def create_matched_genomes():
"""Create list of GenomePair objects."""
gnm1 = genome.Genome()
gnm1.id = "Trixie"
gnm1.annotation_status = "draft"
gnm2 = genome.Genome()
gnm2.id = "Trixie"
gnm_pair1 = genomepair.GenomePair()
gnm_pair1.genome1 = gnm1
gnm_pair1.genome2 = gnm2
gnm3 = genome.Genome()
gnm3.id = "Alice"
gnm3.annotation_status = "final"
gnm4 = genome.Genome()
gnm4.id = "Alice"
gnm_pair2 = genomepair.GenomePair()
gnm_pair2.genome1 = gnm3
gnm_pair2.genome2 = gnm4
matched_genomes = [gnm_pair1, gnm_pair2]
return matched_genomes
def setUp(self):
self.ticket1 = ticket.ImportTicket()
self.src1 = source.Source()
self.src1.id = "L5_SRC_1"
self.src2 = source.Source()
self.src2.id = "L5_SRC_2"
self.src3 = source.Source()
self.src3.id = "L5_SRC_3"
self.cds1 = cds.Cds()
self.cds1.id = "L5_CDS_1"
self.cds2 = cds.Cds()
self.cds2.id = "L5_CDS_2"
self.cds3 = cds.Cds()
self.cds3.id = "L5_CDS_3"
self.trna1 = trna.Trna()
self.trna1.id = "L5_TRNA_1"
self.trna2 = trna.Trna()
self.trna2.id = "L5_TRNA_2"
self.trna3 = trna.Trna()
self.trna3.id = "L5_TRNA_3"
self.tmrna1 = tmrna.Tmrna()
self.tmrna1.id = "L5_TMRNA_1"
self.tmrna2 = tmrna.Tmrna()
self.tmrna2.id = "L5_TMRNA_2"
self.tmrna3 = tmrna.Tmrna()
self.tmrna3.id = "L5_TMRNA_3"
self.genome1 = genome.Genome()
self.genome1.type = "flat_file"
self.genome1.cds_features = [self.cds1, self.cds2]
self.genome1.source_features = [self.src1, self.src2]
self.genome1.trna_features = [self.trna1, self.trna2]
self.genome1.tmrna_features = [self.tmrna1, self.tmrna2]
self.genome2 = genome.Genome()
self.genome2.type = "mysql"
self.genome_pair1 = genomepair.GenomePair()
self.genome_pair2 = genomepair.GenomePair()
self.bndl = bundle.Bundle()
self.bndl.ticket = self.ticket1
self.bndl.genome_dict[self.genome1.type] = self.genome1
self.bndl.genome_dict[self.genome2.type] = self.genome2
self.bndl.genome_pair_dict["genome_pair1"] = self.genome_pair1
self.bndl.genome_pair_dict["genome_pair2"] = self.genome_pair2
self.eval_correct1 = evaluation.Evaluation(status="correct")
self.eval_correct2 = evaluation.Evaluation(status="correct")
self.eval_error1 = evaluation.Evaluation(status="error")
self.eval_error2 = evaluation.Evaluation(status="error")
def test_set_genome_pair_2(self):
"""Check that a genome pair is not set if one key is not present."""
self.bndl.ticket = self.tkt
self.bndl.genome_dict[self.genome1.type] = self.genome1
self.bndl.genome_dict[self.genome2.type] = self.genome2
genome_pair = genomepair.GenomePair()
self.bndl.set_genome_pair(genome_pair, "invalid", "flat_file")
self.assertEqual(len(self.bndl.genome_pair_dict.keys()), 0)
def test_set_genome_pair_1(self):
"""Check that a genome pair is set if both keys are present."""
self.bndl.ticket = self.tkt
self.bndl.genome_dict[self.genome1.type] = self.genome1
self.bndl.genome_dict[self.genome2.type] = self.genome2
genome_pair = genomepair.GenomePair()
self.bndl.set_genome_pair(genome_pair, "mysql", "flat_file")
self.assertEqual(
list(self.bndl.genome_pair_dict.keys())[0], "mysql_flat_file")
def setUp(self):
self.genome1 = genome.Genome()
self.genome2 = genome.Genome()
self.tkt = ticket.ImportTicket()
self.genome_pair = genomepair.GenomePair()
self.genome_pair.genome1 = self.genome1
self.genome_pair.genome2 = self.genome2
self.date_jan1 = datetime.strptime('1/1/2000', '%m/%d/%Y')
self.date_feb1 = datetime.strptime('2/1/2000', '%m/%d/%Y')
self.date_feb1_b = datetime.strptime('2/1/2000', '%m/%d/%Y')
def setUp(self):
self.ticket1 = ticket.ImportTicket()
self.src1 = source.Source()
self.src1.id = "L5_SRC_1"
self.src2 = source.Source()
self.src2.id = "L5_SRC_2"
self.src3 = source.Source()
self.src3.id = "L5_SRC_3"
self.cds1 = cds.Cds()
self.cds1.id = "L5_CDS_1"
self.cds2 = cds.Cds()
self.cds2.id = "L5_CDS_2"
self.cds3 = cds.Cds()
self.cds3.id = "L5_CDS_3"
self.genome1 = genome.Genome()
self.genome1.type = "flat_file"
self.genome1.cds_features.append(self.cds1)
self.genome1.cds_features.append(self.cds2)
self.genome1.source_features.append(self.src1)
self.genome1.source_features.append(self.src2)
self.genome2 = genome.Genome()
self.genome2.type = "mysql"
self.genome_pair1 = genomepair.GenomePair()
self.genome_pair2 = genomepair.GenomePair()
self.bndl = bundle.Bundle()
self.bndl.ticket = self.ticket1
self.bndl.genome_dict[self.genome1.type] = self.genome1
self.bndl.genome_dict[self.genome2.type] = self.genome2
self.bndl.genome_pair_dict["genome_pair1"] = self.genome_pair1
self.bndl.genome_pair_dict["genome_pair2"] = self.genome_pair2
self.eval_correct1 = eval.Eval(status="correct")
self.eval_correct2 = eval.Eval(status="correct")
self.eval_error1 = eval.Eval(status="error")
self.eval_error2 = eval.Eval(status="error")
def match_genomes(mysqldb_dict, phagesdb_dict):
"""Match MySQL database genome data to PhagesDB genome data.
Both dictionaries:
Key = PhageID
Value = pdm_utils genome object"""
# Generate phage_id sets and match sets.
phagesdb_ids = phagesdb_dict.keys()
mysqldb_ids = mysqldb_dict.keys()
matched_ids = mysqldb_ids & phagesdb_ids
unmatched_mysqldb_ids = mysqldb_ids - phagesdb_ids
unmatched_phagesdb_ids = phagesdb_ids - mysqldb_ids
matched_genomes = []
for id in matched_ids:
gnm_pair = genomepair.GenomePair()
gnm_pair.genome1 = mysqldb_dict[id]
gnm_pair.genome2 = phagesdb_dict[id]
matched_genomes.append(gnm_pair)
unmatched_mysqldb_authored_genomes = {}
for id in unmatched_mysqldb_ids:
gnm = mysqldb_dict[id]
if gnm.annotation_author == 1:
unmatched_mysqldb_authored_genomes[id] = gnm
print("\nSummary of genome matching:")
print(f"{len(matched_ids):>6}: genome(s) matched.")
print(f"{len(unmatched_mysqldb_ids):>6}: MySQL genome(s) not matched.")
print(f"{len(unmatched_phagesdb_ids):>6}: PhagesDB genome(s) not matched.")
count = len(unmatched_mysqldb_authored_genomes.keys())
if count > 0:
print(f"{count} Hatfull-authored unmatched MySQL genome(s):")
for key in unmatched_mysqldb_authored_genomes.keys():
print(key)
return (matched_genomes, unmatched_phagesdb_ids)
def match_genomes(dict1, dict2):
"""Match MySQL database genome data to PhagesDB genome data.
Both dictionaries:
Key = PhageID
Value = pdm_utils genome object"""
# Generate phage_id sets and match sets.
d2_keys = dict2.keys()
d1_keys = dict1.keys()
matched_keys = d1_keys & d2_keys
d1_unmatched_keys = d1_keys - d2_keys
d2_unmatched_keys = d2_keys - d1_keys
matched_genomes = []
for key in matched_keys:
gnm_pair = genomepair.GenomePair()
gnm_pair.genome1 = dict1[key]
gnm_pair.genome2 = dict2[key]
matched_genomes.append(gnm_pair)
# Only unmatched with AnnotationAuthor = 1
unmatched_d1_genomes = {}
for key in d1_unmatched_keys:
gnm = dict1[key]
if gnm.annotation_author == 1:
unmatched_d1_genomes[key] = gnm
results = {
"match_count": len(matched_keys),
"d1_unmatch_count": len(d1_unmatched_keys),
"d2_unmatch_count": len(d2_unmatched_keys),
"d1_unmatch_aa1": unmatched_d1_genomes.keys()
}
print_match_results(results)
return (matched_genomes, d2_unmatched_keys)