def test_get_most_likely(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = 'examples/evol/data/protamine/PRM1/paml/'
tree.link_to_evol_model (WRKDIR + 'paml/M1/M1.out', 'M1')
tree.link_to_evol_model (WRKDIR + 'paml/M2/M2.out', 'M2')
self.assertEqual(round(tree.get_most_likely ('M2','M1'),16),
round(6.3280740347111373e-10,16))
def test_paml_parser(self):
alignments = [' 3 6\nseq1\tyo\nATGATG\nseq2\nCTGATG\nseq3\nATGTTT\n',
' 3 6\n>seq1\nATGATG\n>seq2\t|prout\nCTGATG\n>seq3\nATGTTT\n',
'>seq1 \nATGATG\n>seq2\nCTGATG\n>seq3\nATGTTT\n']
for ali in alignments:
t = EvolTree('((seq1,seq2),seq3);')
t.link_to_alignment(ali)
self.assertEqual((t & 'seq1').nt_sequence, 'ATGATG')
self.assertEqual((t & 'seq2').nt_sequence, 'CTGATG')
self.assertEqual((t & 'seq3').nt_sequence, 'ATGTTT')
def test_run_codeml(self):
if which('codeml'):
tree = EvolTree('((seq1,seq2),seq3);')
tree.link_to_alignment('>seq1\nATGCTG\n>seq2\nATGCTG\n>seq3\nTTGATG\n')
tree.run_model('fb')
self.assert_('CODONML' in tree.get_evol_model('fb').run)
self.assert_('Time used:' in tree.get_evol_model('fb').run)
self.assert_('end of tree file' in tree.get_evol_model('fb').run)
self.assert_('lnL' in tree.get_evol_model('fb').run)
self.assert_(tree.get_descendants()[0].w > 0)
def test_marking_trees(self):
TREE_PATH = ETEPATH + '/examples/evol/data/S_example/'
tree = EvolTree (TREE_PATH + 'measuring_S_tree.nw')
self.assertEqual(tree.write(),
'((Hylobates_lar,(Gorilla_gorilla,Pan_troglodytes)),Papio_cynocephalus);')
tree.mark_tree ([1, 3, 7] + [2, 6], marks=['#1']*3 + ['#2']*2, verbose=True)
self.assertEqual(tree.write().replace(' ', ''),
'((Hylobates_lar#2,(Gorilla_gorilla#1,Pan_troglodytes#1)#1)#2,Papio_cynocephalus);')
tree.mark_tree (map (lambda x: x.node_id, tree.get_descendants()),
marks=[''] * len (tree.get_descendants()), verbose=False)
self.assertEqual(tree.write().replace(' ', ''),
'((Hylobates_lar,(Gorilla_gorilla,Pan_troglodytes)),Papio_cynocephalus);')
def test_run_slr(self):
if which('Slr'):
tree = EvolTree('((seq1,seq2),seq3);')
tree.link_to_alignment('>seq1\nCTGATTCTT\n>seq2\nCTGATTCTT\n>seq3\nATGATTCTT\n')
tree.run_model('SLR')
self.assert_('Sitewise Likelihood R' in tree.get_evol_model('SLR').run)
self.assert_('Positively selected s' in tree.get_evol_model('SLR').run)
self.assert_('Conserved sites' in tree.get_evol_model('SLR').run)
self.assert_('lnL' in tree.get_evol_model('SLR').run)
def test_call_histface(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = 'examples/evol/data/protamine/PRM1/paml/'
tree.link_to_alignment (WRKDIR + 'alignments.fasta_ali')
tree.link_to_evol_model (WRKDIR + 'paml/M2/M2.out', 'M2.a')
col = {'NS' : 'grey', 'RX' : 'black',
'RX+': 'grey', 'CN' : 'black',
'CN+': 'grey', 'PS' : 'black', 'PS+': 'black'}
col2 = {'NS' : 'white', 'RX' : 'white',
'RX+': 'white', 'CN' : 'white',
'CN+': 'white', 'PS' : 'white', 'PS+': 'white'}
M2a = tree.get_evol_model('M2.a')
try:
import PyQt4
except ImportError:
pass
else:
M2a.set_histface (up=False, kind='stick', hlines=[1.0, 0.3],
hlines_col=['red','grey'], header='ugliest face')
M2a.set_histface (up=False, kind='curve', colors=col2,errors=True,
hlines = [2.5, 1.0, 4.0, 0.5],
header = 'Many lines, error boxes, background black',
hlines_col=['orange', 'yellow', 'red', 'cyan'])
M2a.set_histface (up=False, kind='bar', hlines = [1.0, 0.3],
hlines_col=['black','grey'],colors=col)
self.assertEqual(str(type(M2a.properties['histface'])),
"<class 'ete2.treeview.faces.SequencePlotFace'>")
def test_deep_copy(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = 'examples/evol/data/protamine/PRM1/paml/'
tree.link_to_evol_model (WRKDIR + 'paml/fb/fb.out', 'fb')
fba = deepcopy (tree.get_evol_model('fb'))
tree._models['fb.a'] = fba
self.assertEqual(str(tree.get_evol_model('fb.a')),
str(tree.get_evol_model('fb'))
)
def test_pickling(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = ETEPATH + '/examples/data/protamine/PRM1/paml/'
tree.link_to_alignment (WRKDIR + 'alignments.fasta_ali')
tree.link_to_evol_model (WRKDIR + 'paml/M2/M2.out', 'M2.a')
out = open('blip.pik', 'w')
dump (tree, out)
out.close()
out = open('blip.pik')
tree2 = load (out)
out.close()
os.remove('blip.pik')
self.assertEqual(str(tree2.get_evol_model('M2.a')),
str(tree.get_evol_model('M2.a'))
)
def test_labelling_tree(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = 'examples/evol/data/protamine/PRM1/paml/'
random_swap(tree)
tree.link_to_evol_model (WRKDIR + 'paml/fb/fb.out', 'fb')
self.assert_(check_annotation (tree))
def test_load_model(self):
tree = EvolTree (WRKDIR + 'tree.nw')
tree.workdir = 'examples/evol/data/protamine/PRM1/paml/'
tree.link_to_evol_model (WRKDIR + 'paml/fb/fb.out', 'fb')
tree.link_to_evol_model (WRKDIR + 'paml/M1/M1.out', 'M1')
tree.link_to_evol_model (WRKDIR + 'paml/M2/M2.out', 'M2')
tree.link_to_evol_model (WRKDIR + 'paml/M7/M7.out', 'M7')
tree.link_to_evol_model (WRKDIR + 'paml/M8/M8.out', 'M8')
tree.link_to_alignment (WRKDIR + 'alignments.fasta_ali')
self.assertEqual(sorted(tree._models.keys()),
sorted(['fb', 'M1', 'M2', 'M7', 'M8']))
self.assertEqual(len (tree.get_evol_model('M2').branches), 194)
self.assertEqual(tree.get_evol_model('fb').lnL, -3265.316569)
self.assert_('proportions' in str(tree.get_evol_model('M2')))
self.assert_('p2=' in str(tree.get_evol_model('M2')))
self.assert_('proportions' not in str(tree.get_evol_model('fb')))
self.assert_(' #193' in str(tree.get_evol_model('fb')))
def parse_paml(pamout, model):
'''
parser function for codeml files,
with values of w,dN,dS etc... dependending of the model
tested.
'''
# if multiple dataset in same file we divide the outfile and model.name+x
if not '*' in str(model.properties['params']['ndata']):
divide_data(pamout, model)
return
all_lines = open(pamout).readlines()
# if we do not have tree, load it
if model._tree == None:
from ete2.evol import EvolTree
model._tree = EvolTree(re.findall('\(.*\);', ''.join(all_lines))[2])
model._tree._label_as_paml()
# starts parsing
for i, line in enumerate(all_lines):
if line is '\n':
continue
# codon frequency
if line.startswith('Codon frequencies under model'):
model.stats['codonFreq'] = []
for j in xrange(16):
line = map(float, re.findall('\d\.\d+', all_lines[i + j + 1]))
model.stats['codonFreq'] += [line]
continue
if not model.stats.has_key('codonFreq'):
continue
######################
# start serious staff
line = line.rstrip()
# lnL and number of parameters
if line.startswith('lnL'):
try:
line = re.sub('.* np: *(\d+)\): +(-\d+\.\d+).*', '\\1 \\2',
line)
model.stats['np'] = int(line.split()[0])
model.stats['lnL'] = float(line.split()[1])
except ValueError:
line = re.sub('.* np: *(\d+)\): +(nan).*', '\\1 \\2', line)
model.stats['np'] = int(line.split()[0])
model.stats['lnL'] = float('-inf')
continue
# get labels of internal branches
if line.count('..') >= 2:
labels = re.findall('\d+\.\.\d+', line + ' ')
_check_paml_labels(model._tree, labels, pamout, model)
continue
# retrieve kappa
if line.startswith('kappa '):
try:
model.stats['kappa'] = float(
re.sub('.*(\d+\.\d+).*', '\\1', line))
except ValueError:
model.stats['kappa'] = 'nan'
# retrieve dS dN t w N S and if present, errors. from summary table
if line.count('..') == 1 and line.startswith(' '):
if not re.match(' +\d+\.\.\d+ +\d+\.\d+ ', line):
if re.match(' +( +\d+\.\d+){8}', all_lines[i + 1]):
_get_values(model,
line.split()[0] + ' ' + all_lines[i + 1])
continue
_get_values(model, line)
continue
#MY_PATH = '/home/francisco/toolbox/ete2-codeml/doc/tutorial/examples/'
MY_PATH = ''
TREE_PATH = MY_PATH + re.sub('\./', '', TREE_PATH)
ALG_PATH = MY_PATH + re.sub('\./', '', ALG_PATH )
###
# load tree
print '\n ----> we create a EvolTree object, and give to him a topology, from',
print TREE_PATH
out = True
while out == True:
try:
T = EvolTree(TREE_PATH)
out = False
except:
sys.stderr.write('Bad path for working directory. Enter new path or quit("Q"):\n')
PATH = raw_input('')
if PATH.startswith('q') or PATH.startswith('Q'):
sys.exit()
TREE_PATH = "./measuring_%s_tree.nw" % (typ)
ALG_PATH = "./alignment_%s_measuring_evol.fasta" % (typ)
TREE_PATH = PATH + re.sub('\./', '', TREE_PATH)
ALG_PATH = PATH + re.sub('\./', '', ALG_PATH )
print T
print '\n ----> and an alignment from: \n'+ALG_PATH+'\n\n'
T.link_to_alignment(ALG_PATH)
from ete2.evol import EvolTree
import sys, re
typ = 'S'
#while typ != 'L' and typ != 'S':
# typ = raw_input (\
# "choose kind of example [L]ong or [S]hort, hit [L] or [S]:\n")
TREE_PATH = "data/S_example/measuring_%s_tree.nw" % (typ)
ALG_PATH = "data/S_example/alignment_%s_measuring_evol.fasta" % (typ)
WORKING_PATH = "data/S_example/paml/"
MY_PATH = ''
TREE_PATH = MY_PATH + re.sub('\./', '', TREE_PATH)
ALG_PATH = MY_PATH + re.sub('\./', '', ALG_PATH )
T = EvolTree (TREE_PATH)
T.link_to_alignment (ALG_PATH)
T.workdir = (WORKING_PATH)
T.link_to_evol_model(T.workdir + '/fb/out','fb')
T.link_to_evol_model(T.workdir + '/M1/out','M1')
T.link_to_evol_model(T.workdir + '/M2/out','M2')
T.show(histfaces=['M2'])
sys.stderr.write('\n\nThe End.\n\n')
