Esempio n. 1
0
    def __call__(self, threshold, alphabet_len=None, use_loose=True):
        """Returns a MotifResults object containing a list of significant Motifs
        
            - When a Module is found that is less than the threshold, a new 
                Motif object is created and added to the list of significant
                motifs.
        """
        # Create new MotifResults object
        k_word_results = MotifResults()
        k_word_results.Modules = []
        k_word_results.Motifs = []
        # Add alignment to MotifResults
        k_word_results.Alignment = self.Alignment
        # Give each module a unique ID
        module_id = 0
        # For each module
        for i, module in enumerate(self.Modules):

            if use_loose:
                p_curr = self.getPValueLoose(module, alphabet_len)
            else:
                p_curr = self.getPValueStrict(module, alphabet_len)
            # If the P value is less than or equal to the threshold
            if p_curr <= threshold:
                # Add the P value to the current module
                module.Pvalue = p_curr
                module.ID = module_id
                module_id += 1
                # Append the module to the k_word_results Modules list
                k_word_results.Modules.append(module)
                # Create a Motif from the module and append to Motifs list
                k_word_results.Motifs.append(Motif(module))
        return k_word_results
Esempio n. 2
0
    def __call__(self, threshold, alphabet_len=None, use_loose=True):
        """Returns a MotifResults object containing a list of significant Motifs
        
            - When a Module is found that is less than the threshold, a new 
                Motif object is created and added to the list of significant
                motifs.
        """
        #Create new MotifResults object
        k_word_results = MotifResults()
        k_word_results.Modules = []
        k_word_results.Motifs = []
        #Add alignment to MotifResults
        k_word_results.Alignment = self.Alignment
        #Give each module a unique ID
        module_id = 0
        #For each module
        for i, module in enumerate(self.Modules):

            if use_loose:
                p_curr = self.getPValueLoose(module, alphabet_len)
            else:
                p_curr = self.getPValueStrict(module, alphabet_len)
            #If the P value is less than or equal to the threshold
            if p_curr <= threshold:
                #Add the P value to the current module
                module.Pvalue = p_curr
                module.ID = module_id
                module_id += 1
                #Append the module to the k_word_results Modules list
                k_word_results.Modules.append(module)
                #Create a Motif from the module and append to Motifs list
                k_word_results.Motifs.append(Motif(module))
        return k_word_results
Esempio n. 3
0
def GibbsParser(lines, truncate_len=None, strict=True):
    """Returns a MotifResults object given a Gibbs Sampler results file.
    
        - only works with results from command line version of Gibbs Sampler
    """
    try:
        #Get sequence and motif blocks
        sequence_block, motif_block = get_sequence_and_motif_blocks(lines)
    except Exception as e:
        if strict:
            raise e
        else:
            return None
    #Create MotifResults object
    gibbs_motif_results = MotifResults()

    #Get sequence map
    sequence_map = get_sequence_map(sequence_block)
    #Get motif blocks
    motif_blocks = get_motif_blocks(motif_block)
    #Get modules
    for module in motif_blocks:
        if module[1] == 'No Motifs Detected':
            print("No Modules detcted!!", module[0])
            continue
        for cur_smod in build_module_objects(module, sequence_map, truncate_len=truncate_len):
            gibbs_motif_results.Modules.append(cur_smod)
    module_ids_to_int(gibbs_motif_results.Modules)
    
    for ct, module in enumerate(gibbs_motif_results.Modules):
        gibbs_motif_results.Motifs.append(Motif(module))
    gibbs_motif_results.Alphabet=module.Alphabet
    return gibbs_motif_results
Esempio n. 4
0
def GibbsParser(lines, truncate_len=None, strict=True):
    """Returns a MotifResults object given a Gibbs Sampler results file.
    
        - only works with results from command line version of Gibbs Sampler
    """
    try:
        #Get sequence and motif blocks
        sequence_block, motif_block = get_sequence_and_motif_blocks(lines)
    except Exception as e:
        if strict:
            raise e
        else:
            return None
    #Create MotifResults object
    gibbs_motif_results = MotifResults()

    #Get sequence map
    sequence_map = get_sequence_map(sequence_block)
    #Get motif blocks
    motif_blocks = get_motif_blocks(motif_block)
    #Get modules
    for module in motif_blocks:
        if module[1] == 'No Motifs Detected':
            print("No Modules detcted!!", module[0])
            continue
        for cur_smod in build_module_objects(module,
                                             sequence_map,
                                             truncate_len=truncate_len):
            gibbs_motif_results.Modules.append(cur_smod)
    module_ids_to_int(gibbs_motif_results.Modules)

    for ct, module in enumerate(gibbs_motif_results.Modules):
        gibbs_motif_results.Motifs.append(Motif(module))
    gibbs_motif_results.Alphabet = module.Alphabet
    return gibbs_motif_results
Esempio n. 5
0
def MemeParser(lines, allowed_ids=[]):
    """Returns a MotifResults object given a MEME results file.
    """
    warnings = []
    #Create MotifResults object
    meme_motif_results = MotifResults()
    #Get main block and alphabet
    main_block, alphabet = getDataBlock(lines)
    #Add alphabet to MotifResults object
    meme_motif_results.MolType = alphabet
    #Get command line summary block and module blocks
    command_block, module_blocks = getCommandModuleBlocks(main_block)
    if command_block:
        #Extract command line data and put in dict
        parameters_list = extractCommandLineData(command_block)
        #Add parameters dict to MotifResults object parameters
        meme_motif_results.Parameters = parameters_list
    #make sure modules were found
    if len(module_blocks) > 0:
        #Get Summary of motifs block
        summary_block = getSummaryBlock(module_blocks[-1])
        #Extract summary data and get summary_dict
        summary_dict = extractSummaryData(summary_block)
        seq_names = summary_dict['CombinedP'].keys()
        if allowed_ids:
            remap_dict,warning = make_remap_dict(seq_names,allowed_ids)
            if warning:
                warnings.append(warning)
            sd = {}
            for k,v in summary_dict['CombinedP'].items():
                sd[remap_dict[k]]=v
            summary_dict['CombinedP']=sd
        else:
            remap_dict = dict(zip(seq_names,seq_names))
        
        #Add summary dict to MotifResults object
        meme_motif_results.Results = summary_dict
        
        #Add warnings to MotifResults object
        meme_motif_results.Results['Warnings']=warnings
        
        #Get blocks for each module
        module_blocks = getModuleDataBlocks(module_blocks)
        #Extract modules and put in MotifResults.Modules list
        for module in module_blocks:
            meme_motif_results.Modules.append(extractModuleData(module,\
                alphabet,remap_dict))
        for module in meme_motif_results.Modules:
            meme_motif_results.Motifs.append(Motif(module))
    return meme_motif_results
Esempio n. 6
0
 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)
Esempio n. 7
0
def MemeParser(lines, allowed_ids=[]):
    """Returns a MotifResults object given a MEME results file.
    """
    warnings = []
    #Create MotifResults object
    meme_motif_results = MotifResults()
    #Get main block and alphabet
    main_block, alphabet = getDataBlock(lines)
    #Add alphabet to MotifResults object
    meme_motif_results.MolType = alphabet
    #Get command line summary block and module blocks
    command_block, module_blocks = getCommandModuleBlocks(main_block)
    if command_block:
        #Extract command line data and put in dict
        parameters_list = extractCommandLineData(command_block)
        #Add parameters dict to MotifResults object parameters
        meme_motif_results.Parameters = parameters_list
    #make sure modules were found
    if len(module_blocks) > 0:
        #Get Summary of motifs block
        summary_block = getSummaryBlock(module_blocks[-1])
        #Extract summary data and get summary_dict
        summary_dict = extractSummaryData(summary_block)
        seq_names = list(summary_dict['CombinedP'].keys())
        if allowed_ids:
            remap_dict, warning = make_remap_dict(seq_names, allowed_ids)
            if warning:
                warnings.append(warning)
            sd = {}
            for k, v in list(summary_dict['CombinedP'].items()):
                sd[remap_dict[k]] = v
            summary_dict['CombinedP'] = sd
        else:
            remap_dict = dict(list(zip(seq_names, seq_names)))

        #Add summary dict to MotifResults object
        meme_motif_results.Results = summary_dict

        #Add warnings to MotifResults object
        meme_motif_results.Results['Warnings'] = warnings

        #Get blocks for each module
        module_blocks = getModuleDataBlocks(module_blocks)
        #Extract modules and put in MotifResults.Modules list
        for module in module_blocks:
            meme_motif_results.Modules.append(extractModuleData(module,\
                alphabet,remap_dict))
        for module in meme_motif_results.Modules:
            meme_motif_results.Motifs.append(Motif(module))
    return meme_motif_results
Esempio n. 8
0
    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),
            }
Esempio n. 9
0
def GibbsParser(lines, truncate_len=None, strict=True):
    """Returns a MotifResults object given a Gibbs Sampler results file.
    
        - only works with results from command line version of Gibbs Sampler
    """
    try:
        #Get sequence and motif blocks
        sequence_block, motif_block = get_sequence_and_motif_blocks(lines)
    except Exception, e:
        if strict:
            raise e
        else:
            return None
    #Create MotifResults object
    gibbs_motif_results = MotifResults()

    #Get sequence map
    sequence_map = get_sequence_map(sequence_block)
    #Get motif blocks
    motif_blocks = get_motif_blocks(motif_block)
    #Get modules
    for module in motif_blocks:
        if module[1] == 'No Motifs Detected':
            print "No Modules detcted!!", module[0]
            continue
        for cur_smod in build_module_objects(module,
                                             sequence_map,
                                             truncate_len=truncate_len):
            gibbs_motif_results.Modules.append(cur_smod)
    module_ids_to_int(gibbs_motif_results.Modules)
Esempio n. 10
0
def GibbsParser(lines, truncate_len=None, strict=True):
    """Returns a MotifResults object given a Gibbs Sampler results file.
    
        - only works with results from command line version of Gibbs Sampler
    """
    try:
        #Get sequence and motif blocks
        sequence_block, motif_block = get_sequence_and_motif_blocks(lines)
    except Exception, e:
        if strict:
            raise e
        else:
            return None
    #Create MotifResults object
    gibbs_motif_results = MotifResults()

    #Get sequence map
    sequence_map = get_sequence_map(sequence_block)
    #Get motif blocks
    motif_blocks = get_motif_blocks(motif_block)
    #Get modules
    for module in motif_blocks:
        if module[1] == 'No Motifs Detected':
            print "No Modules detcted!!", module[0]
            continue
        for cur_smod in build_module_objects(module, sequence_map, truncate_len=truncate_len):
            gibbs_motif_results.Modules.append(cur_smod)
    module_ids_to_int(gibbs_motif_results.Modules)
    
    for ct, module in enumerate(gibbs_motif_results.Modules):