def test_GeneScores_add_from_file_bug_list(self):
"""
GeneScores class: Test add_from_file bug list
"""
gene_scores=store.GeneScores()
gene_scores.add_from_file(cfg.genetable_file)
# Test the bug list is as expected
self.assertEqual(sorted(cfg.genetable_file_bug_scores.keys()),sorted(gene_scores.bug_list()))
def test_GeneScores_add_from_file_scores(self):
"""
GeneScores class: Test add_from_file scores
"""
gene_scores=store.GeneScores()
gene_scores.add_from_file(cfg.genetable_file)
# Test the scores for all bugs and genes
for bug in cfg.genetable_file_bug_scores:
self.assertDictEqual(cfg.genetable_file_bug_scores[bug],gene_scores.scores_for_bug(bug))
def test_GeneScores_add_from_file_id_mapping_bug_list(self):
"""
GeneScores class: Test add_from_file bug list with id mapping
"""
gene_scores = store.GeneScores()
gene_scores.add_from_file(
cfg.genetable_file, id_mapping_file=cfg.id_mapping_gene_table_file)
# Test the bug list is as expected
self.assertEqual(
sorted(cfg.genetable_file_bug_scores_id_mapping.keys()),
sorted(gene_scores.bug_list()))
def test_GeneScores_get_score(self):
"""
GeneScores class: Test get_score function
"""
gene_scores=store.GeneScores()
bug1_scores={"gene1":1,"gene2":2}
gene_scores.add(bug1_scores,"bug1")
bug2_scores={"gene1":1,"gene2":2}
gene_scores.add(bug2_scores,"bug2")
self.assertEqual(gene_scores.get_score("bug1","gene2"),2)
def test_GeneScores_add(self):
"""
GeneScores class: Test add function
"""
gene_scores=store.GeneScores()
bug1_scores={"gene1":1,"gene2":2}
gene_scores.add(bug1_scores,"bug1")
bug2_scores={"gene1":1,"gene2":2}
gene_scores.add(bug2_scores,"bug2")
self.assertEqual(gene_scores.count_genes_for_bug("bug1"),2)
def test_gene_families_tsv_output(self):
"""
Test the gene families function and the blast config indexes
Test UniRef50_unknown is read in and used for gene scores but not printed
Test the tsv output
"""
# create a set of alignments
alignments = store.Alignments()
# load the usearch output
file_handle = open(cfg.usearch_file)
for line in file_handle:
if not re.search("^#", line):
data = line.strip().split(config.blast_delimiter)
referenceids = data[config.blast_reference_index].split("|")
queryid = data[config.blast_query_index]
evalue = float(data[config.blast_evalue_index])
alignments.add(referenceids[1], 1, queryid, evalue,
referenceids[0])
file_handle.close()
# set the output format
config.output_format = "tsv"
# set the location of the file to write to as a temp file
file_out, gene_families_file = tempfile.mkstemp()
os.close(file_out)
config.genefamilies_file = gene_families_file
# create gene_scores instance
gene_scores = store.GeneScores()
# obtain the gene families
gene_families_file = quantify_families.gene_families(
alignments, gene_scores)
# check the gene families output is as expected
self.assertTrue(
filecmp.cmp(gene_families_file,
cfg.gene_familes_file,
shallow=False))
# delete the temp file
utils.remove_temp_file(gene_families_file)
def test_GeneScores_add_second_set(self):
"""
GeneScores class: Test add function
Test adding a second set of scores to bug set
"""
gene_scores=store.GeneScores()
bug1_scores={"gene1":1,"gene2":2}
gene_scores.add(bug1_scores,"bug1")
bug1_scores_2={"gene3":1,"gene4":2, "gene2":22}
gene_scores.add(bug1_scores_2,"bug1")
self.assertEqual(gene_scores.count_genes_for_bug("bug1"),4)
def test_GeneScores_scores_for_bug(self):
"""
GeneScores class: Test scores_for_bug
"""
gene_scores=store.GeneScores()
bug1_scores={"gene1":1,"gene2":2}
gene_scores.add(bug1_scores,"bug1")
bug1_scores_2={"gene3":1,"gene4":2, "gene2":22}
gene_scores.add(bug1_scores_2,"bug1")
# Test that the most recent score for gene2 is returned
self.assertDictEqual(gene_scores.scores_for_bug("bug1"),
{"gene1":1,"gene2":22,"gene3":1,"gene4":2})
def test_GeneScores_get_score_second_set(self):
"""
GeneScores class: Test get_score function
Test getting the score for a second set of scores added to bug set
"""
gene_scores=store.GeneScores()
bug1_scores={"gene1":1,"gene2":2}
gene_scores.add(bug1_scores,"bug1")
bug1_scores_2={"gene3":1,"gene4":2, "gene2":22}
gene_scores.add(bug1_scores_2,"bug1")
# Test that the most recent score for gene2 is returned
self.assertEqual(gene_scores.get_score("bug1","gene2"),22)
def test_Alignments_compute_gene_scores_double_gene_double_query(self):
"""
Test the compute_gene_scores function
Test two hits to gene with more than one hit per query
"""
# create a set of hits
# bug, reference, reference_length, query, evalue = hit
eval1 = 1e-4
eval2 = 3e-7
eval3 = 2e-10
eval4 = 2e-10
gene1_length = 2
gene2_length = 3
gene3_length = 4
# Create a set of alignments
alignments_store = store.Alignments()
alignments_store.add("gene1", gene1_length, "query1", eval1, "bug1")
alignments_store.add("gene2", gene2_length, "query1", eval2, "bug1")
alignments_store.add("gene2", gene2_length, "query2", eval3, "bug1")
alignments_store.add("gene3", gene3_length, "query3", eval4, "bug1")
gene_scores_store = store.GeneScores()
# compute gene scores
alignments_store.convert_alignments_to_gene_scores(gene_scores_store)
# gene1
hit1_score = math.exp(-eval1)
hit2_score = math.exp(-eval2)
query1_sum = hit1_score + hit2_score
# convert lengths to per kb
gene2_length = gene2_length / 1000.0
# gene2
hit3_score = math.exp(-eval3)
query2_sum = hit3_score
gene_score = hit3_score / query2_sum / gene2_length + hit2_score / query1_sum / gene2_length
self.assertAlmostEqual(gene_scores_store.get_score("bug1", "gene2"),
gene_score,
places=12)
def test_GeneScores_add_from_file_gene_list(self):
"""
GeneScores class: Test add_from_file gene list
"""
gene_scores=store.GeneScores()
gene_scores.add_from_file(cfg.genetable_file)
# Create a list of all of the genes in the table
genes={}
for bug in cfg.genetable_file_bug_scores:
for gene in cfg.genetable_file_bug_scores[bug]:
genes[gene]=1
# Test the gene list is as expected
self.assertEqual(sorted(genes.keys()),sorted(gene_scores.gene_list()))
def test_Alignments_compute_gene_scores_single_gene_single_query(self):
"""
Test the compute_gene_scores function
Test one hit for gene with one hit for query
"""
# create a set of hits
eval1 = 1e-4
eval2 = 3e-7
eval3 = 2e-10
eval4 = 2e-10
gene1_length = 2
gene2_length = 3
gene3_length = 4
# Create a set of alignments
alignments_store = store.Alignments()
alignments_store.add("gene1", gene1_length, "query1", eval1, "bug1")
alignments_store.add("gene2", gene2_length, "query1", eval2, "bug1")
alignments_store.add("gene2", gene2_length, "query2", eval3, "bug1")
alignments_store.add("gene3", gene3_length, "query3", eval4, "bug1")
gene_scores_store = store.GeneScores()
# compute gene scores
alignments_store.convert_alignments_to_gene_scores(gene_scores_store)
# convert lengths to per kb
gene3_length = gene3_length / 1000.0
# gene3
hit4_score = math.exp(-eval4)
query3_sum = hit4_score
gene_score = hit4_score / query3_sum / gene3_length
self.assertEqual(gene_scores_store.get_score("bug1", "gene3"),
gene_score)