def pipeline(task, wkname, conf=None):
logindent(2)
# Points to npr parameters according to task properties
if not task:
source_seqtype = "aa" if "aa" in GLOBALS["seqtypes"] else "nt"
npr_conf = IterConfig(conf, wkname,
len(GLOBALS["target_species"]),
source_seqtype)
cogconf, cogclass = npr_conf.cog_selector
initial_task = cogclass(GLOBALS["target_species"], set(),
source_seqtype, conf, cogconf)
initial_task.main_tree = main_tree = None
initial_task.threadid = generate_runid()
initial_task.configid = initial_task.threadid
initial_task.target_wkname = wkname
# Register node
db.add_node(initial_task.threadid, initial_task.nodeid,
initial_task.cladeid, initial_task.targets,
initial_task.outgroups)
new_tasks = [initial_task]
else:
conf = GLOBALS[task.configid]
npr_conf = IterConfig(conf, wkname, task.size, task.seqtype)
new_tasks = process_task(task, wkname, npr_conf, conf['_nodeinfo'])
process_new_tasks(task, new_tasks, conf)
logindent(-2)
return new_tasks
def __init__(self, nodeid, task_type, task_name, base_args=None, extra_args=None):
# I want every tree merge instance to be unique (avoids recycling and
# undesired collisions between trees from different threads containing
# the same topology, so I create a random checksum to compute taskid
extra_args = {} if not extra_args else dict(extra_args)
extra_args["_treechecksum"] = generate_runid()
Task.__init__(self, nodeid, task_type, task_name, base_args, extra_args)
def pipeline(task, wkname, conf=None):
logindent(2)
if not task: # in this case, conf is expected
source_seqtype = "aa" if "aa" in GLOBALS["seqtypes"] else "nt"
all_seqs = GLOBALS["target_sequences"]
initial_task = Msf(set(all_seqs), set(),
seqtype=source_seqtype)
initial_task.main_tree = None
initial_task.threadid = generate_runid()
initial_task.configid = initial_task.threadid
initial_task.target_wkname = wkname
# Register node
db.add_node(initial_task.threadid, initial_task.nodeid,
initial_task.cladeid, initial_task.target_seqs,
initial_task.out_seqs)
new_tasks = [initial_task]
else:
conf = GLOBALS[task.configid]
npr_conf = IterConfig(conf, wkname, task.size, task.seqtype)
new_tasks = process_task(task, wkname, npr_conf, conf["_nodeinfo"])
process_new_tasks(task, new_tasks, conf)
logindent(-2)
return new_tasks
def load_jobs(self):
# I want a single phylognetic tree for each cog
from nprlib.workflow.genetree import pipeline
for co in self.cogs:
# Register a new msf task for each COG, using the same
# config file but opening an new tree reconstruction
# thread.
job = Msf(set(co), set(), seqtype = self.seqtype)
job.main_tree = None
job.threadid = generate_runid()
job.configid = self.conf["_configid"]
# This converts the job in a workflow job. As soon as a
# task is done, it will be automatically processed and the
# new tasks will be registered as new jobs.
job.task_processor = pipeline
job.target_wkname = self.genetree_workflow
self.jobs.append(job)
self.cog_ids.add(job.nodeid)