def test_create_readme(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.version = '1.0'
task = ChallengeDataTask(temp_dir, params)
blast = BlastNFilterTask(temp_dir, params)
blast.create_dir()
sfile = blast.get_blastnfilter_summary_file()
f = open(sfile, 'w')
f.write('hello there\n')
f.flush()
f.close()
task._create_readme(temp_dir)
readme = os.path.join(temp_dir,
ChallengeDataTask.README_TXT_FILE)
self.assertEqual(os.path.isfile(readme), True)
f = open(readme, 'r')
found = False
for line in f:
print line
if re.match('^hello there.*$', line):
found = True
break
f.close()
self.assertEqual(found, True)
finally:
shutil.rmtree(temp_dir)
def test_create_readme_version_empty_summary(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.version = '1.0'
task = ChallengeDataTask(temp_dir, params)
blast = BlastNFilterTask(temp_dir, params)
blast.create_dir()
sfile = blast.get_blastnfilter_summary_file()
open(sfile, 'a').close()
task._create_readme(temp_dir)
readme = os.path.join(temp_dir,
ChallengeDataTask.README_TXT_FILE)
self.assertEqual(os.path.isfile(readme), True)
f = open(readme, 'r')
found = False
for line in f:
if re.match('^Celpprunner version: 1.0.*$', line):
found = True
break
f.close()
self.assertEqual(found, True)
finally:
shutil.rmtree(temp_dir)
def generate_reports(theargs):
"""Generates reports
"""
celpp_years = util.get_all_celpp_years(theargs.celppdir)
if theargs.outdir is None:
raise Exception('--outdir must be set')
if not os.path.isdir(theargs.outdir):
os.makedirs(theargs.outdir)
f = open(os.path.join(theargs.outdir, BLASTNFILTER_SUMMARY_CSV), 'w')
f.write('Week #, Year, Complexes, Dockable complexes, Dockable monomers, '
'Targets Found\n')
for year in celpp_years:
logger.info('Examining year ' + year)
for week in util.get_all_celpp_weeks(
os.path.join(theargs.celppdir, year)):
logger.debug('Examining week ' + week)
the_dir = os.path.join(theargs.celppdir, year,
util.DATA_SET_WEEK_PREFIX + week)
blast = BlastNFilterTask(the_dir, theargs)
summary = blast.get_blastnfilter_summary()
f.write(summary.get_csv() + '\n')
f.flush()
f.close()
def test_get_task_list_for_stage_for_scoring_stage_with_twofound(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
blasttask = BlastNFilterTask(temp_dir, params)
blasttask.create_dir()
open(os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE), 'a').close()
params.latest_weekly = temp_dir
glidedir = os.path.join(temp_dir,
EvaluationTaskFactory.DOCKSTAGE_PREFIX +
'glide')
os.mkdir(glidedir)
open(os.path.join(glidedir, D3RTask.COMPLETE_FILE), 'a').close()
freddir = os.path.join(temp_dir,
EvaluationTaskFactory.DOCKSTAGE_PREFIX +
'fred')
os.mkdir(freddir)
open(os.path.join(freddir, D3RTask.COMPLETE_FILE), 'a').close()
task_list = celpprunner.get_task_list_for_stage(params,
'evaluation')
self.assertEqual(len(task_list), 2)
finally:
shutil.rmtree(temp_dir)
def test_run_fails_cause_genchallenge_fails(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.genchallenge = 'false'
params.pdbdb = '/foo'
params.version = '1'
blastnfilter = BlastNFilterTask(temp_dir, params)
blastnfilter.create_dir()
open(os.path.join(blastnfilter.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
chall = ChallengeDataTask(temp_dir, params)
chall.run()
self.assertEqual(chall.get_error(),
'Non zero exit code: 1 received. Standard out: ' +
' Standard error: ')
# test file gets created
errfile = os.path.join(chall.get_dir(),
D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(errfile), True)
stderr = os.path.join(chall.get_dir(),
'false.stderr')
self.assertEqual(os.path.isfile(stderr), True)
stdout = os.path.join(chall.get_dir(),
'false.stdout')
self.assertEqual(os.path.isfile(stdout), True)
finally:
shutil.rmtree(temp_dir)
def _create_readme(self, path):
"""Creates readme.txt file for task
"""
ver = 'Unknown'
try:
ver = self.get_args().version
except AttributeError:
logger.warning('Version unset using Unknown')
f = open(os.path.join(path, ChallengeDataTask.README_TXT_FILE), 'w')
f.write(ChallengeDataTask.README_BODY.format(
version=ver,
week=self._week_num,
year=self._year))
blast = BlastNFilterTask(self.get_path(), self.get_args())
summary_file = blast.get_blastnfilter_summary_file()
# append summary.txt file
if os.path.isfile(summary_file):
sumfile = open(summary_file, 'r')
for line in sumfile:
f.write(line)
sumfile.close()
f.flush()
f.close()
def test_run_with_can_run_already_set_false(self):
params = D3RParameters()
params.blastnfilter = 'false'
params.postanalysis = 'false'
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(None, params)
blasttask._can_run = False
blasttask.run()
def test_blastnfilter_summary_week_and_year(self):
blast = BlastNFilterTask('/foo', D3RParameters())
summary = BlastNFilterSummary('/foo/2018/dataset.week.4'
'/' + blast.get_dir_name())
self.assertEqual(summary.get_week_number(), '4')
self.assertEqual(summary.get_year(), '2018')
self.assertEqual(summary.get_csv(), '4,2018,0,0,0,0')
def test_run_success_with_ftp_upload(self):
temp_dir = tempfile.mkdtemp()
try:
script = self.create_gen_challenge_script(temp_dir)
params = D3RParameters()
params.genchallenge = script
params.pdbdb = '/foo'
params.version = '1'
blastnfilter = BlastNFilterTask(temp_dir, params)
blastnfilter.create_dir()
open(os.path.join(blastnfilter.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
chall = ChallengeDataTask(temp_dir, params)
mockftp = D3RParameters()
mockftp.put = Mock(side_effect=[3, 5])
ftp = FtpFileTransfer(None)
ftp.set_remote_challenge_dir('/challenge')
ftp.set_connection(mockftp)
chall.set_file_transfer(ftp)
dimport = DataImportTask(temp_dir, params)
dimport.create_dir()
ctsv = dimport.get_crystalph_tsv()
f = open(ctsv, 'w')
f.write('crystal')
f.flush()
f.close()
nonpoly = dimport.get_nonpolymer_tsv()
f = open(nonpoly, 'w')
f.write('nonpoly')
f.flush()
f.close()
seq = dimport.get_sequence_tsv()
f = open(seq, 'w')
f.write('seq')
f.flush()
f.close()
chall.run()
self.assertEqual(chall.get_error(), None)
# verify test files get created
errfile = os.path.join(chall.get_dir(),
D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(errfile), False)
compfile = os.path.join(chall.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(compfile), True)
ftp.disconnect()
finally:
shutil.rmtree(temp_dir)
def test_get_blastnfilter_summary(self):
"""Tests BlastNFilterTask.get_blastnfilter_summary call
"""
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
blasttask = BlastNFilterTask(temp_dir, params)
summary = blasttask.get_blastnfilter_summary()
self.assertEqual(summary.get_complexes(), 0)
finally:
shutil.rmtree(temp_dir)
def __init__(self, path, args):
super(ChallengeDataTask, self).__init__(path, args)
self.set_name('challengedata')
# Make stage number one higher then BlastNFilter Stage
blast = BlastNFilterTask(path, args)
self.set_stage(blast.get_stage() + 1)
self.set_status(D3RTask.UNKNOWN_STATUS)
self._challenge_tarball_filename = None
self._week_num = util.get_celpp_week_number_from_path(self.get_path())
self._year = util.get_celpp_year_from_path(self.get_dir())
def test_get_evaluation_tasks_on_dir_with_lower_stages_dirs(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
dataimport = DataImportTask(temp_dir, params)
blast = BlastNFilterTask(temp_dir, params)
os.mkdir(os.path.join(temp_dir, dataimport.get_dir_name()))
os.mkdir(os.path.join(temp_dir, blast.get_dir_name()))
stf = EvaluationTaskFactory(temp_dir, params)
task_list = stf.get_evaluation_tasks()
self.assertEquals(len(task_list), 0)
finally:
shutil.rmtree(temp_dir)
def test_constructor(self):
params = D3RParameters()
blasttask = BlastNFilterTask('ha', params)
self.assertEqual(blasttask.get_name(), 'blastnfilter')
self.assertEqual(blasttask.get_path(), 'ha')
self.assertEqual(blasttask.get_stage(), 3)
self.assertEqual(blasttask.get_status(), D3RTask.UNKNOWN_STATUS)
self.assertEqual(blasttask.get_error(), None)
def can_run(self):
"""Determines if task can actually run
This method first verifies the docking task
has `D3RTask.COMPLETE_STATUS` for
status. The method then verifies this task does
not already exist. If above is not true then self.set_error()
is set with information about the issue
:return: True if can run otherwise False
"""
self._can_run = False
self._error = None
self._docktask.update_status_from_filesystem()
if self._docktask.get_status() != D3RTask.COMPLETE_STATUS:
logger.info('Cannot run ' + self.get_name() + ' task ' +
'because ' + self._docktask.get_name() + ' task' +
' has a status of ' + self._docktask.get_status())
self.set_error(self._docktask.get_name() + ' task has ' +
self._docktask.get_status() + ' status')
return False
# check blast
blastnfilter = BlastNFilterTask(self._path, self._args)
blastnfilter.update_status_from_filesystem()
if blastnfilter.get_status() != D3RTask.COMPLETE_STATUS:
logger.info('Cannot run ' + self.get_name() + 'task ' +
'because ' + blastnfilter.get_name() + 'task' +
'has a status of ' + blastnfilter.get_status())
self.set_error(blastnfilter.get_name() + ' task has ' +
blastnfilter.get_status() + ' status')
return False
# check this task is not complete and does not exist
self.update_status_from_filesystem()
if self.get_status() == D3RTask.COMPLETE_STATUS:
logger.debug("No work needed for " + self.get_name() + " task")
return False
if self.get_status() != D3RTask.NOTFOUND_STATUS:
logger.warning(self.get_name() + " task was already " +
"attempted, but there was a problem")
self.set_error(self.get_dir_name() + ' already exists and ' +
'status is ' + self.get_status())
return False
self._can_run = True
return True
def test_run_fails_cause_genchallenge_not_set(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.version = '1'
blastnfilter = BlastNFilterTask(temp_dir, params)
blastnfilter.create_dir()
open(os.path.join(blastnfilter.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
chall = ChallengeDataTask(temp_dir, params)
chall.run()
self.assertEqual(chall.get_error(),
'genchallenge not set')
# test files get created
self.assertEqual(os.path.isdir(chall.get_dir()),
True)
errfile = os.path.join(chall.get_dir(),
D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(errfile), True)
finally:
shutil.rmtree(temp_dir)
def test_run_fails_cause_genchallenge_is_not_found(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.genchallenge = '/bin/doesnotexist'
params.pdbdb = '/foo'
params.version = '1'
blastnfilter = BlastNFilterTask(temp_dir, params)
blastnfilter.create_dir()
open(os.path.join(blastnfilter.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
chall = ChallengeDataTask(temp_dir, params)
chall.run()
self.assertEqual(chall.get_error(),
'Caught Exception trying to run ' +
'/bin/doesnotexist --candidatedir ' +
blastnfilter.get_dir() + ' --pdbdb ' +
'/foo --outdir ' +
chall.get_dir() +
'/' + chall.get_celpp_challenge_data_dir_name() +
' : [Errno 2] No such file or directory')
# test files get created
errfile = os.path.join(chall.get_dir(),
D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(errfile), True)
finally:
shutil.rmtree(temp_dir)
def test_can_run(self):
temp_dir = tempfile.mkdtemp()
try:
# no blast task found so it cannot run
params = D3RParameters()
task = ChallengeDataTask(temp_dir, params)
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(),
'blastnfilter task has notfound status')
# blastn filter running
blastnfilter = BlastNFilterTask(temp_dir, params)
blastnfilter.create_dir()
open(os.path.join(blastnfilter.get_dir(), D3RTask.START_FILE),
'a').close()
task = ChallengeDataTask(temp_dir, params)
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(),
'blastnfilter task has start status')
# blastnfilter failed
error_file = os.path.join(blastnfilter.get_dir(),
D3RTask.ERROR_FILE)
open(error_file, 'a').close()
task = ChallengeDataTask(temp_dir, params)
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(),
'blastnfilter task has error status')
# blastnfilter success
os.remove(error_file)
open(os.path.join(blastnfilter.get_dir(),
D3RTask.COMPLETE_FILE), 'a').close()
task = ChallengeDataTask(temp_dir, params)
self.assertEqual(task.can_run(), True)
self.assertEqual(task.get_error(), None)
# proteinligprep task exists already
task = ChallengeDataTask(temp_dir, params)
task.create_dir()
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(),
task.get_dir_name() +
' already exists and status is unknown')
# proteinlibprep already complete
task = ChallengeDataTask(temp_dir, params)
open(os.path.join(task.get_dir(),
D3RTask.COMPLETE_FILE), 'a').close()
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(), None)
finally:
shutil.rmtree(temp_dir)
def test_run_with_exception(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.blastnfilter = 'falseasdfasdf'
params.postanalysis = 'true'
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
blasttask.run()
self.assertEqual(blasttask.get_status(), D3RTask.ERROR_STATUS)
self.assertEqual(blasttask.get_error().startswith('Caught'), True)
self.assertNotEqual(blasttask.get_error(), None)
finally:
shutil.rmtree(temp_dir)
def test_update_status_from_filesystem(self):
params = D3RParameters()
task = BlastNFilterTask(None, params)
test_task.try_update_status_from_filesystem(self, task)
def main():
blasttask = BlastNFilterTask('', p)
dataimport = DataImportTask('', p)
challenge = ChallengeDataTask('', p)
glide = GlideTask('', p)
makedb = MakeBlastDBTask('', p)
prot = ProteinLigPrepTask('', p)
vina = AutoDockVinaTask('', p)
chimeraprep = ChimeraProteinLigPrepTask('', p)
desc = """
Version {version}
Runs the 9 stages (makedb, import, blast, challengedata,
proteinligprep, {chimeraprep}, extsubmission, glide, vina, &
evaluation) of CELPP processing pipeline
(http://www.drugdesigndata.org)
CELPP processing pipeline relies on a set of directories
with specific structure. The pipeline runs a set of stages
Each stage has a numerical value and a name. The numerical
value denotes order and the stage name identifies separate
tasks to run in the stage.
The filesystem structure of the stage is:
stage.<stage number>.<task name>
The stage(s) run are defined via the required --stage flag.
To run multiple stages serially just pass a comma delimited
list to the --stage flag. Example: --stage import,blast
NOTE: When running multiple stages serially the program will
not run subsequent stages if a task in a stage fails.
Also note order matters, ie putting blast,import will
cause celpprunner.py to run blast stage first.
This program drops a pid lockfile
(celpprunner.<stage>.lockpid) in celppdir to prevent duplicate
invocation.
When run, this program will examine the stage and see
if work can be done. If stage is complete or previous
steps have not completed, the program will exit silently.
If previous steps have failed or current stage already
exists in an error or uncomplete state then program will
report the error via email using addresses set in --email
flag. Errors will also be reported via stderr/stdout.
The program will also exit with nonzero exit code.
This program utilizes simple token files to denote stage
completion. If within the stage directory there is a:
'{complete}' file - then stage is done and no other
checking is done.
'error' file - then stage failed.
'start' file - then stage is running.
Notification of stage start and end will be sent to
addresses set via --email flag.
Unless --customweekdir is set, this program will
examine the 'celppdir' (last argument passed on
commandline) to find the latest directory with this path:
<year>/dataset.week.#
The program will find the latest <year> and within
that year the dataset.week.# with highest #. The output
directories created will be put within this directory.
Setting --customweekdir will cause program to use 'celppdir'
path.
Setting the --createweekdir flag will instruct this
program to create a new directory for the current
celpp week/year before running any stage processing.
NOTE: CELPP weeks start on Friday and end on Thursday
and week # follows ISO8601 rules so week numbers
at the end and start of the year are a bit
wonky.
Breakdown of behavior of program is defined by
value passed with --stage flag:
If --stage '{createchallenge}'
This is NOT a stage, but has the same effect as
calling --stage makedb,import,blast,challengedata
The four stages that need to run to generate the challenge
data package.
If --stage 'makedb'
In this stage the file {pdb_seqres} is downloaded from
an ftp site set by --pdbsequrl.
This file is then gunzipped and NCBI makeblastdb
(set by --makeblastdb) is run on it to create a blast
database. The files are stored in {makeblastdb_dirname}
If --stage 'import'
In this stage 4 files are downloaded from urls specified
by --compinchi and --pdbfileurl flags on the commandline
into {dataimport_dirname} directory.
The tsv files are (--pdbfileurl flag sets url to
download these files from):
{nonpolymer_tsv}
{sequence_tsv}
{crystal_tsv}
The Components ich file is (--compinchi flag sets base url to
download this file from):
{compinchi_ich}
This stage will just wait and retry if any of the tsv files
have NOT been updated since the start of the current
celpp week as determined by a HEAD request. To bypass
this delay add --skipimportwait flag. --importsleep denotes
the time to wait before re-examining the update time of the
tsv files and --importretry sets number of times to retry
before giving up.
If --stage 'blast'
Verifies {dataimport_dirname} exists and has '{complete}'
file. Also verifies {makeblastdb_dirname} exists and has
'{complete}' file. If both conditions are met then the
'blast' stage is run which invokes script set by
--blastnfilter flag and output stored in
{blast_dirname}.
Requires --pdbdb to be set to a directory with valid PDB
database files.
Note: --blastnfilter script is killed after time set with
--blastnfiltertimeout flag.
If --stage 'challengedata'
Verifies {blast_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --genchallenge flag to create a challenge dataset
file. The --pdbdb flag must also be set when calling this
stage. If --ftpconfig is set with {challengepath} field then
this stage will also upload the challenge dataset tarfile
to the ftp server with path set by {challengepath}. The
code will also upload a {latest_txt} file containing name
of the tarfile to the same destination overwriting any
{latest_txt} file that already exists.
Example file for --ftpconfig:
{host} some.ftp.com
{user} bob
{passn} mypass
{path} /celpp
{challengepath} /challenge
{submissionpath} /submissions
If --stage '{chimeraprep}'
Verifies {challenge_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --chimeraprep flag to prepare pdb and inchi files
storing output in {chimeraprep_dirname}. --pdbdb flag
must also be set when calling this stage.
If --stage 'proteinligprep'
Verifies {challenge_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --proteinligprep flag to prepare pdb and inchi files
storing output in {proteinligprep_dirname}. --pdbdb flag
must also be set when calling this stage.
If --stage 'extsubmission'
Connects to server specified by --ftpconfig and downloads
external docking submissions from {submissionpath} on remote
server.
Submissions should be named:
celpp_weekXX_YYYY_dockedresults_ZZZZ.tar.gz as documented here:
https://github.com/drugdata/d3r/wiki/Proposed-challenge-docked\
-results-file-structure
For each submission a directory named stage.X.ZZZZ.extsubmission
will be created and uncompressed contents of package will be
stored in that directory. If data does not conform properly
'error' file will be placed in directory denoting failure
If --stage 'glide'
Verifies {proteinligprep_dirname} exists and has a '{complete}'
file within it. If complete, this stage runs which invokes
program set in --glide flag to perform docking via glide
storing output in {glide_dirname}
If --stage 'vina'
Verifies {proteinligprep_dirname} exists and has a '{complete}'
file within it. If complete, this stage runs which invokes
program set in --vina flag to perform docking via AutoDock Vina
storing output in {vina_dirname}
If --stage 'evaluation'
Finds all stage.{dockstage}.<algo> directories with '{complete}'
files in them which do not end in name '{webdata}' and runs
script set via --evaluation parameter storing the result of
the script into stage.{evalstage}.<algo>.evaluation. --pdbdb flag
must also be set when calling this stage.
""".format(makeblastdb_dirname=makedb.get_dir_name(),
dataimport_dirname=dataimport.get_dir_name(),
blast_dirname=blasttask.get_dir_name(),
challenge_dirname=challenge.get_dir_name(),
createchallenge=CREATE_CHALLENGE,
proteinligprep_dirname=prot.get_dir_name(),
glide_dirname=glide.get_dir_name(),
vina_dirname=vina.get_dir_name(),
dockstage=str(glide.get_stage()),
evalstage=str(glide.get_stage() + 1),
complete=blasttask.COMPLETE_FILE,
chimeraprep_dirname=chimeraprep.get_dir_name(),
chimeraprep=CHIMERA_PREP,
compinchi_ich=DataImportTask.COMPINCHI_ICH,
pdb_seqres=MakeBlastDBTask.PDB_SEQRES_TXT_GZ,
nonpolymer_tsv=DataImportTask.NONPOLYMER_TSV,
sequence_tsv=DataImportTask.SEQUENCE_TSV,
crystal_tsv=DataImportTask.CRYSTALPH_TSV,
webdata=EvaluationTaskFactory.WEB_DATA_SUFFIX,
latest_txt=ChallengeDataTask.LATEST_TXT,
host=FtpFileTransfer.HOST,
user=FtpFileTransfer.USER,
passn=FtpFileTransfer.PASS,
path=FtpFileTransfer.PATH,
challengepath=FtpFileTransfer.CHALLENGEPATH,
submissionpath=FtpFileTransfer.SUBMISSIONPATH,
version=d3r.__version__)
theargs = _parse_arguments(desc, sys.argv[1:])
theargs.program = sys.argv[0]
theargs.version = d3r.__version__
util.setup_logging(theargs)
try:
run_stages(theargs)
except Exception:
logger.exception("Error caught exception")
sys.exit(2)
def test_parse_blastnfilter_output_for_hit_stats(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
blasttask = BlastNFilterTask(temp_dir, params)
blasttask.create_dir()
# no summary.txt file
self.assertEquals(
blasttask._parse_blastnfilter_output_for_hit_stats(),
'\n# txt files found: 0\n\nOutput from ' + 'summary.txt\n')
csv_file = os.path.join(blasttask.get_dir(), '4zyc.txt')
f = open(csv_file, 'w')
f.write('NEED TO PUT REAL DATA IN HERE\n')
f.flush()
f.close()
self.assertEquals(
blasttask._parse_blastnfilter_output_for_hit_stats(),
'\n# txt files found: 1\n\nOutput from ' + 'summary.txt\n')
csv_file = os.path.join(blasttask.get_dir(), '4qqq.txt')
f = open(csv_file, 'w')
f.write('NEED TO PUT REAL DATA IN HERE\n')
f.flush()
f.close()
res = blasttask._parse_blastnfilter_output_for_hit_stats()\
.rstrip('\n')
res.index('# txt files found: 2')
res.index('Output from summary.txt')
csv_file = os.path.join(blasttask.get_dir(), '4abc.txt')
f = open(csv_file, 'w')
f.write('NEED TO PUT REAL DATA IN HERE\n')
f.flush()
f.close()
res = blasttask._parse_blastnfilter_output_for_hit_stats()\
.rstrip('\n')
res.index('# txt files found: 3')
res.index('Output from summary.txt')
finally:
shutil.rmtree(temp_dir)
def test_generate_reports_with_one_week_in_one_year(self):
temp_dir = tempfile.mkdtemp()
try:
theargs = D3RParameters()
theargs.celppdir = temp_dir
outdir = os.path.join(temp_dir, 'outdir')
theargs.outdir = outdir
yeardir = os.path.join(temp_dir, '2016')
os.mkdir(yeardir)
weekdir = os.path.join(yeardir, 'dataset.week.10')
os.mkdir(weekdir)
blast = BlastNFilterTask(temp_dir, theargs)
blastdir = os.path.join(weekdir, blast.get_dir_name())
os.mkdir(blastdir)
# create dummary summary.txt file
f = open(os.path.join(blastdir, BlastNFilterTask.SUMMARY_TXT), 'w')
f.write('INPUT SUMMARY\n')
f.write(' entries: 221\n')
f.write(' complexes: 178\n')
f.write(' dockable complexes: 95\n')
f.write(' monomers: 145\n')
f.write(' dockable monomers: 71\n')
f.write(' multimers: 76\n')
f.write(' dockable multimers: 24\n\n')
f.write('FILTERING CRITERIA\n')
f.write(' No. of query sequences <= 1\n')
f.write(' No. of dockable ligands = 1\n')
f.write(' Percent identity >= 0.95\n')
f.write(' Percent Coverage >= 0.9\n')
f.write(' No. of hit sequences <= 4\n')
f.write(' Structure determination method: '
' x-ray diffraction\n\n')
f.write('OUTPUT SUMMARY\n')
f.write(' Targets found: 67\n')
f.write(' Target: 5fz7|Sequences: 1|Hits: 94|Candidates: 17|'
'Elected:4|PDBids: 5fz7,5fyz,5a3p,5a1f\n')
f.flush()
f.close()
celppreports.generate_reports(theargs)
self.assertEqual(os.path.isdir(outdir), True)
csv_file = os.path.join(outdir, 'blastnfilter.summary.csv')
self.assertEqual(os.path.isfile(csv_file), True)
# check csv file
f = open(csv_file, 'r')
header = f.readline()
self.assertEqual(
header, 'Week #, Year, Complexes, Dockable '
'complexes, Dockable monomers, '
'Targets Found\n')
data = f.readline()
self.assertEqual(data, '10,2016,178,95,71,67\n')
finally:
shutil.rmtree(temp_dir)
def test_run_with_error(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.blastnfilter = 'false'
params.postanalysis = 'true'
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
blasttask.run()
self.assertEqual(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEqual(blasttask.get_error(), None)
complete_file = os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(complete_file), True)
error_file = os.path.join(blasttask.get_dir(), D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(error_file), False)
std_err_file = os.path.join(blasttask.get_dir(), 'false.stderr')
self.assertEqual(os.path.isfile(std_err_file), True)
std_out_file = os.path.join(blasttask.get_dir(), 'false.stdout')
self.assertEqual(os.path.isfile(std_out_file), True)
finally:
shutil.rmtree(temp_dir)
def test_run_fail_due_to_time_but_with_txt_file(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
foo_script = os.path.join(temp_dir, 'foo.py')
params.blastnfilter = foo_script
params.postanalysis = '/bin/echo'
params.pdbdb = '/pdbdb'
params.blastnfiltertimeout = 1
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
txt_file = os.path.join(blasttask.get_dir(), '4za4.txt')
txt_contents = (
'query, 4za4\\n' + 'ph, 7.4\\n' + 'ligand, 4LU\\n\\n' +
'inchi, InChI=1S/C22H29N4O9P/c1-10-7-12-16-' +
'15(11(10)2)22(3,4)5-6-25(16)17-19(23-21(31)' +
'24-20(17)30)26(12)8-13(27)18(29)14(28)9-35-' +
'36(32,33)34/h6-7,13-14,18,27-29H,5,8-9H2,1-' +
'4H3,(H3-,23,24,30,31,32,33,34)/p+1/t13-,14+' + ',18-/m0/s1\\n'
'largest, 4zz3, 4PP\\n' + 'smallest, 3ax3, 4LP\\n' +
'holo, 2ax1, XDN\\n' + 'apo, 2ll3, GSS\\n')
# create fake blastnfilter script that makes csv files
f = open(foo_script, 'w')
f.write('#! /usr/bin/env python\n\n')
f.write('import time\n')
f.write('f = open(\'' + txt_file + '\', \'w\')\n')
f.write('f.write(\'' + txt_contents + '\\n\')\n')
f.write('f.flush()\nf.close()\n')
f.write('time.sleep(360)\n')
f.flush()
f.close()
os.chmod(foo_script, stat.S_IRWXU)
blasttask.run()
self.assertEqual(blasttask.get_error(), None)
self.assertEqual(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
complete_file = os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(complete_file), True)
std_err_file = os.path.join(blasttask.get_dir(), 'foo.py.stderr')
self.assertEqual(os.path.isfile(std_err_file), True)
std_out_file = os.path.join(blasttask.get_dir(), 'foo.py.stdout')
self.assertEqual(os.path.isfile(std_out_file), True)
res = blasttask.get_email_log().rstrip('\n')
res.index('/foo.py')
res.index('# txt files found: 1')
res.index('Output from summary.txt')
finally:
shutil.rmtree(temp_dir)
def test_get_txt_files(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
blast_task = BlastNFilterTask(temp_dir, params)
# try missing directory
self.assertEquals(len(blast_task.get_txt_files()), 0)
blast_task.create_dir()
# try empty directory
self.assertEquals(len(blast_task.get_txt_files()), 0)
# try where only summary.txt exists
summary_file = os.path.join(blast_task.get_dir(), 'summary.txt')
open(summary_file, 'a').close()
self.assertEquals(len(blast_task.get_txt_files()), 0)
# try non txt files only
os.remove(summary_file)
open(os.path.join(blast_task.get_dir(), 'foo.csv'), 'a').close()
self.assertEquals(len(blast_task.get_txt_files()), 0)
# try 1 txt file
open(os.path.join(blast_task.get_dir(), '4vwx.txt'), 'a').close()
self.assertEquals(len(blast_task.get_txt_files()), 1)
# try 1 summary.txt and 1 txt file
summary_file = os.path.join(blast_task.get_dir(), 'summary.txt')
open(summary_file, 'a').close()
self.assertEquals(len(blast_task.get_txt_files()), 1)
# try multiple txt files
open(os.path.join(blast_task.get_dir(), '5vwx.txt'), 'a').close()
self.assertEquals(len(blast_task.get_txt_files()), 2)
finally:
shutil.rmtree(temp_dir)
def test_run_with_blast_success_postanalysis_success_no_summary_file(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.blastnfilter = 'true'
params.postanalysis = '/bin/echo'
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
blasttask.run()
self.assertEqual(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEqual(blasttask.get_error(), None)
complete_file = os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(complete_file), True)
std_err_file = os.path.join(blasttask.get_dir(), 'echo.stderr')
self.assertEqual(os.path.isfile(std_err_file), True)
std_out_file = os.path.join(blasttask.get_dir(), 'echo.stdout')
dataimport = DataImportTask(temp_dir, params)
f = open(std_out_file, 'r')
echo_out = f.read().replace('\n', '')
echo_out.index('--compinchi ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.COMPINCHI_ICH))
echo_out.index(' ' +
os.path.join(temp_dir, blasttask.get_dir_name()))
f.close()
self.assertEqual(os.path.isfile(std_out_file), True)
self.assertEquals(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEquals(
os.path.exists(os.path.join(blasttask.get_dir(),
'true.stderr')), True)
self.assertEquals(
os.path.exists(os.path.join(blasttask.get_dir(),
'true.stdout')), True)
res = blasttask.get_email_log().rstrip('\n')
res.index('/bin/echo')
res.index('# txt files found: 0')
res.index('Output from summary.txt')
finally:
shutil.rmtree(temp_dir)
def run(self):
"""Runs EvaluationTask after verifying dock was good
Method requires can_run() to be called before hand with
successful outcome
Otherwise method invokes D3RTask.start then this method
creates a directory and invokes evaluation script. Upon
completion results are analyzed and success or error status
is set appropriately and D3RTask.end is invoked
"""
super(EvaluationTask, self).run()
if self._can_run is False:
logger.debug(self.get_dir_name() +
' cannot run cause _can_run flag '
'is False')
return
try:
logger.debug('evaluation set to ' + self.get_args().evaluation)
except AttributeError:
self.set_error('evaluation not set')
self.end()
return
try:
logger.debug('pdbdb set to ' + self.get_args().pdbdb)
except AttributeError:
self.set_error('pdbdb not set')
self.end()
return
try:
evaltimeout = self.get_args().evaluationtimeout
logger.debug('Setting evaluation timeout to ' + str(evaltimeout))
except AttributeError:
evaltimeout = None
try:
killdelay = self.get_args().evaluationtimeoutkilldelay
logger.debug('Setting evaluation kill delay to ' + str(killdelay))
except AttributeError:
killdelay = 60
blastnfilter = BlastNFilterTask(self._path, self._args)
challenge = ChallengeDataTask(self._path, self._args)
challdir = os.path.join(challenge.get_dir(),
challenge.get_celpp_challenge_data_dir_name())
#
# --pdbdb <path to pdb.extracted> --dockdir <stage.4.glide> \
# --outdir <path to stage.5.glide.evaluation>
#
cmd_to_run = (self.get_args().evaluation + ' --pdbdb ' +
self.get_args().pdbdb + ' --dockdir ' +
self._docktask.get_dir() + ' --blastnfilterdir ' +
blastnfilter.get_dir() + ' --challengedir ' + challdir +
' --outdir ' + self.get_dir())
eval_name = os.path.basename(self.get_args().evaluation)
ecode = self.run_external_command(eval_name,
cmd_to_run,
False,
timeout=evaltimeout,
kill_delay=killdelay)
# write out evaluate exit code file
self._write_evaluate_exitcode_file(ecode)
# attempt to send evaluation email
try:
self._emailer.send_evaluation_email(self)
self.append_to_email_log(self._emailer.get_message_log())
except Exception as e:
logger.exception('Caught exception trying to send evaluation '
'email')
self.append_to_email_log('Caught exception trying to send '
'evaluation email ' + str(e) + '\n')
# assess the result
self.end()
def test_get_uploadable_files(self):
temp_dir = tempfile.mkdtemp()
try:
# test where directory doesn't even exist
params = D3RParameters()
task = BlastNFilterTask(temp_dir, params)
self.assertEqual(task.get_uploadable_files(), [])
# test on empty dir
task.create_dir()
# test with blastn filter log
logfile = os.path.join(task.get_dir(),
BlastNFilterTask.BLASTNFILTER_LOG)
open(logfile, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 1)
self.assertEqual(flist[0], logfile)
# test with dockable.xslx
dockable = os.path.join(task.get_dir(),
BlastNFilterTask.DOCKABLE_XSLX)
open(dockable, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 2)
flist.index(dockable)
# test with summary.txt
summary = os.path.join(task.get_dir(),
BlastNFilterTask.SUMMARY_TXT)
open(summary, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 3)
flist.index(summary)
# test with 1 txt file
txt_one = os.path.join(task.get_dir(), '1fcz.txt')
open(txt_one, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 4)
flist.index(txt_one)
# test with 2 txt files
txt_two = os.path.join(task.get_dir(), '4asd.txt')
open(txt_two, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 5)
flist.index(txt_two)
# test with additional stderr/stdout files
errfile = os.path.join(task.get_dir(), 'blastnfilter.py.stderr')
open(errfile, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 6)
flist.index(errfile)
outfile = os.path.join(task.get_dir(), 'blastnfilter.py.stdout')
open(outfile, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 7)
flist.index(outfile)
flist.index(errfile)
flist.index(txt_two)
flist.index(txt_one)
flist.index(summary)
flist.index(dockable)
flist.index(logfile)
finally:
shutil.rmtree(temp_dir)
def test_can_run(self):
tempDir = tempfile.mkdtemp()
try:
# try where makeblastdb is not complete
params = D3RParameters()
blastTask = BlastNFilterTask(tempDir, params)
self.assertEqual(blastTask.can_run(), False)
# try where makeblastdb failed
blastDb = MakeBlastDBTask(tempDir, params)
blastDb.create_dir()
errorFile = os.path.join(blastDb.get_path(),
blastDb.get_dir_name(),
D3RTask.ERROR_FILE)
open(errorFile, 'a').close()
self.assertEqual(blastTask.can_run(), False)
self.assertEqual(blastTask.get_error(),
'makeblastdb task has error status')
# try where data import is not complete
completeFile = os.path.join(blastDb.get_path(),
blastDb.get_dir_name(),
D3RTask.COMPLETE_FILE)
open(completeFile, 'a').close()
self.assertEqual(blastTask.can_run(), False)
self.assertEqual(blastTask.get_error(),
'dataimport task has ' + 'notfound status')
# try where data import failed
dataImport = DataImportTask(tempDir, params)
dataImport.create_dir()
errorFile = os.path.join(dataImport.get_path(),
dataImport.get_dir_name(),
D3RTask.ERROR_FILE)
open(errorFile, 'a').close()
self.assertEqual(blastTask.can_run(), False)
self.assertEqual(blastTask.get_error(),
'dataimport task has error status')
# try where blast can run
os.remove(errorFile)
completeFile = os.path.join(dataImport.get_dir(),
D3RTask.COMPLETE_FILE)
open(completeFile, 'a').close()
self.assertEqual(blastTask.can_run(), True)
self.assertEqual(blastTask.get_error(), None)
# try where blast exists
blastTask.create_dir()
self.assertEqual(blastTask.can_run(), False)
self.assertEqual(
blastTask.get_error(),
blastTask.get_dir_name() + ' already exists and' +
' status is unknown')
# try where blast is complete
completeFile = os.path.join(blastTask.get_path(),
blastTask.get_dir_name(),
D3RTask.COMPLETE_FILE)
open(completeFile, 'a').close()
self.assertEqual(blastTask.can_run(), False)
self.assertEqual(blastTask.get_error(), None)
finally:
shutil.rmtree(tempDir)
def test_run_with_blast_success_useoldseq_and_postanalysis_fail(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.blastnfilter = '/bin/echo'
params.postanalysis = os.path.join(temp_dir, 'foo.py')
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
txt_file = os.path.join(blasttask.get_dir(), 'summary.txt')
txt_contents = ('INPUT SUMMARY\\n' + ' sequences: 177\\n' +
' complexes: 149\\n')
# create fake blastnfilter script that makes csv files
f = open(params.postanalysis, 'w')
f.write('#! /usr/bin/env python\n\n')
f.write('f = open(\'' + txt_file + '\', \'w\')\n')
f.write('f.write(\'' + txt_contents + '\\n\')\n')
f.write('f.flush()\nf.close()\n')
f.flush()
f.close()
os.chmod(params.postanalysis, stat.S_IRWXU)
blasttask.run()
self.assertEqual(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEqual(blasttask.get_error(), None)
complete_file = os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(complete_file), True)
std_err_file = os.path.join(blasttask.get_dir(), 'echo.stderr')
self.assertEqual(os.path.isfile(std_err_file), True)
std_out_file = os.path.join(blasttask.get_dir(), 'echo.stdout')
dataimport = DataImportTask(temp_dir, params)
makeblast = MakeBlastDBTask(temp_dir, params)
f = open(std_out_file, 'r')
echo_out = f.read().replace('\n', '')
echo_out.index('--nonpolymertsv ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.NONPOLYMER_TSV))
echo_out.index(' --sequencetsv ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.OLDSEQUENCE_TSV))
echo_out.index(' --pdbblastdb ' +
os.path.join(temp_dir, makeblast.get_dir_name()))
echo_out.index(' --compinchi ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.COMPINCHI_ICH))
echo_out.index(' --outdir ' +
os.path.join(temp_dir, blasttask.get_dir_name()))
echo_out.index(' --crystalpH ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.CRYSTALPH_TSV))
echo_out.index(' --pdbdb /pdbdb ')
f.close()
self.assertEqual(os.path.isfile(std_out_file), True)
self.assertEquals(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEquals(
os.path.exists(
os.path.join(blasttask.get_dir(), 'foo.py.stderr')), True)
self.assertEquals(
os.path.exists(
os.path.join(blasttask.get_dir(), 'foo.py.stdout')), True)
res = blasttask.get_email_log().rstrip('\n')
res.index('/bin/echo')
res.index('# txt files found: 0')
res.index('Output from summary.txt')
res.index(' sequences: 177')
res.index(' complexes: 149')
res.index(dataimport.get_sequence_tsv() +
' file not found falling back to ' +
dataimport.get_oldsequence_tsv())
finally:
shutil.rmtree(temp_dir)
