def runBlast(self,short=False):
lS = []
for s in SeqIO.parse(open(self.query),'fasta'):
if short and len(s) <= 30:
lS.append(s)
elif not short:lS.append(s)
self._maxsubstatus = len(lS)
for seqs in slice_it(lS,10):
if self.killed:
logger.debug('Exiting for a kill signal')
return False
if len(seqs) == 0:
continue
self._substatus += len(seqs)
self.updateStatus(sub=True)
if short:
query = os.path.join(self._room,
'KEGGshort_%d.faa'%self._substatus)
out = os.path.join(self._room,
'KEGGshort_%d.xml'%self._substatus)
else:
query = os.path.join(self._room,
'KEGG_%d.faa'%self._substatus)
out = os.path.join(self._room,'KEGG_%d.xml'%self._substatus)
self.out.append(out)
# If recovery, skip the unnecessary scans
if ( self.recover and os.path.exists(query) and
os.path.exists(out)):
# Last test: can it be parsed?
try:
self._blast.parseBlast(out)
for hits in self._blast.getHits(self.evalue):
pass
logger.debug('Skipping slice %s because has already been done'
%query)
continue
except:
pass
oseqs = SeqIO.write(seqs,open(query,'w'),'fasta')
if oseqs != len(seqs):
logger.warning('Query splitting error! Expected %d, '+
'Printed %d'%(len(seqs),oseqs))
if short:
res = self._blast.runBlast(query, self.db, out,
evalue = self.evalue,
ncpus = self.ncpus, task='blastp-short')
else:
res = self._blast.runBlast(query, self.db, out,
evalue = self.evalue,
ncpus = self.ncpus)
if not res:
return False
return True
def plotClusters(X, labels, params=None, method='', prefix='clusters'):
from ductape.common.utils import slice_it
import matplotlib.pyplot as plt
import matplotlib.cm as cm
X = np.array(X)
labels = np.array(labels)
labels_unique = np.unique(labels)
n_clusters_ = len(labels_unique)
figidx = 1
fig = plt.figure(1)
fig.clf()
for x, y in product(range(len(X[0])), repeat=2):
ax = fig.add_subplot(len(X[0]),len(X[0]),figidx)
if figidx%len(X[0]) == 1:
if not params:
ax.set_ylabel(x, rotation='horizontal')
else:
ax.set_ylabel(params[x], rotation='horizontal')
if abs((len(X[0])*len(X[0])) % figidx - len(X[0])) <= len(X[0]):
if not params:
ax.set_xlabel(y)
else:
ax.set_xlabel(params[y])
figidx += 1
color = dict()
j = 0
for i in slice_it(range(255), cols=n_clusters_):
color[j] = cm.RdYlGn(i[0])
j += 1
for k in range(n_clusters_):
my_members = labels == k
ax.plot(X[my_members, y], X[my_members, x], '.', color=color[k])
fig.subplots_adjust(wspace=0, hspace=0)
for ax in fig.axes:
ax.get_xaxis().set_ticks([])
ax.get_yaxis().set_ticks([])
fig.suptitle('Clusters (%s, %s): %d' % (prefix, method, n_clusters_))
fig.savefig('%s_%s.png'%(prefix,method),dpi=300)
def plotClusters(X, labels, params=None, method='', prefix='clusters'):
from ductape.common.utils import slice_it
import matplotlib.pyplot as plt
import matplotlib.cm as cm
X = np.array(X)
labels = np.array(labels)
labels_unique = np.unique(labels)
n_clusters_ = len(labels_unique)
figidx = 1
fig = plt.figure(1)
fig.clf()
for x, y in product(range(len(X[0])), repeat=2):
ax = fig.add_subplot(len(X[0]), len(X[0]), figidx)
if figidx % len(X[0]) == 1:
if not params:
ax.set_ylabel(x, rotation='horizontal')
else:
ax.set_ylabel(params[x], rotation='horizontal')
if abs((len(X[0]) * len(X[0])) % figidx - len(X[0])) <= len(X[0]):
if not params:
ax.set_xlabel(y)
else:
ax.set_xlabel(params[y])
figidx += 1
color = dict()
j = 0
for i in slice_it(range(255), cols=n_clusters_):
color[j] = cm.RdYlGn(i[0])
j += 1
for k in range(n_clusters_):
my_members = labels == k
ax.plot(X[my_members, y], X[my_members, x], '.', color=color[k])
fig.subplots_adjust(wspace=0, hspace=0)
for ax in fig.axes:
ax.get_xaxis().set_ticks([])
ax.get_yaxis().set_ticks([])
fig.suptitle('Clusters (%s, %s): %d' % (prefix, method, n_clusters_))
fig.savefig('%s_%s.png' % (prefix, method), dpi=300)
def prepareColors(dReacts, colorrange):
if len(dReacts) == 0:
return {}
maximum = max([dReacts[x] for x in dReacts])
hexs = {}
prev = '#FFFFFF'
i = 1
for color in slice_it(colorrange, cols=maximum):
if len(color) == 0:
hexs[i] = prev
else:
hexs[i] = rgb_to_hex(tuple([int(round(x*255))
for x in color[-1][:3]])).upper()
prev = hexs[i]
i += 1
return hexs
def runBlast(self, short=False):
lS = []
for s in SeqIO.parse(open(self.query), 'fasta'):
if short and len(s) <= 30:
lS.append(s)
elif not short:
lS.append(s)
self._maxsubstatus = len(lS)
for seqs in slice_it(lS, 10):
if self.killed:
logger.debug('Exiting for a kill signal')
return False
if len(seqs) == 0:
continue
self._substatus += len(seqs)
self.updateStatus(sub=True)
if short:
query = os.path.join(self._room,
'KEGGshort_%d.faa' % self._substatus)
out = os.path.join(self._room,
'KEGGshort_%d.xml' % self._substatus)
else:
query = os.path.join(self._room,
'KEGG_%d.faa' % self._substatus)
out = os.path.join(self._room, 'KEGG_%d.xml' % self._substatus)
self.out.append(out)
# If recovery, skip the unnecessary scans
if (self.recover and os.path.exists(query)
and os.path.exists(out)):
# Last test: can it be parsed?
try:
self._blast.parseBlast(out)
for hits in self._blast.getHits(self.evalue):
pass
logger.debug(
'Skipping slice %s because has already been done' %
query)
continue
except:
pass
oseqs = SeqIO.write(seqs, open(query, 'w'), 'fasta')
if oseqs != len(seqs):
logger.warning('Query splitting error! Expected %d, ' +
'Printed %d' % (len(seqs), oseqs))
if short:
res = self._blast.runBlast(query,
self.db,
out,
evalue=self.evalue,
ncpus=self.ncpus,
task='blastp-short')
else:
res = self._blast.runBlast(query,
self.db,
out,
evalue=self.evalue,
ncpus=self.ncpus)
if not res:
return False
return True