def __init__(self, nodeid, alg_phylip_file, constrain_id, model,
seqtype, conf, confname, parts_id=None):
GLOBALS["citator"].add(PHYML_CITE)
base_args = OrderedDict({
"--model": "",
"--no_memory_check": "",
"--quiet": "",
"--constraint_tree": ""})
self.confname = confname
self.conf = conf
self.constrain_tree = None
if constrain_id:
self.constrain_tree = db.get_dataid(constrain_id, DATATYPES.constrain_tree)
self.alg_phylip_file = alg_phylip_file
TreeTask.__init__(self, nodeid, "tree", "Phyml",
base_args, conf[confname])
if seqtype == "aa":
self.model = model or conf[confname]["_aa_model"]
elif seqtype == "nt":
self.model = model or conf[confname]["_nt_model"]
self.seqtype = seqtype
self.lk = None
self.init()
def __init__(self, nodeid, alg_file, constrain_id, model, seqtype, conf, confname, parts_id=None):
GLOBALS["citator"].add(FASTTREE_CITE)
self.confname = confname
self.conf = conf
self.alg_phylip_file = alg_file
self.constrain_tree = None
if constrain_id:
self.constrain_tree = db.get_dataid(constrain_id, DATATYPES.constrain_alg)
self.alg_basename = basename(self.alg_phylip_file)
self.seqtype = seqtype
self.tree_file = ""
if model:
log.warning("FastTree does not support model selection")
self.model = None
self.lk = None
base_args = OrderedDict()
base_args["-nopr"] = ""
if self.seqtype == "nt":
base_args["-gtr -nt"] = ""
elif self.seqtype == "aa":
pass
else:
raise ValueError("Unknown seqtype %s" % self.seqtype)
TreeTask.__init__(self, nodeid, "tree", "FastTree", base_args, self.conf[confname])
self.init()
def __init__(self, nodeid, alg_file, constrain_id, model,
seqtype, conf, confname, parts_id=None):
GLOBALS["citator"].add(RAXML_CITE)
base_args = OrderedDict()
self.bootstrap = conf[confname].get("_bootstrap", None)
model = model or conf[confname]["_aa_model"]
self.confname = confname
self.conf = conf
self.alg_phylip_file = alg_file
try:
self.constrain_tree = db.get_dataid(constrain_id, DATATYPES.constrain_tree)
except ValueError:
self.constrain_tree = None
self.partitions_file = parts_id
TreeTask.__init__(self, nodeid, "tree", "RaxML",
base_args, conf[confname])
max_cores = GLOBALS["_max_cores"]
appname = conf[confname]["_app"]
if max_cores > 1:
threads = conf["threading"].get("raxml-pthreads")
if threads > 1:
appname = appname.replace("raxml", "raxml-pthreads")
raxml_bin = conf["app"][appname]
else:
appname = appname.replace("raxml-pthreads", "raxml")
threads = 1
raxml_bin = conf["app"][appname]
self.raxml_bin = raxml_bin
self.threads = threads
self.seqtype = seqtype
# Process raxml options
method = conf[confname].get("_method", "GAMMA").upper()
if seqtype.lower() == "aa":
self.model_string = 'PROT%s%s' %(method, model.upper())
self.model = model
elif seqtype.lower() == "nt":
self.model_string = 'GTR%s' %method
self.model = "GTR"
else:
raise ValueError("Unknown seqtype %s", seqtype)
#inv = conf[confname].get("pinv", "").upper()
#freq = conf[confname].get("ebf", "").upper()
self.init()
def finish(self):
lks = []
j = self.jobs[0]
tree_file = os.path.join(j.jobdir,
self.alg_phylip_file+"_phyml_tree.txt")
stats_file = os.path.join(j.jobdir,
self.alg_phylip_file+"_phyml_stats.txt")
m = re.search('Log-likelihood:\s+(-?\d+\.\d+)',
open(stats_file).read())
lk = float(m.groups()[0])
stats = {"lk": lk}
tree = PhyloTree(tree_file)
TreeTask.store_data(self, tree.write(), stats)
def finish(self):
node_info = self.conf["_nodeinfo"][self.nodeid]
target_seqs = node_info.get("target_seqs", set())
out_seqs = node_info.get("out_seqs", set())
all_seqs = list(target_seqs | out_seqs)
if len(all_seqs) == 1:
newick = "%s;" %(all_seqs[0])
elif len(all_seqs) == 2:
newick = "(%s, %s);" %(all_seqs[0], all_seqs[1])
else:
newick = "(%s, (%s));" %(all_seqs[0], ','.join(all_seqs[1:]))
TreeTask.store_data(self, newick, {})
def __init__(self, nodeid, alg_file, constrain_id, model, seqtype,
conf, confname, parts_id=None):
self.confname = confname
self.conf = conf
self.alg_phylip_file = alg_file
self.constrain_tree = None
if constrain_id:
self.constrain_tree = db.get_dataid(constrain_id, DATATYPES.constrain_alg)
self.alg_basename = basename(self.alg_phylip_file)
self.seqtype = seqtype
self.tree_file = ""
self.model = None
self.lk = None
TreeTask.__init__(self, nodeid, "tree", "DummyTree", {}, {})
self.init()
def finish(self):
#first job is the raxml tree
def parse_alrt(match):
dist = match.groups()[0]
support = float(match.groups()[1])/100.0
return "%g:%s" %(support, dist)
if self.bootstrap == "alrt":
alrt_tree_file = os.path.join(self.alrt_job.jobdir,
"RAxML_fastTreeSH_Support." + self.alrt_job.args["-n"])
raw_nw = open(alrt_tree_file).read()
try:
nw, nsubs = re.subn(":(\d+\.\d+)\[(\d+)\]", parse_alrt, raw_nw, flags=re.MULTILINE)
except TypeError:
raw_nw = raw_nw.replace("\n","")
nw, nsubs = re.subn(":(\d+\.\d+)\[(\d+)\]", parse_alrt, raw_nw)
if nsubs == 0:
log.warning("alrt values were not detected in raxml tree!")
tree = Tree(nw)
elif self.bootstrap == "alrt_phyml":
alrt_tree_file = os.path.join(self.alrt_job.jobdir,
self.alg_phylip_file +"_phyml_tree.txt")
tree = Tree(alrt_tree_file)
else:
alrt_tree_file = os.path.join(self.bootd_job.jobdir,
"RAxML_bipartitions." + self.bootd_job.args["-n"])
nw = open(alrt_tree_file).read()
tree = Tree(nw)
tree.support = 100
for n in tree.traverse():
if n.support >1:
n.support /= 100.
else:
n.support = 0
TreeTask.store_data(self, tree.write(), {})
def finish(self):
job = self.jobs[-1]
t = Tree(job.stdout_file)
TreeTask.store_data(self, t.write(), {})