def run_checkM_lineage_wf(self, ctx, params):
"""
:param params: instance of type "CheckMLineageWfParams" (input_ref -
reference to the input Assembly or BinnedContigs data (could be
expanded to include Genome objects as well)) -> structure:
parameter "input_ref" of String, parameter "workspace_name" of
String, parameter "save_output_dir" of type "boolean" (A boolean -
0 for false, 1 for true. @range (0, 1)), parameter
"save_plots_dir" of type "boolean" (A boolean - 0 for false, 1 for
true. @range (0, 1))
:returns: instance of type "CheckMLineageWfResult" -> structure:
parameter "report_name" of String, parameter "report_ref" of String
"""
# ctx is the context object
# return variables are: result
#BEGIN run_checkM_lineage_wf
print('--->\nRunning kb_Msuite.run_checkM_lineage_wf\nparams:')
print(json.dumps(params, indent=1))
cmu = CheckMUtil(self.config)
result = cmu.run_checkM_lineage_wf(params)
#END run_checkM_lineage_wf
# At some point might do deeper type checking...
if not isinstance(result, dict):
raise ValueError('Method run_checkM_lineage_wf return value ' +
'result is not type dict as required.')
# return the results
return [result]
def test_output_plotting(self):
cmu = CheckMUtil(self.cfg, self.ctx)
plots_dir = os.path.join(self.scratch, 'plots_1')
html_dir = os.path.join(self.scratch, 'html_1')
tetra_file = os.path.join(self.scratch, 'tetra_1.tsv')
cmu.build_checkM_lineage_wf_plots(self.input_dir, self.output_dir,
plots_dir, self.all_seq_fasta,
tetra_file)
self.assertTrue(os.path.isdir(plots_dir))
self.assertTrue(
os.path.isfile(os.path.join(plots_dir, 'bin_qa_plot.png')))
self.assertTrue(
os.path.isfile(
os.path.join(plots_dir, 'NewBins.001.ref_dist_plots.png')))
self.assertTrue(
os.path.isfile(
os.path.join(plots_dir, 'NewBins.002.ref_dist_plots.png')))
self.assertTrue(os.path.isfile(tetra_file))
ob = OutputBuilder(self.output_dir, plots_dir, self.scratch,
self.callback_url)
os.makedirs(html_dir)
res = ob.build_html_output_for_lineage_wf(html_dir, 'MyCheckMOutput')
self.assertIn('shock_id', res)
self.assertIn('name', res)
self.assertIn('description', res)
self.assertEqual(res['name'], 'CheckM_Plot.html')
def setUpClass(cls):
token = environ.get('KB_AUTH_TOKEN', None)
config_file = environ.get('KB_DEPLOYMENT_CONFIG', None)
cls.cfg = {}
config = ConfigParser()
config.read(config_file)
for nameval in config.items('kb_Msuite'):
cls.cfg[nameval[0]] = nameval[1]
# Getting username from Auth profile for token
authServiceUrl = cls.cfg['auth-service-url']
auth_client = _KBaseAuth(authServiceUrl)
user_id = auth_client.get_user(token)
# WARNING: don't call any logging methods on the context object,
# it'll result in a NoneType error
cls.ctx = MethodContext(None)
cls.ctx.update({
'token':
token,
'user_id':
user_id,
'provenance': [{
'service': 'kb_Msuite',
'method': 'please_never_use_it_in_production',
'method_params': []
}],
'authenticated':
1
})
cls.wsURL = cls.cfg['workspace-url']
cls.wsClient = workspaceService(cls.wsURL)
cls.serviceImpl = kb_Msuite(cls.cfg)
cls.scratch = cls.cfg['scratch']
cls.callback_url = os.environ['SDK_CALLBACK_URL']
cls.checkm_runner = CheckMUtil(cls.cfg)
def run_checkM(self, ctx, params):
"""
:param params: instance of type "CheckMInputParams" (Runs CheckM as a
command line local function. subcommand - specify the subcommand
to run; supported options are lineage_wf, tetra, bin_qa_plot,
dist_plot bin_folder - folder with fasta files representing each
contig (must end in .fna) out_folder - folder to store output
plots_folder - folder to save plots to seq_file - the full
concatenated FASTA file (must end in .fna) of all contigs in your
bins, used just for running the tetra command tetra_File - specify
the output/input tetra nucleotide frequency file (generated with
the tetra command) dist_value - when running dist_plot, set this
to a value between 0 and 100 threads - number of threads
reduced_tree - if set to 1, run checkM with the reduced_tree flag,
which will keep memory limited to less than 16gb (otherwise needs
40+ GB, which NJS worker nodes do have) quiet - pass the --quiet
parameter to checkM, but doesn't seem to work for all subcommands)
-> structure: parameter "subcommand" of String, parameter
"bin_folder" of String, parameter "out_folder" of String,
parameter "plots_folder" of String, parameter "seq_file" of
String, parameter "tetra_file" of String, parameter "dist_value"
of Long, parameter "threads" of Long, parameter "reduced_tree" of
type "boolean" (A boolean - 0 for false, 1 for true. @range (0,
1)), parameter "quiet" of type "boolean" (A boolean - 0 for false,
1 for true. @range (0, 1))
"""
# ctx is the context object
#BEGIN run_checkM
print('--->\nRunning kb_Msuite.run_checkM\nparams:')
print(json.dumps(params, indent=1))
for key, value in list(params.items()):
if isinstance(value, str):
params[key] = value.strip()
if 'subcommand' not in params:
raise ValueError(
'"subcommand" parameter field must be specified ' +
'(to one of lineage_wf, tetra, bin_qa_plot, dist_plot, etc)')
checkM_runner = CheckMUtil(self.config, ctx)
checkM_runner.run_checkM(params['subcommand'], params)
#END run_checkM
pass
def setUpClass(cls):
token = environ.get('KB_AUTH_TOKEN', None)
config_file = environ.get('KB_DEPLOYMENT_CONFIG', None)
cls.cfg = {}
config = ConfigParser()
config.read(config_file)
for nameval in config.items('kb_Msuite'):
cls.cfg[nameval[0]] = nameval[1]
# Getting username from Auth profile for token
authServiceUrl = cls.cfg['auth-service-url']
auth_client = _KBaseAuth(authServiceUrl)
user_id = auth_client.get_user(token)
# WARNING: don't call any logging methods on the context object,
# it'll result in a NoneType error
cls.ctx = MethodContext(None)
cls.ctx.update({
'token':
token,
'user_id':
user_id,
'provenance': [{
'service': 'kb_Msuite',
'method': 'please_never_use_it_in_production',
'method_params': []
}],
'authenticated':
1
})
cls.wsURL = cls.cfg['workspace-url']
cls.wsClient = workspaceService(cls.wsURL)
cls.serviceImpl = kb_Msuite(cls.cfg)
cls.scratch = cls.cfg['scratch']
cls.callback_url = os.environ['SDK_CALLBACK_URL']
cls.checkm_runner = CheckMUtil(cls.cfg, cls.ctx)
suffix = int(time.time() * 1000)
cls.wsName = "test_kb_Msuite_" + str(suffix)
cls.ws_info = cls.wsClient.create_workspace({'workspace': cls.wsName})
cls.au = AssemblyUtil(os.environ['SDK_CALLBACK_URL'])
cls.setAPI = SetAPI(url=cls.cfg['srv-wiz-url'], token=cls.ctx['token'])
cls.gfu = GenomeFileUtil(os.environ['SDK_CALLBACK_URL'],
service_ver='dev')
cls.mu = MetagenomeUtils(os.environ['SDK_CALLBACK_URL'])
# stage an input and output directory
"""
cls.input_dir = os.path.join(cls.scratch, 'input_1')
cls.output_dir = os.path.join(cls.scratch, 'output_1')
cls.all_seq_fasta = os.path.join(cls.scratch, 'all_seq.fna')
shutil.copytree(os.path.join('data', 'example_out', 'input'), cls.input_dir)
shutil.copytree(os.path.join('data', 'example_out', 'output'), cls.output_dir)
shutil.copy(os.path.join('data', 'example_out', 'all_seq.fna'), cls.all_seq_fasta)
"""
# prepare WS data
cls.prepare_data()
def setUpClass(cls):
token = environ.get('KB_AUTH_TOKEN', None)
config_file = environ.get('KB_DEPLOYMENT_CONFIG', None)
test_time_stamp = int(time.time() * 1000)
cls.cfg = {}
config = ConfigParser()
config.read(config_file)
for nameval in config.items('kb_Msuite'):
cls.cfg[nameval[0]] = nameval[1]
# Getting username from Auth profile for token
authServiceUrl = cls.cfg['auth-service-url']
auth_client = _KBaseAuth(authServiceUrl)
user_id = auth_client.get_user(token)
# WARNING: don't call any logging methods on the context object,
# it'll result in a NoneType error
cls.ctx = MethodContext(None)
cls.ctx.update({
'token':
token,
'user_id':
user_id,
'provenance': [{
'service': 'kb_Msuite',
'method': 'please_never_use_it_in_production',
'method_params': []
}],
'authenticated':
1
})
cls.wsURL = cls.cfg['workspace-url']
cls.wsClient = Workspace(cls.wsURL)
cls.serviceImpl = kb_Msuite(cls.cfg)
cls.callback_url = os.environ['SDK_CALLBACK_URL']
cls.scratch = cls.cfg['scratch']
cls.appdir = cls.cfg['appdir']
cls.test_data_dir = os.path.join(cls.scratch, 'test_data')
cls.suffix = test_time_stamp
cls.checkm_runner = CheckMUtil(cls.cfg, cls.ctx)
cls.wsName = "test_kb_Msuite_" + str(cls.suffix)
cls.ws_info = cls.wsClient.create_workspace({'workspace': cls.wsName})
cls.au = AssemblyUtil(os.environ['SDK_CALLBACK_URL'])
cls.gfu = GenomeFileUtil(os.environ['SDK_CALLBACK_URL'],
service_ver='dev')
cls.mu = MetagenomeUtils(os.environ['SDK_CALLBACK_URL'])
cls.setAPI = SetAPI(url=cls.cfg['srv-wiz-url'], token=cls.ctx['token'])
cls.kr = KBaseReport(os.environ['SDK_CALLBACK_URL'])
cls.data_loaded = False
def run_checkM_lineage_wf_withFilter(self, ctx, params):
"""
:param params: instance of type "CheckMLineageWf_withFilter_Params"
(input_ref - reference to the input BinnedContigs data) ->
structure: parameter "input_ref" of String, parameter
"workspace_name" of String, parameter "reduced_tree" of type
"boolean" (A boolean - 0 for false, 1 for true. @range (0, 1)),
parameter "save_output_dir" of type "boolean" (A boolean - 0 for
false, 1 for true. @range (0, 1)), parameter "save_plots_dir" of
type "boolean" (A boolean - 0 for false, 1 for true. @range (0,
1)), parameter "completeness_perc" of Double, parameter
"contamination_perc" of Double, parameter
"output_filtered_binnedcontigs_obj_name" of String, parameter
"threads" of Long
:returns: instance of type "CheckMLineageWf_withFilter_Result" ->
structure: parameter "report_name" of String, parameter
"report_ref" of String, parameter "binned_contig_obj_ref" of type
"obj_ref" ("WS_ID/OBJ_ID/VER")
"""
# ctx is the context object
# return variables are: result
#BEGIN run_checkM_lineage_wf_withFilter
print(
'--->\nRunning kb_Msuite.run_checkM_lineage_wf_withFilter\nparams:'
)
print(json.dumps(params, indent=1))
cmu = CheckMUtil(self.config, ctx)
result = cmu.run_checkM_lineage_wf(params)
#END run_checkM_lineage_wf_withFilter
# At some point might do deeper type checking...
if not isinstance(result, dict):
raise ValueError(
'Method run_checkM_lineage_wf_withFilter return value ' +
'result is not type dict as required.')
# return the results
return [result]