in set(features_by_contig.keys()) - contig_ids:
self.warn(
f"Sequence name {cid} does not match a sequence id in the FASTA file."
f"{len(features_by_contig[cid])} features will not be imported."
)
if self.strict:
raise ValueError(
"Every feature sequence id must match a fasta sequence id")
prot_fasta_path = f"{self.cfg.sharedFolder}/{params['genome_name']}_protein.fasta"
# if is a metagenome, the following function writes a protein fasta
self._process_cdss(prot_fasta_path)
# save assembly file
'''
Metagenome Changes:
if we want to pass more stuff to AssemblyUtil, do here.
TODO: add flag to save_assembly_from_fasta
'''
if self.is_metagenome:
genome_type = "metagenome"
else:
genome_type = params.get('genome_type', 'isolate')
if params.get('existing_assembly_ref'):
assembly_ref = params['existing_assembly_ref']
ret = self.dfu.get_objects({'object_refs':
[assembly_ref]})['data'][0]
assembly_obj_type = ret['info'][2].split('-')[0]
valid_assembly_types = [
"KBaseGenomeAnnotations.Assembly", "KBaseGenomes.ContigSet"
]
if assembly_obj_type not in valid_assembly_types:
raise ValueError(
f"{assembly_ref} is not a reference to an assembly")
assembly_data = ret['data']
# should do more thorough check of sequences.
if not validate_lists_have_same_elements(
assembly_data['contigs'].keys(), contig_ids):
raise ValueError(
f"provided assembly with ref {assembly_ref} does not "
"have matching contig ids to provided input fasta.")
logging.info(f"Using supplied assembly: {assembly_ref}")
else:
assembly_ref = self.au.save_assembly_from_fasta({
'file': {
'path': input_fasta_file
},
'workspace_name':
params['workspace_name'],
'assembly_name':
params['genome_name'] + ".assembly",
'type':
genome_type,
})
assembly_data = self.dfu.get_objects({
'object_refs': [assembly_ref],
'ignore_errors': 0
})['data'][0]['data']
# generate genome info
genome = self._gen_genome_info(assembly_ref, assembly_data,
input_gff_file, molecule_type,
prot_fasta_path, params)
if self.spoof_gene_count > 0:
self.warn(warnings['spoofed_genome'].format(self.spoof_gene_count))
genome['suspect'] = 1
if self.warnings:
genome['warnings'] = self.warnings
return genome
@staticmethod
def _location(in_feature):
in_feature['strand'] = in_feature['strand'].replace(
"-1", "-").translate(strand_table)
if in_feature['strand'] == '+':
start = in_feature['start']
elif in_feature['strand'] == '-':
start = in_feature['end']
else:
raise ValueError('Invalid feature strand: {}'.format(
in_feature['strand']))
return [
in_feature['contig'], start, in_feature['strand'],
in_feature['end'] - in_feature['start'] + 1
]
@staticmethod
def _validate_import_file_params(params):
"""
validate_import_file_params:
validates params passed to FastaGFFToGenome.import_file method
"""
# check for required parameters
for p in ['workspace_name', 'genome_name', 'fasta_file', 'gff_file']:
if p not in params:
raise ValueError(f'"{p}" parameter is required, but missing')
# one and only one of 'path', or 'shock_id' is required
for key in ('fasta_file', 'gff_file'):
file = params[key]
if not isinstance(file, dict):
raise ValueError(f'Required "{key}" field must be a map/dict')
sources = ('path', 'shock_id')
n_valid_fields = sum(1 for f in sources if file.get(f))
print(f"inputs: {n_valid_fields}")
if n_valid_fields < 1:
raise ValueError(
f'Required "{key}" field must include one source: '
f'{", ".join(sources)}')
if n_valid_fields > 1:
raise ValueError(
f'Required "{key}" field has too many sources specified: '
f'{", ".join(file.keys())}')
if params.get('genetic_code'):
if not (isinstance(params['genetic_code'], int)
and 0 < params['genetic_code'] < 32):
raise ValueError(
"Invalid genetic code specified: {}".format(params))
def _set_parsed_params(self, params):
logging.info('Setting params')
default_params = {
'taxon_wsname': self.cfg.raw['taxon-workspace-name'],
'scientific_name': 'unknown_taxon',
'source': 'User',
'release': None,
'metadata': {},
'source_id': 'unknown',
}
default_params.update(params)
logging.info(json.dumps(default_params, indent=1))
return default_params
def _stage_input(self, params, input_directory):
"""
stage_input: Setup the input_directory by fetching the files and uncompressing if needed
"""
file_paths = dict()
for key in ('fasta_file', 'gff_file'):
file = params[key]
file_path = None
'''
below seems like weird if statement
'''
if file.get('path') is not None:
local_file_path = file['path']
file_path = os.path.join(input_directory,
os.path.basename(local_file_path))
logging.info(
f'Moving file from {local_file_path} to {file_path}')
# Metagenome Updates
# not sure if we have to be careful about moving the objects
# around
if os.path.isfile(local_file_path):
shutil.copy2(local_file_path, file_path)
else:
raise FileNotFoundError(
f"Input {key} file {local_file_path} not found")
err_msg = "Shutil copy unsucessful"
elif file.get('shock_id') is not None:
# handle shock file
logging.info(f'Downloading file from SHOCK node: '
f'{self.cfg.sharedFolder}-{file["shock_id"]}')
sys.stdout.flush()
file_name = self.dfu.shock_to_file({
'file_path': input_directory,
'shock_id': file['shock_id']
})['node_file_name']
file_path = os.path.join(input_directory, file_name)
err_msg = "Shock retrieval"
# extract the file if it is compressed
'''
Metagenome Changes:
may have to make check here to see if the the file is too big for
working dir.
'''
if file_path is not None:
logging.info("staged input file =" + file_path)
sys.stdout.flush()
if not os.path.isfile(file_path):
raise FileNotFoundError(f"{file_path} not a file")
dfUtil_result = self.dfu.unpack_file({'file_path': file_path})
file_paths[key] = dfUtil_result['file_path']
err_msg = "DataFielUtil 'unpack_file' function call"
else:
raise ValueError(
'No valid files could be extracted based on the input')
if not os.path.isfile(file_path):
raise ValueError(f"{err_msg} for {key} file to {file_path}")
return file_paths
def _retrieve_gff_file(self, input_gff_file):
"""
_retrieve_gff_file: retrieve info from gff_file
"""
logging.info("Reading GFF file")
feature_list = collections.defaultdict(list) # type: dict
is_patric = 0
'''
Metagenome Changes:
the lines below iterate through the entire gff input file, which
for a Metagenome may be an issue.
! Only a problem if there are space limits on processing in this
request
'''
for current_line in open(input_gff_file):
if current_line.isspace(
) or current_line == "" or current_line.startswith("#"):
continue
# Split line
try:
(contig_id, source_id, feature_type, start, end, score, strand,
phase, attributes) = current_line.split('\t')
except ValueError:
raise ValueError(f"unable to parse {current_line}")
''' Do Metagenomes need this phytozome/PATRIC stuff??'''
# Checking to see if Phytozome
if "phytozome" in source_id.lower():
self.is_phytozome = True
# Checking to see if Phytozome
if "PATRIC" in source_id:
is_patric = True
# PATRIC prepends their contig ids with some gibberish
if is_patric and "|" in contig_id:
contig_id = contig_id.split("|", 1)[1]
# Populating basic feature object
ftr: dict = {
'contig': contig_id,
'source': source_id,
'type': feature_type,
'start': int(start),
'end': int(end),
'score': score,
'strand': strand,
'phase': phase,
'attributes': collections.defaultdict(list)
}
# Populating with attribute key-value pair
# This is where the feature id is from
for attribute in attributes.split(";"):
attribute = attribute.strip()
# Sometimes empty string
if not attribute:
continue
# Use of 1 to limit split as '=' character can also be made available later
# Sometimes lack of "=", assume spaces instead
if "=" in attribute:
key, value = attribute.split("=", 1)
elif " " in attribute:
key, value = attribute.split(" ", 1)
else:
logging.debug(f'Unable to parse {attribute}')
continue
ftr['attributes'][make_snake_case(key)].append(
parse.unquote(value.strip('"')))
ftr['attributes']['raw'] = attributes
if "id" in ftr['attributes']:
ftr['ID'] = ftr['attributes']['id'][0]
if "parent" in ftr['attributes']:
ftr['Parent'] = ftr['attributes']['parent'][0]
feature_list[contig_id].append(ftr)
# Some GFF/GTF files don't use "ID" so we go through the possibilities
feature_list = self._add_missing_identifiers(feature_list)
# Most bacterial files have only CDSs
# In order to work with prokaryotic and eukaryotic gene structure synonymously
# Here we add feature dictionaries representing the parent gene and mRNAs
# feature_list = self._add_missing_parents(feature_list)
# Phytozome has the annoying habit of editing their identifiers so we fix them
if self.is_phytozome:
self._update_phytozome_features(feature_list)
# All identifiers need to be checked so that they follow the same general rules
# Rules are listed within the function itself
feature_list = self._update_identifiers(feature_list)
return feature_list
def _add_missing_identifiers(self, feature_list):
logging.info("Adding missing identifiers")
# General rule is to iterate through a range of possibilities if "ID" is missing
for contig in feature_list:
for i, feat in enumerate(feature_list[contig]):
if "ID" not in feature_list[contig][i]:
# all of the following are not guaranteed to be unique ID's
# for key in ("transcriptid", "proteinid", "pacid",
# "parent", "name", 'transcript_id'):
for key in ("protein_id", "name", "pacid", "parent"):
if key in feature_list[contig][i]['attributes']:
feature_list[contig][i]['ID'] = feature_list[
contig][i]['attributes'][key][0]
break
if feat['type'] not in self.skip_types:
self.feature_counts[feat['type']] += 1
# If the process fails, throw an error
if "ID" not in feature_list[contig][i]:
feat[
'ID'] = f"{feat['type']}_{self.feature_counts[feat['type']]}"
return feature_list
def _add_missing_parents(self, feature_list):
# General rules is if CDS or RNA missing parent, add them
for contig in feature_list:
ftrs = feature_list[contig]
new_ftrs = []
for i in range(len(ftrs)):
if ftrs[i]["type"] in self.skip_types:
continue
if "Parent" not in ftrs[i]:
# Assuming parent doesn't exist at all, so create de novo instead of trying to find it
if "RNA" in ftrs[i]["type"] or "CDS" in ftrs[i]["type"]:
new_gene_ftr = copy.deepcopy(ftrs[i])
new_gene_ftr["type"] = "gene"
ftrs[i]["Parent"] = new_gene_ftr["ID"]
new_ftrs.append(new_gene_ftr)
if "CDS" in ftrs[i]["type"]:
new_rna_ftr = copy.deepcopy(ftrs[i])
new_rna_ftr["type"] = "mRNA"
new_ftrs.append(new_rna_ftr)
ftrs[i]["Parent"] = new_rna_ftr["ID"]
new_ftrs.append(ftrs[i])
feature_list[contig] = new_ftrs
return feature_list
@staticmethod
def _update_phytozome_features(feature_list):
# General rule is to use the "Name" field where possible
# And update parent attribute correspondingly
for contig in feature_list:
feature_position_dict = {}
for i in range(len(feature_list[contig])):
# Maintain old_id for reference
# Sometimes ID isn't available, so use PACid
old_id = None
for key in ("id", "pacid"):
if key in feature_list[contig][i]['attributes']:
old_id = feature_list[contig][i]['attributes'][key][0]
break
if old_id is None:
continue
# Retain old_id
feature_position_dict[old_id] = i
# Clip off the increment on CDS IDs so fragments of the same
# CDS share the same ID
if "CDS" in feature_list[contig][i]["ID"]:
feature_list[contig][i]["ID"] = feature_list[contig][i][
"ID"].rsplit('.', 1)[0]
# In Phytozome, gene and mRNA have "Name" field, CDS do not
if "name" in feature_list[contig][i]['attributes']:
feature_list[contig][i]["ID"] = feature_list[contig][i][
'attributes']['name'][0]
if "Parent" in feature_list[contig][i]:
# Update Parent to match new ID of parent ftr
feature_list[contig][i]["Parent"] = feature_list[contig][
feature_position_dict[feature_list[contig][i]
["Parent"]]]["ID"]
return feature_list
def _update_identifiers(self, feature_list):
# General rules:
# 1) Genes keep identifier
# 2) RNAs keep identifier only if its different from gene, otherwise append ".mRNA"
# 3) CDS always uses RNA identifier with ".CDS" appended
mRNA_parent_dict = dict()
for contig in feature_list:
for ftr in feature_list[contig]:
if ftr["type"] in self.skip_types:
continue
if "Parent" in ftr:
# Retain old_id of parents
old_id = ftr["ID"]
if ftr["ID"] == ftr["Parent"] or "CDS" in ftr["type"]:
ftr["ID"] = ftr["Parent"] + "." + ftr["type"]
# link old to new ids for mRNA to use with CDS
if "RNA" in ftr["type"]:
mRNA_parent_dict[old_id] = ftr["ID"]
return feature_list
def _check_location_order(self, locations):
"""If order looks good return None.
If out of order return warning
If on multiple strands return warning"""
strand = None
last_start = 0
for location in locations:
if strand is None:
strand = location[2]
elif strand != location[2]:
return warnings["both_strand_coordinates"]
if strand == "-":
locations = reversed(locations)
for location in locations:
if last_start > location[1]:
return warnings["out_of_order"]
else:
last_start = location[1]
return None
def _create_ontology_event(self, ontology_type):
"""Creates the ontology_event if necessary
Returns the index of the ontology event back."""
if ontology_type not in self.ont_mappings:
raise ValueError(
"{} is not a supported ontology".format(ontology_type))
if "event_index" not in self.ont_mappings[ontology_type]:
self.ont_mappings[ontology_type]['event_index'] = len(
self.ontology_events)
if ontology_type == "GO":
ontology_ref = "KBaseOntology/gene_ontology"
elif ontology_type == "PO":
ontology_ref = "KBaseOntology/plant_ontology"
else:
ontology_ref = f"KBaseOntology/{ontology_type.lower()}_ontology"
self.ontology_events.append({
"method": "GenomeFileUtils Genbank uploader from annotations",
"method_version": self.version,
"timestamp": self.time_string,
"id": ontology_type,
"ontology_ref": ontology_ref
})
return self.ont_mappings[ontology_type]['event_index']
def _get_ontology_db_xrefs(self, feature):
"""Splits the ontology info from the other db_xrefs"""
ontology = collections.defaultdict(dict) # type: dict
db_xrefs = []
# these are keys are formatted strangely and require special parsing
for key in ("go_process", "go_function", "go_component"):
ontology_event_index = self._create_ontology_event("GO")
for term in feature.get(key, []):
sp = term.split(" - ")
ontology['GO'][sp[0]] = [ontology_event_index]
self.ontologies_present['GO'][
sp[0]] = self.ont_mappings['GO'].get(sp[0], '')
# CATH terms are not distinct from EC numbers so myst be found by key
for term in feature.get('cath_funfam', []) + feature.get('cath', []):
for ref in term.split(','):
ontology['CATH'][ref] = [self._create_ontology_event("CATH")]
self.ontologies_present['CATH'][ref] = self.ont_mappings[
'CATH'].get(ref, '')
search_keys = [
'ontology_term', 'db_xref', 'dbxref', 'product_source', 'tigrfam',
'pfam', 'cog', 'go', 'po', 'ko'
]
ont_terms = [] # type: list
# flatten out into list of values
for key in search_keys:
if key in feature:
ont_terms += [x for y in feature[key] for x in y.split(',')]
for ref in ont_terms:
if ref.startswith('GO:'):
ontology['GO'][ref] = [self._create_ontology_event("GO")]
self.ontologies_present['GO'][ref] = self.ont_mappings[
'GO'].get(ref, '')
elif ref.startswith('PO:'):
ontology['PO'][ref] = [self._create_ontology_event("PO")]
self.ontologies_present['PO'][ref] = self.ont_mappings[
'PO'].get(ref, '')
elif ref.startswith('KO:'):
ontology['KO'][ref] = [self._create_ontology_event("KO")]
self.ontologies_present['KO'][ref] = self.ont_mappings[
'KO'].get(ref, '')
elif ref.startswith('COG'):
ontology['COG'][ref] = [self._create_ontology_event("COG")]
self.ontologies_present['COG'][ref] = self.ont_mappings[
'COG'].get(ref, '')
elif ref.startswith('PF'):
ontology['PFAM'][ref] = [self._create_ontology_event("PFAM")]
self.ontologies_present['PFAM'][ref] = self.ont_mappings[
'PFAM'].get(ref, '')
elif ref.startswith('TIGR'):
ontology['TIGRFAM'][ref] = [
self._create_ontology_event("TIGRFAM")
]
self.ontologies_present['TIGRFAM'][ref] = self.ont_mappings[
'TIGRFAM'].get(ref, '')
elif ":" not in ref:
db_xrefs.append(tuple(["Unknown_Source", ref]))
else:
db_xrefs.append(tuple(ref.split(":", 1)))
return dict(ontology), db_xrefs
'''
Metagenome Changes:
okay looks like this might be the real meat of it
'''
def _transform_feature(self, contig, in_feature):
"""Converts a feature from the gff ftr format into the appropriate
format for a genome object """
def _aliases(feat):
keys = ('locus_tag', 'old_locus_tag', 'protein_id',
'transcript_id', 'gene', 'ec_number', 'gene_synonym')
alias_list = []
for key in keys:
if key in feat['attributes']:
alias_list.extend([(key, val)
for val in feat['attributes'][key]])
return alias_list
if in_feature['start'] < 1 or in_feature['end'] > len(contig):
self.warn(
f"Feature with invalid location for specified contig: {in_feature}"
)
if self.strict:
raise ValueError(
"Features must be completely contained within the Contig in the "
f"Fasta file. Feature: in_feature")
return
feat_seq = contig.seq[in_feature['start'] -
1:in_feature['end']].upper()
if in_feature['strand'] in {'-', '-1'}:
feat_seq = feat_seq.reverse_complement()
# if the feature ID is duplicated (CDS or transpliced gene) we only
# need to update the location and dna_sequence
if in_feature.get('ID') in self.feature_dict:
existing = self.feature_dict[in_feature['ID']]
existing['location'].append(self._location(in_feature))
existing['dna_sequence'] = existing.get('dna_sequence',
'') + str(feat_seq)
existing['dna_sequence_length'] = len(existing['dna_sequence'])
return
# The following is common to all the feature types
out_feat = {
"id": in_feature.get('ID'),
"type": in_feature['type'],
"location": [self._location(in_feature)],
"dna_sequence": str(feat_seq),
"dna_sequence_length": len(feat_seq),
"md5": hashlib.md5(str(feat_seq).encode('utf8')).hexdigest(),
"warnings": [],
"flags": [],
}
# add optional fields
if 'note' in in_feature['attributes']:
out_feat['note'] = in_feature['attributes']["note"][0]
ont, db_xrefs = self._get_ontology_db_xrefs(in_feature['attributes'])
if ont:
out_feat['ontology_terms'] = ont
aliases = _aliases(in_feature)
if aliases:
out_feat['aliases'] = aliases
if db_xrefs:
out_feat['db_xrefs'] = db_xrefs
if 'product' in in_feature['attributes']:
out_feat['functions'] = in_feature['attributes']["product"]
if 'product_name' in in_feature['attributes']:
if "functions" in out_feat:
out_feat['functions'].extend(
in_feature['attributes']["product_name"])
else:
out_feat['functions'] = in_feature['attributes'][
"product_name"]
if 'function' in in_feature['attributes']:
out_feat['functional_descriptions'] = in_feature['attributes'][
"function"]
if 'inference' in in_feature['attributes']:
GenomeUtils.parse_inferences(in_feature['attributes']['inference'])
if 'trans-splicing' in in_feature['attributes'].get('exception', []):
out_feat['flags'].append('trans_splicing')
if 'pseudo' in in_feature['attributes'].get('exception', []):
out_feat['flags'].append('pseudo')
if 'ribosomal-slippage' in in_feature['attributes'].get(
'exception', []):
out_feat['flags'].append('ribosomal_slippage')
parent_id = in_feature.get('Parent', '')
if parent_id and parent_id not in self.feature_dict:
raise ValueError(
f"Parent ID: {parent_id} was not found in feature ID list.")
# if the feature is a exon or UTR, it will only be used to update the
# location and sequence of it's parent, we add the info to it parent
# feature but not the feature dict
if in_feature['type'] in self.skip_types:
if parent_id and in_feature['type'] in {
'exon', 'five_prime_UTR', 'three_prime_UTR'
}:
parent = self.feature_dict[parent_id]
if in_feature['type'] not in parent:
parent[in_feature['type']] = []
parent[in_feature['type']].append(out_feat)
return
# add type specific features
elif 'gene' in in_feature['type']:
out_feat['protein_translation_length'] = 0
out_feat['cdss'] = []
elif in_feature['type'] == 'CDS':
if parent_id:
parent = self.feature_dict[parent_id]
if 'cdss' in parent: # parent must be a gene
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings[
"genes_CDS_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"].append(warnings[
"CDS_fail_child_of_gene_coordinate_validation"].
format(parent_id))
parent['cdss'].append(in_feature['ID'])
out_feat['parent_gene'] = parent_id
else: # parent must be mRNA
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings["mRNA_fail_parent_coordinate_validation"].
format(out_feat["id"])
]
out_feat["warnings"].append(warnings[
"CDS_fail_child_of_mRNA_coordinate_validation"].
format(parent_id))
parent['cds'] = in_feature['ID']
out_feat['parent_mrna'] = parent_id
parent_gene = self.feature_dict[parent['parent_gene']]
parent_gene['cdss'].append(in_feature['ID'])
out_feat['parent_gene'] = parent['parent_gene']
# keep track of CDSs for post processing
self.cdss.add(out_feat['id'])
elif in_feature['type'] == 'mRNA':
if parent_id:
parent = self.feature_dict[parent_id]
if 'mrnas' not in parent:
parent['mrnas'] = []
if 'cdss' in parent: # parent must be a gene
parent['mrnas'].append(in_feature['ID'])
out_feat['parent_gene'] = parent_id
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings["genes_mRNA_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"].append(
warnings["mRNAs_parent_gene_fails_location_validation"]
.format(parent_id))
else:
out_feat["type"] = in_feature['type']
# this prevents big misc_features from blowing up the genome size
if out_feat['dna_sequence_length'] > MAX_MISC_FEATURE_SIZE:
del out_feat['dna_sequence']
if parent_id:
parent = self.feature_dict[parent_id]
if 'children' not in parent:
parent['children'] = []
parent['children'].append(out_feat['id'])
out_feat['parent_gene'] = parent_id
if not is_parent(parent, out_feat):
parent["warnings"] = parent.get('warnings', []) + [
warnings[
"generic_parents_child_fails_location_validation"].
format(out_feat["id"])
]
out_feat["warnings"].append(warnings[
"generic_childs_parent_fails_location_validation"].
format(parent_id))
# cleanup empty optional arrays
for key in ['warnings', 'flags']:
if not out_feat[key]:
del out_feat[key]
self.feature_dict[out_feat['id']] = out_feat
def _process_cdss(self, prot_fasta_path):
"""Because CDSs can have multiple fragments, it's necessary to go
back over them to calculate a final protein sequence"""
if self.is_metagenome:
prot_fasta = {} # type: dict
untranslatable_prot = set()
for cds_id in self.cdss:
cds = self.feature_dict[cds_id]
try:
prot_seq = str(
Seq(cds['dna_sequence']).translate(self.code_table,
cds=True).strip("*"))
except TranslationError as e:
cds['warnings'] = cds.get('warnings', []) + [str(e)]
# NOTE: we may need a different way of handling this for metagenomes.
prot_seq = ""
if self.is_metagenome:
untranslatable_prot.add(cds_id)
if self.is_metagenome:
if prot_seq != "":
protein_id = ""
if cds.get("aliases"):
aliases = cds['aliases']
for key, val in aliases:
if key == "protein_id":
protein_id = val
if not protein_id:
protein_id = cds['id'] # assign to some default
else:
# log a warning here?
pass
# TODO: update header to reflect what we actually want people
# to see.
if protein_id in prot_fasta:
prot_fasta[protein_id][0] += "|" + cds['id']
else:
fasta_seq_data = ">" + protein_id + " cds_ids:" + cds[
'id']
prot_fasta[protein_id] = [fasta_seq_data, prot_seq]
else:
pass
else:
cds.update({
"protein_translation":
prot_seq,
"protein_md5":
hashlib.md5(prot_seq.encode('utf8')).hexdigest(),
"protein_translation_length":
len(prot_seq),
})
if 'parent_gene' in cds:
parent_gene = self.feature_dict[cds['parent_gene']]
# no propigation for now
propagate_cds_props_to_gene(cds, parent_gene,
self.is_metagenome)
elif self.generate_genes:
spoof = copy.copy(cds)
spoof['type'] = 'gene'
spoof['id'] = cds['id'] + "_gene"
spoof['cdss'] = [cds['id']]
spoof['warnings'] = [
warnings['spoofed_gene'].format(cds['id'])
]
self.feature_dict[spoof['id']] = spoof
cds['parent_gene'] = spoof['id']
self.spoof_gene_count += 1
else:
raise ValueError(warnings['no_spoof'])
self.feature_dict[cds['id']] = cds
if self.is_metagenome:
with open(prot_fasta_path, 'w') as fid:
for key, line in prot_fasta.items():
fid.write('\n'.join(line))
# do something with 'untranslatable_prot'
def _update_from_exons(self, feature):
"""This function updates the sequence and location of a feature based
on it's UTRs, CDSs and exon information"""
# note that start and end here are in direction of translation
def start(loc):
return loc[0][1]
def end(loc):
if loc[-1][2] == "+":
return loc[-1][1] + loc[-1][3] + 1
else:
return loc[-1][1] - loc[-1][3] - 1
if 'exon' in feature:
# update the feature with the exon locations and sequences
feature['location'] = [x['location'][0] for x in feature['exon']]
feature['dna_sequence'] = "".join(x['dna_sequence']
for x in feature['exon'])
feature['dna_sequence_length'] = len(feature['dna_sequence'])
# construct feature location from utrs and cdss if present
elif 'cds' in feature:
cds = [copy.deepcopy(self.feature_dict[feature['cds']])]
locs = [] # type: list
seq = ""
for frag in feature.get('five_prime_UTR', []) + cds + \
feature.get('three_prime_UTR', []):
# merge into last location if adjacent
if locs and abs(end(locs) - start(frag['location'])) == 1:
# extend the location length by the length of the first
# location in the fragment
first = frag['location'].pop(0)
locs[-1][3] += first[3]
locs.extend(frag['location'])
seq += frag['dna_sequence']
feature['location'] = locs
feature['dna_sequence'] = seq
feature['dna_sequence_length'] = len(seq)
# remove these properties as they are no longer needed
for x in ['five_prime_UTR', 'three_prime_UTR', 'exon']:
feature.pop(x, None)
else:
ValueError(
'Feature {feature["id"]} must contain either exon or cds data to '
'construct an accurate location and sequence')
def _gen_genome_info(self, assembly_ref, assembly, input_gff_file,
molecule_type, prot_fasta_path, params):
"""
_gen_genome_info: generate genome info
Here is the meat of the saving operation.
Genome Fields:
features: protein encoding genes
cdss:
mrnas: mrna sequences
non_coding_features: everything that doesn't fall into 'features',
'cdss', 'mrnas'
"""
features = []
cdss = []
mrnas = []
non_coding_features = []
genome = {
"id": params.get('genome_name'),
"scientific_name": params.get('scientific_name', "Unknown"),
"assembly_ref": assembly_ref,
'molecule_type': molecule_type,
"gc_content": assembly["gc_content"],
"dna_size": assembly["dna_size"],
'md5': assembly['md5'],
'num_contigs': len(assembly['contigs']),
'ontologies_present': dict(self.ontologies_present),
'ontology_events': self.ontology_events,
}
if self.is_metagenome:
metagenome_fields = [
("publications", []),
("external_source_origination_date", None),
("original_source_file_name", None),
("notes", None),
# NOTE: in the future environment should use an ontology.
("environment", None),
] # type: list
for field, default in metagenome_fields:
genome[field] = params.get(field, default)
# save protein fasta to shock
prot_to_shock = self.dfu.file_to_shock({
'file_path': prot_fasta_path,
'make_handle': 1,
'pack': 'gzip'
})
genome['protein_handle_ref'] = prot_to_shock['handle']['hid']
genome['contig_ids'], genome['contig_lengths'] = zip(
*[(k, v['length']) for k, v in assembly['contigs'].items()])
if self.is_metagenome:
genome['source'], _ = self.gi.determine_tier(params.get('source'))
else:
genome['source'], genome['genome_tiers'] = self.gi.determine_tier(
params.get('source'))
# Set taxonomy-related fields in the genome data
if params.get('taxon_id'):
GenomeUtils.set_taxon_data(int(params['taxon_id']),
self.re_api_url, genome)
else:
GenomeUtils.set_default_taxon_data(genome)
# handle optional fields
for key in ('release', 'genetic_code', 'genome_type', 'source_id'):
if params.get(key):
genome[key] = params[key]
# Phytozome gff files are not compatible with the RNASeq Pipeline
# so it's better to build from the object than cache the file
if self.is_phytozome or self.is_metagenome:
gff_file_to_shock = self.dfu.file_to_shock({
'file_path': input_gff_file,
'make_handle': 1,
'pack': "gzip"
})
genome['gff_handle_ref'] = gff_file_to_shock['handle']['hid']
for feature in self.feature_dict.values():
self.feature_counts[feature['type']] += 1
if 'exon' in feature or feature['type'] == 'mRNA':
self._update_from_exons(feature)
# Test if location order is in order.
is_transpliced = "flags" in feature and "trans_splicing" in feature[
"flags"]
if not is_transpliced and len(feature["location"]) > 1:
# Check the order only if not trans_spliced and has more than 1 location.
location_warning = self._check_location_order(
feature["location"])
if location_warning is not None:
feature["warnings"] = feature.get('warnings',
[]) + [location_warning]
contig_len = genome["contig_lengths"][genome["contig_ids"].index(
feature["location"][0][0])]
feature = check_full_contig_length_or_multi_strand_feature(
feature, is_transpliced, contig_len, self.skip_types)
# sort features into their respective arrays
if feature['type'] == 'CDS':
if not self.is_metagenome:
del feature['type']
cdss.append(feature)
elif feature['type'] == 'mRNA':
if not self.is_metagenome:
del feature['type']
mrnas.append(feature)
elif feature['type'] == 'gene':
# remove duplicates that may arise from CDS info propagation
for key in ('functions', 'aliases', 'db_xrefs'):
if key in feature:
feature[key] = list(set(feature[key]))
if feature['cdss']:
if not self.is_metagenome:
del feature['type']
self.feature_counts["protein_encoding_gene"] += 1
features.append(feature)
else:
feature.pop('mrnas', None)
feature.pop('cdss', None)
feature.pop('protein_translation_length', None)
self.feature_counts["non_coding_gene"] += 1
non_coding_features.append(feature)
else:
non_coding_features.append(feature)
# if input is metagenome, save features, cdss, non_coding_features, and
# mrnas to shock
if self.is_metagenome:
# TODO: make this section more efficient by editing the above.
metagenome_features = features + cdss + mrnas + non_coding_features
genome['num_features'] = len(metagenome_features)
genome_name = params['genome_name']
json_file_path = f'{self.cfg.sharedFolder}/{genome_name}_features.json'
# save to json files first
with open(json_file_path, 'w') as fid:
json.dump(metagenome_features, fid)
# write json to shock
json_to_shock = self.dfu.file_to_shock({
'file_path': json_file_path,
'make_handle': 1,
'pack': 'gzip'
})
self.feature_counts["non_coding_features"] = len(
non_coding_features)
genome['features_handle_ref'] = json_to_shock['handle']['hid']
# remove json file to avoid disk overload
os.remove(json_file_path)
# delete python objects to reduce overhead
del metagenome_features
del features, cdss, mrnas, non_coding_features
else:
# TODO determine whether we want to deepcopy here instead of reference.
genome['features'] = features
genome['cdss'] = cdss
genome['mrnas'] = mrnas
genome['non_coding_features'] = non_coding_features
self.feature_counts["non_coding_features"] = len(
genome['non_coding_features'])
if self.warnings:
genome['warnings'] = self.warnings
genome['feature_counts'] = dict(self.feature_counts)
return genome