def test_basic_depends_on(self):
q = JobQueue(1)
g = CommandGraph(q)
q.set_command_graph(g)
c1 = CommandNode("sleep 1", "1")
c2 = CommandNode("echo hello", "2", stdout="test.txt")
g.add_node(command_node=c1, depends_on=[c2])
assert set(g.nodes()) == set(['1', '2'])
g.start()
g.finish_block()
os.remove('test.txt')
def test_two_depends_on(self):
q = JobQueue(1)
g = CommandGraph(q)
q.set_command_graph(g)
c1 = CommandNode("sleep 1", "1")
c2 = CommandNode("echo hello", "2", stdout="test.txt")
c3 = CommandNode("echo hello", "3", stdout="test2.txt")
g.add_node(command_node=c1, depends_on=[c3, c2])
g.add_node(command_node=c3, depends_on=[c2])
g.start()
g.finish_block()
os.remove("test.txt")
os.remove("test2.txt")
def bwa_samse(args, config, sai, fq):
"""
Single-ended read processing for bwa
"""
reference = config['reference']['fasta']
cmd = __BWA_SAMSE__.format(config['bwa']['executable'], reference, sai, fq)
id_string = os.path.basename(fq) + ".samse"
stdout_output = os.path.basename(fq) + ".sam"
return CommandNode(cmd, id_string, stdout_output, args.temp_directory)
def bwa_sampe(args, config, sai1, sai2, fq1, fq2):
"""
Paired-end sample processing.
"""
reference = config['reference']['fasta']
cmd = __BWA_SAMPE__.format(config['bwa']['executable'], reference, sai1,
sai2, fq1, fq2)
id_string = os.path.basename(fq1) + ".samse"
stdout_output = os.path.basename(fq1) + ".sam"
return CommandNode(cmd, id_string, stdout_output, args.temp_directory)
def ancient_filter(args, config, bam_file):
"""
Run the ancient filtration procedure
"""
bam_prefix = bam_file.split('.bam')[0]
bam_out_name = bam_prefix + ".ancient_filter.bam"
id_string = bam_prefix + "ancient_filter"
cmd = __ANCIENT_FILTER__.format(config['scripts_dir']['directory'],
args.ancient_downweight_number, bam_file,
bam_out_name)
return CommandNode(cmd, id_string, bam_out_name, args.temp_directory)
def map_damage(args, config, bam_file):
"""
Create the mapdamage command for use in the pipeline
"""
bam_prefix = bam_file.split('.bam')[0]
id_string = bam_prefix + "_filter_unique_bam"
cmd = __MAP_DAMAGE__.format(
os.path.join(config['map_damage']['executable']), bam_file,
os.path.abspath(os.path.join(args.temp_directory, bam_prefix)),
config['reference']['fasta'])
bam_out_name = os.path.join(bam_prefix, bam_prefix + '.rescaled.bam')
return CommandNode(cmd, id_string, bam_out_name, args.temp_directory)
def bwa_aln(args, config, fq):
"""
Aln algorithm
"""
seeding_off = args.bwa_seeding
reference = config['reference']['fasta']
if seeding_off:
cmd = __BWA_ALN__.format(config['bwa']['executable'],
args.bwa_edit_distance, args.bwa_gap_opens,
'1024', reference, fq)
else:
cmd = __BWA_ALN_NO_SEED__.format(config['bwa']['executable'],
args.bwa_edit_distance,
args.bwa_gap_opens, reference, fq)
id_string = os.path.basename(fq) + 'aln'
stdout_output = os.path.basename(fq) + '.sai'
return CommandNode(cmd, id_string, stdout_output, args.temp_directory)
def picard_rg(args, config, bam_file):
sample_id = bam_file.split('_')[0]
bam_out_name = bam_file.split('.bam')[0] + ".rg.bam"
id_string = bam_out_name + "_rg_bam"
cmd = __READ__GROUPS__.format(config['java']['executable'], config['picard']['jar'], bam_file, bam_out_name, sample_id)
return CommandNode(cmd, id_string, bam_out_name, args.temp_directory)
def adapter_removal(config, args, fq1 ,fq2):
fq1o = os.path.abspath(fq1)
fq2o = os.path.abspath(fq2)
cmd = __ADAPTER_REMOVAL__.format(config['adapter_removal']['executable'], fq1o, fq2o, fq1, fq2, args.adapt_min_length, args.adapt_mismatch_rate ,args.adapt_min_qual, args.adapt_alignment_length)
job_id = fq1 + ".adapter_removal"
return CommandNode(cmd, job_id, None, args.temp_directory)
def test_create_command_node(self):
c = CommandNode("echo hello", "1")
assert (str(c) == "1")
assert (c.command == "echo hello")