def test_run_pipeline_no_rqcfilter(self):
# load some data here.
output = self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": "MyNewAssembly",
"workspace_name": util.get_ws_name(),
"skip_rqcfilter": 1,
"debug": 1
})[0]
self.assertIn('report_name', output)
self.assertIn('report_ref', output)
self.assertIn('assembly_upa', output)
pprint(output)
# again, but save cleaned reads
output = self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": "MyNewAssembly",
"workspace_name": util.get_ws_name(),
"skip_rqcfilter": 1,
"debug": 1,
"cleaned_reads_name": "MyCleanedReads"
})[0]
self.assertIn("report_name", output)
self.assertIn("report_ref", output)
self.assertIn("assembly_upa", output)
self.assertIn("cleaned_reads_upa", output)
pprint(output)
def test_make_report_ok(self):
pipeline_output = {
"bbmap": {
"stats_file": self.bbmap_stats_file,
"coverage_file": self.cov_stats,
"command": "bbmap cmd",
"version_string": "bbmap version"
},
"stats": {
"stats_txt": self.assembly_stats_file,
"stats_tsv": self.assembly_stats_tsv,
"command": "stats command",
"version_string": "stats version"
},
"rqcfilter": {
"run_log": self.rqcfilter_log,
"command": "rqcfilter command",
"version_string": "rqcfilter version"
},
"reads_info_prefiltered": {
"count": 10000,
"command": "reads_info_prefiltered command",
"version_string": "reads info version"
},
"reads_info_filtered": {
"count": 8000,
"command": "reads_info_filtered command",
"version_string": "reads info version"
},
"reads_info_corrected": {
"count": 7000,
"command": "reads_info_corrected command",
"version_string": "reads info version"
},
"bfc": {
"command": "bfc command",
"version_string": "bfc version"
},
"seqtk": {
"command": "seqtk command",
"version_string": "seqtk verseion"
},
"spades": {
"command": "spades command",
"version_string": "spades version"
},
"agp": {
"command": "agp command",
"version_string": "agp version"
},
}
report_info = self._get_report_util().make_report(
pipeline_output, util.get_ws_name(), []
)
self.assertIn('report_ref', report_info)
self.assertIn('report_name', report_info)
def test_run_pipeline_mismatched_inputs(self):
with self.assertRaises(ValueError):
self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": "MyNewAssembly",
"workspace_name": util.get_ws_name(),
"skip_rqcfilter": True,
"filtered_reads_name": "MyFilteredReads"
})
def test_run_pipeline_missing_inputs(self):
with self.assertRaises(ValueError):
self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": None,
"output_assembly_name": "MyNewAssembly",
"workspace_name": util.get_ws_name()
})
with self.assertRaises(ValueError):
self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": None,
"workspace_name": util.get_ws_name()
})
with self.assertRaises(ValueError):
self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": "MyNewAssembly",
"workspace_name": None
})
def test_run_pipeline_default_ok(self):
# load some data here.
output = self.getImpl().run_mg_assembly_pipeline(
self.getContext(), {
"reads_upa": self.reads_upa,
"output_assembly_name": "MyNewAssembly",
"workspace_name": util.get_ws_name(),
"debug": 1
})[0]
self.assertIn('report_name', output)
self.assertIn('report_ref', output)
self.assertIn('assembly_upa', output)
pprint(output)
def test_upload_assembly_bad_inputs(self):
"""
Missing or malformed inputs to file.upload_assembly(). Should fail properly.
"""
file_util = self._get_file_util()
with self.assertRaises(ValueError) as cm:
file_util.upload_assembly(None, util.get_ws_name(), "MyAssembly")
self.assertIn("must be defined", str(cm.exception))
with self.assertRaises(ValueError) as cm:
file_util.upload_assembly("not_a_file", util.get_ws_name(),
"MyAssembly")
self.assertIn("does not exist", str(cm.exception))
with self.assertRaises(ValueError) as cm:
file_util.upload_assembly(self.scratch_assembly, None,
"MyAssembly")
self.assertIn("must be defined", str(cm.exception))
with self.assertRaises(ValueError) as cm:
file_util.upload_assembly(self.scratch_assembly,
util.get_ws_name(), None)
self.assertIn("must be defined", str(cm.exception))
with self.assertRaises(Exception):
file_util.upload_assembly(self.scratch_assembly, "not_a_workspace",
"MyAssembly")
def test_upload_assembly_ok(self):
"""
happy test - all fields properly made.
"""
file_util = self._get_file_util()
asm_upa = file_util.upload_assembly(self.scratch_assembly,
util.get_ws_name(), "MyAssembly")
self.assertIsNotNone(asm_upa)
ws = util.get_ws_client()
asm_info = ws.get_object_info3({'objects': [{
'ref': asm_upa
}]})['infos'][0]
self.assertEquals("MyAssembly", asm_info[1])
self.assertIn("KBaseGenomeAnnotations.Assembly", asm_info[2])
def test_make_report_bad_inputs(self):
pipeline_output = {
"bbmap": {
"stats_file": self.bbmap_stats_file,
"coverage_file": self.cov_stats,
"command": "bbmap cmd",
"version_string": "bbmap version"
},
"stats": {
"stats_txt": self.assembly_stats_file,
"stats_tsv": self.assembly_stats_tsv,
"command": "stats command",
"version_string": "stats version"
},
"rqcfilter": {
"run_log": self.rqcfilter_log,
"command": "rqcfilter command",
"version_string": "rqcfilter version"
},
"reads_info_prefiltered": {
"count": 10000,
"command": "reads_info_prefiltered command",
"version_string": "reads info version"
},
"reads_info_filtered": {
"count": 8000,
"command": "reads_info_filtered command",
"version_string": "reads info version"
},
"reads_info_corrected": {
"count": 7000,
"command": "reads_info_corrected command",
"version_string": "reads info version"
},
"bfc": {
"command": "bfc command",
"version_string": "bfc version"
},
"seqtk": {
"command": "seqtk command",
"version_string": "seqtk verseion"
},
"spades": {
"command": "spades command",
"version_string": "spades version"
},
"agp": {
"command": "agp command",
"version_string": "agp version"
},
}
ru = self._get_report_util()
with self.assertRaises(AssertionError) as err:
ru.make_report(None, util.get_ws_name(), [])
self.assertIn("Pipeline output not found!", str(err.exception))
for key in ["reads_info_prefiltered", "reads_info_filtered", "reads_info_corrected"]:
output_copy = deepcopy(pipeline_output)
del output_copy[key]
with self.assertRaises(AssertionError) as err:
ru.make_report(output_copy, util.get_ws_name(), [])
self.assertIn("Required reads info '{}' is not present!".format(key), str(err.exception))
with self.assertRaises(AssertionError) as err:
ru.make_report(pipeline_output, None, [])
self.assertIn("A workspace name is required", str(err.exception))