def test_init(self):
"""Init should properly initialize Module object."""
module = Module(
data={(self.modules_no_e[0].Location.SeqId, self.modules_no_e[0].Location.Start): self.modules_no_e[0]}
)
self.assertEqual(module.Template, None)
self.assertEqual(module.Alphabet, ASCII.Alphabet)
self.assertEqual(module.Pvalue, None)
self.assertEqual(module.Evalue, None)
self.assertEqual(module.keys(), [("seq0", 1)])
self.assertEqual(module.values(), [ModuleInstance(self.sequences[0], self.locations[0], self.Pvalues[0])])
def test_init(self):
"""Init should properly initialize Module object."""
module = Module(data={(self.modules_no_e[0].Location.SeqId,
self.modules_no_e[0].Location.Start): \
self.modules_no_e[0]})
self.assertEqual(module.Template, None)
self.assertEqual(module.Alphabet, ASCII.Alphabet)
self.assertEqual(module.Pvalue, None)
self.assertEqual(module.Evalue, None)
self.assertEqual(module.keys(),[('seq0',1)])
self.assertEqual(module.values(),[ModuleInstance(self.sequences[0],
self.locations[0],
self.Pvalues[0])])
def __call__(self, word_length):
"""Builds a dict of all Modules and a list of their order.
- module_dict is {module pattern:Module object}
- module_order is a list in descending order of their count.
"""
#Dictionary keying k-word to Module
self.ModuleDict = {}
#For each sequence in the alignment
for key, seq in self.Alignment.items():
#For each position in seq till end - word_length
for i in range(0, len(seq) - word_length + 1):
#Get the current k-word
word = seq[i:i + word_length].tostring()
#Create a location object
location = Location(key, i, i + word_length)
#Create a ModuleInstance
curr_instance = ModuleInstance(word, location)
#Check to see if pattern is already in dict
if word in self.ModuleDict:
#Add instance to Module
self.ModuleDict[word][(key, i)] = curr_instance
#Not in dict
else:
#Create a new module and add to dict
self.ModuleDict[word]=Module({(key,i):curr_instance},\
MolType=self.MolType)
#Get list of counts
module_counts = \
[(len(mod.Names),word) for word,mod in self.ModuleDict.items()]
#Sort and put in descending order
module_counts.sort()
module_counts.reverse()
#Get list of only the words in descending order
self.ModuleOrder = [word for i, word in module_counts]
def test_cmp(self):
"""Module objects should sort properly with __cmp__ overwritten."""
pvals_sorted = [3e-010, 0.002,
0.0020100000000000001,
0.0094000000000000004,
0.10000000000000001,
0.59999999999999998,
0.59999999999999998]
evals_sorted = [.9,
.02,
.900001,
.09,
.006,
.0000003,
.0200000001,
]
modules = []
for instance, pvalue, evalue in zip(self.modules_no_e,
self.Pvalues,
self.Evalues):
modules.append(Module({(instance.Location.SeqId,
instance.Location.Start):instance},
Pvalue=pvalue,
Evalue=evalue))
modules.sort()
for ans, p, e in zip(modules, pvals_sorted, evals_sorted):
self.assertEqual(ans.Pvalue, p)
self.assertEqual(ans.Evalue, e)
def build_module_objects(motif_block, sequence_map, truncate_len=None):
"""Returns module object given a motif_block and sequence_map.
- motif_block is list of lines resulting from calling get_motif_blocks
- sequence_map is the mapping between Gibbs sequence numbering and
sequence id from fasta file.
"""
#Get motif id
motif_id = motif_block[0].strip().split()[-1]
#Get motif_list
motif_list = get_motif_sequences(motif_block)
#Get motif p-value
motif_p = get_motif_p_value(motif_block)
#Guess alphabet from motif sequences
alphabet = guess_alphabet(motif_list)
#Create Module object(s)
gibbs_module = {}
module_keys = ["1"]
for motif in motif_list:
seq_id = str(sequence_map[motif[0]])
if truncate_len:
seq_id = seq_id[:truncate_len]
start = motif[1]
seq = motif[2]
sig = motif[3]
motif_num = "1"
#Create Location object
location = Location(seq_id, start, start + len(seq))
#Create ModuleInstance
mod_instance = ModuleInstance(seq, location, sig)
cur_key = (seq_id, start)
gibbs_module[(seq_id, start)] = mod_instance
gibbs_mod = Module(gibbs_module, MolType=alphabet)
gibbs_mod.Pvalue = motif_p
gibbs_mod.ID = motif_id + module_keys[0]
yield gibbs_mod
def build_module_objects(motif_block, sequence_map, truncate_len=None):
"""Returns module object given a motif_block and sequence_map.
- motif_block is list of lines resulting from calling get_motif_blocks
- sequence_map is the mapping between Gibbs sequence numbering and
sequence id from fasta file.
"""
#Get motif id
motif_id = motif_block[0].strip().split()[-1]
#Get motif_list
motif_list = get_motif_sequences(motif_block)
#Get motif p-value
motif_p = get_motif_p_value(motif_block)
#Guess alphabet from motif sequences
alphabet = guess_alphabet(motif_list)
#Create Module object(s)
gibbs_module = {}
module_keys = ["1"]
for motif in motif_list:
seq_id = str(sequence_map[motif[0]])
if truncate_len:
seq_id = seq_id[:truncate_len]
start = motif[1]
seq = motif[2]
sig = motif[3]
motif_num = "1"
#Create Location object
location = Location(seq_id, start, start + len(seq))
#Create ModuleInstance
mod_instance = ModuleInstance(seq,location,sig)
cur_key = (seq_id,start)
gibbs_module[(seq_id,start)]=mod_instance
gibbs_mod = Module(gibbs_module,MolType=alphabet)
gibbs_mod.Pvalue = motif_p
gibbs_mod.ID = motif_id + module_keys[0]
yield gibbs_mod
def extractModuleData(module_data, alphabet, remap_dict):
"""Creates Module object given module_data list.
- Only works on 1 module at a time: only pass in data from one module.
"""
#Create Module object
meme_module = {}
#Only keep first 3 elements of the list
module_data = module_data[:3]
#Get Module general information: module_data[0]
#Only need to keep first line
general_dict = getModuleGeneralInfo(module_data[0][0])
module_length = int(general_dict['width'])
#Get ModuleInstances: module_data[2]
instance_data = module_data[2][4:-2]
for i in xrange(len(instance_data)):
instance_data[i] = instance_data[i].split()
#Create a ModuleInstance object and add it to Module for each instance
for instance in instance_data:
seqId = remap_dict[instance[0]]
start = int(instance[1])-1
Pvalue = float(instance[2])
sequence = instance[4]
#Create Location object for ModuleInstance
location = Location(seqId, start, start + module_length)
#Create ModuleInstance
mod_instance = ModuleInstance(sequence,location,Pvalue)
#Add ModuleInstance to Module
meme_module[(seqId,start)] = mod_instance
meme_module = Module(meme_module, MolType=alphabet)
#Get Multilevel Consensus Sequence
meme_module.ConsensusSequence = getConsensusSequence(module_data[1])
#Pull out desired values from dict
meme_module.Llr = int(general_dict['llr'])
meme_module.Evalue = float(general_dict['E-value'])
meme_module.ID = general_dict['MOTIF']
return meme_module
def test_init(self):
"""Init should properly initialize Motif object."""
module = Module({
('a',3): ModuleInstance('guc', Location('a',3,5)),
('b',3): ModuleInstance('guc', Location('b',3,5)),
('c',8): ModuleInstance('guc', Location('c',8,10)),
})
m = Motif(module)
self.assertEqual(m.Modules,[module])
self.assertEqual(m.Info,None)
def test_init(self):
"""Init should properly initialize MotifResults object."""
module = Module({
('a',3): ModuleInstance('guc', Location('a',3,5)),
('b',3): ModuleInstance('guc', Location('b',3,5)),
('c',8): ModuleInstance('guc', Location('c',8,10)),
})
motif = Motif([module])
results = {'key1':'value1','key2':'value2'}
parameters = {'parameter1':1,'parameter2':2}
mr = MotifResults([module],[motif],results,parameters)
self.assertEqual(mr.Modules,[module])
self.assertEqual(mr.Motifs,[motif])
self.assertEqual(mr.Results,results)
self.assertEqual(mr.parameter1,1)
self.assertEqual(mr.parameter2,2)
def extractModuleData(module_data, alphabet, remap_dict):
"""Creates Module object given module_data list.
- Only works on 1 module at a time: only pass in data from one module.
"""
#Create Module object
meme_module = {}
#Only keep first 3 elements of the list
module_data = module_data[:3]
#Get Module general information: module_data[0]
#Only need to keep first line
general_dict = getModuleGeneralInfo(module_data[0][0])
module_length = int(general_dict['width'])
#Get ModuleInstances: module_data[2]
instance_data = module_data[2][4:-2]
for i in range(len(instance_data)):
instance_data[i] = instance_data[i].split()
#Create a ModuleInstance object and add it to Module for each instance
for instance in instance_data:
seqId = remap_dict[instance[0]]
start = int(instance[1]) - 1
Pvalue = float(instance[2])
sequence = instance[4]
#Create Location object for ModuleInstance
location = Location(seqId, start, start + module_length)
#Create ModuleInstance
mod_instance = ModuleInstance(sequence, location, Pvalue)
#Add ModuleInstance to Module
meme_module[(seqId, start)] = mod_instance
meme_module = Module(meme_module, MolType=alphabet)
#Get Multilevel Consensus Sequence
meme_module.ConsensusSequence = getConsensusSequence(module_data[1])
#Pull out desired values from dict
meme_module.Llr = int(general_dict['llr'])
meme_module.Evalue = float(general_dict['E-value'])
meme_module.ID = general_dict['MOTIF']
return meme_module
def setUp(self):
"""SetUp for MotifFormatter class tests."""
self.sequences = [
'accucua',
'caucguu',
'accucua',
'cgacucg',
'cgaucag',
'cuguacc',
'cgcauca',
]
self.locations = [
Location('seq0',1,3),
Location('seq1',2,3),
Location('seq1',1,5),
Location('seq1',5,3),
Location('seq2',3,54),
Location('seq2',54,2),
Location('seq3',4,0),
]
self.Pvalues = [
.1,
.002,
.0000000003,
.6,
.0094,
.6,
.00201,
]
self.Evalues = [
.006,
.02,
.9,
.0200000001,
.09,
.0000003,
.900001,
]
self.modules_no_e = []
for i in xrange(7):
self.modules_no_e.append(ModuleInstance(self.sequences[i],
self.locations[i],
self.Pvalues[i]))
self.module_with_template = Module(
{
(self.modules_no_e[0].Location.SeqId,
self.modules_no_e[0].Location.Start):self.modules_no_e[0],
(self.modules_no_e[1].Location.SeqId,
self.modules_no_e[1].Location.Start):self.modules_no_e[1],
(self.modules_no_e[2].Location.SeqId,
self.modules_no_e[2].Location.Start):self.modules_no_e[2],
(self.modules_no_e[3].Location.SeqId,
self.modules_no_e[3].Location.Start):self.modules_no_e[3],
(self.modules_no_e[4].Location.SeqId,
self.modules_no_e[4].Location.Start):self.modules_no_e[4],
(self.modules_no_e[5].Location.SeqId,
self.modules_no_e[5].Location.Start):self.modules_no_e[5],
(self.modules_no_e[6].Location.SeqId,
self.modules_no_e[6].Location.Start):self.modules_no_e[6],
},
Template = 'accgucg', ID='1'
)
self.modules_with_ids =\
[Module({
('a',3): ModuleInstance('guc', Location('a',3,5)),
('b',3): ModuleInstance('guc', Location('b',3,5)),
('c',8): ModuleInstance('guc', Location('c',8,10)),
},ID='1'),
Module({
('a',7): ModuleInstance('cca', Location('a',7,9)),
('b',7): ModuleInstance('cca', Location('b',7,9)),
('c',11): ModuleInstance('cca',Location('c',11,13)),
},ID='2'),
Module({
('a',10): ModuleInstance('gca',Location('a',10,12)),
('b',10): ModuleInstance('gca',Location('b',10,12)),
('c',14): ModuleInstance('gca',Location('c',14,12)),
},ID='3'),
Module({
('a',13): ModuleInstance('ggg',Location('a',13,15)),
('b',13): ModuleInstance('ggg',Location('b',13,15)),
('c',18): ModuleInstance('ggg',Location('c',18,20)),
},ID='4'),
]
self.motifs_with_ids = map(Motif,self.modules_with_ids)
self.motif_results = MotifResults(Modules=self.modules_with_ids,\
Motifs=self.motifs_with_ids)
self.color_map = {'1':"""background-color: #0000FF; ; font-family: 'Courier New', Courier""",
'2':"""background-color: #FFFF00; ; font-family: 'Courier New', Courier""",
'3':"""background-color: #00FFFF; ; font-family: 'Courier New', Courier""",
'4':"""background-color: #FF00FF; ; font-family: 'Courier New', Courier""",
}
self.color_map_rgb = {
'color_1':(0.0,0.0,1.0),
'color_2':(1.0,1.0,0.0),
'color_3':(0.0,1.0,1.0),
'color_4':(1.0,0.0,1.0),
}
def setUp(self):
"""SetUp for Module class tests."""
self.sequences = [
'accucua',
'caucguu',
'accucua',
'cgacucg',
'cgaucag',
'cuguacc',
'cgcauca',
]
self.locations = [
Location('seq0',1,3),
Location('seq1',2,3),
Location('seq1',1,5),
Location('seq1',5,3),
Location('seq2',3,54),
Location('seq2',54,2),
Location('seq3',4,0),
]
self.Pvalues = [
.1,
.002,
.0000000003,
.6,
.0094,
.6,
.00201,
]
self.Evalues = [
.006,
.02,
.9,
.0200000001,
.09,
.0000003,
.900001,
]
self.modules_no_e = []
for i in xrange(7):
self.modules_no_e.append(ModuleInstance(self.sequences[i],
self.locations[i],
self.Pvalues[i]))
self.modules_p_and_e = []
for i in xrange(7):
self.modules_p_and_e.append(ModuleInstance(self.sequences[i],
self.locations[i],
self.Pvalues[i],
self.Evalues[i]))
self.module_no_template = Module(
{
(self.modules_no_e[0].Location.SeqId,
self.modules_no_e[0].Location.Start):self.modules_no_e[0],
(self.modules_no_e[1].Location.SeqId,
self.modules_no_e[1].Location.Start):self.modules_no_e[1],
(self.modules_no_e[2].Location.SeqId,
self.modules_no_e[2].Location.Start):self.modules_no_e[2],
(self.modules_no_e[3].Location.SeqId,
self.modules_no_e[3].Location.Start):self.modules_no_e[3],
(self.modules_no_e[4].Location.SeqId,
self.modules_no_e[4].Location.Start):self.modules_no_e[4],
(self.modules_no_e[5].Location.SeqId,
self.modules_no_e[5].Location.Start):self.modules_no_e[5],
(self.modules_no_e[6].Location.SeqId,
self.modules_no_e[6].Location.Start):self.modules_no_e[6],
}
)
self.module_with_template = Module(
{
(self.modules_no_e[0].Location.SeqId,
self.modules_no_e[0].Location.Start):self.modules_no_e[0],
(self.modules_no_e[1].Location.SeqId,
self.modules_no_e[1].Location.Start):self.modules_no_e[1],
(self.modules_no_e[2].Location.SeqId,
self.modules_no_e[2].Location.Start):self.modules_no_e[2],
(self.modules_no_e[3].Location.SeqId,
self.modules_no_e[3].Location.Start):self.modules_no_e[3],
(self.modules_no_e[4].Location.SeqId,
self.modules_no_e[4].Location.Start):self.modules_no_e[4],
(self.modules_no_e[5].Location.SeqId,
self.modules_no_e[5].Location.Start):self.modules_no_e[5],
(self.modules_no_e[6].Location.SeqId,
self.modules_no_e[6].Location.Start):self.modules_no_e[6],
},
Template = 'accgucg'
)
