Exemple #1
0
    def set_ref_data(self):
        """Write the reference sequence to a fasta file for this specific target if it does not
        exist.

        Args:
            None
        Returns:
            None
        Raise:
            None
        """

        # Write reference fasta file if needed.
        for i in range(len(self.files['target_ref_fn'])):
            fn = self.files['target_ref_fn'][i]
            direction = "forward" if fn.find("forward") != -1 else "reverse"
            utils.log(self.loggingName, 'info', 'Extracting refseq sequence and writing %s' % fn)
            utils.extract_refseq_fa(self.values, self.paths['ref_data'], self.params.get_param('reference_fasta'), direction, fn)

        # If using blatn for target realignment, the db must be available.
        blastn = self.params.get_param('blast')
        if blastn is not None:
            # Check if blast db files are available for each target.
            if not os.path.isfile(self.files['target_ref_fn'][0] + '.nin'):
                makedb = os.path.join(os.path.split(blastn)[0], 'makeblastdb')  # Create blast db
                cmd = "%s -in %s -dbtype 'nucl' -out %s" % (makedb, self.files['target_ref_fn'][0], self.files['target_ref_fn'][0])
                utils.log(self.loggingName, 'info', 'Creating blast db files for target %s with reference file %s' % (self.name, self.files['target_ref_fn'][0]))
                p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True)
                output, errors = p.communicate()
                if errors != '':
                    utils.log(self.loggingName, 'debug', 'Failed to make blast db files using reference file %s' % self.files['target_ref_fn'][0])
Exemple #2
0
    def set_ref_data(self):

        '''
        '''

        # Write rmask bed file if needed.
        # if not self.params.opts['keep_repeat_regions'] and 'repeat_mask_file' in self.params.opts: 
        #   self.logger.info('Extracting repeat mask regions for target gene %s.' % self.name)
        #   self.repeat_mask = setup_rmask(self.get_values(), self.paths['ref_data'], self.params.opts['repeat_mask_file'])
         
        # Write reference fasta file if needed.
        for target_refseq_fn in self.files['target_ref_fn']:
            direction = "forward"
            if target_refseq_fn.find("forward") == -1:
                direction = "reverse"
            utils.log(self.logging_name, 'info', 'Extracting refseq sequence and writing %s' % target_refseq_fn)
            utils.extract_refseq_fa(self.get_values(), self.paths['ref_data'], self.params.get_param('reference_fasta'), direction, target_refseq_fn, self.params.get_param('buffer_size'))
Exemple #3
0
    def set_ref_data(self):

        '''
        '''

        # Write rmask bed file if needed.
        # if not self.params.opts['keep_repeat_regions'] and 'repeat_mask_file' in self.params.opts:
        #   self.logger.info('Extracting repeat mask regions for target gene %s.' % self.name)
        #   self.repeat_mask = setup_rmask(self.get_values(), self.paths['ref_data'], self.params.opts['repeat_mask_file'])

        # Write reference fasta file if needed.
        for target_refseq_fn in self.files['target_ref_fn']:
            direction = "forward"
            if target_refseq_fn.find("forward") == -1:
                direction = "reverse"
            utils.log(self.logging_name, 'info', 'Extracting refseq sequence and writing %s' % target_refseq_fn)
            utils.extract_refseq_fa(self.get_values(), self.paths['ref_data'], self.params.get_param('reference_fasta'), direction, target_refseq_fn, self.params.get_param('buffer_size'))
Exemple #4
0
    def set_ref_data(self):
        """Write the reference sequence to a fasta file for this specific target if it does not
        exist.

        Args:
            None
        Returns:
            None
        Raise:
            None
        """

        # Write reference fasta file if needed.
        for i in range(len(self.files['target_ref_fn'])):
            fn = self.files['target_ref_fn'][i]
            direction = "forward" if fn.find("forward") != -1 else "reverse"
            utils.log(self.loggingName, 'info',
                      'Extracting refseq sequence and writing %s' % fn)
            utils.extract_refseq_fa(self.values, self.paths['ref_data'],
                                    self.params.get_param('reference_fasta'),
                                    direction, fn)

        # If using blatn for target realignment, the db must be available.
        blastn = self.params.get_param('blast')
        if blastn is not None:
            # Check if blast db files are available for each target.
            if not os.path.isfile(self.files['target_ref_fn'][0] + '.nin'):
                makedb = os.path.join(os.path.split(blastn)[0],
                                      'makeblastdb')  # Create blast db
                cmd = "%s -in %s -dbtype 'nucl' -out %s" % (
                    makedb, self.files['target_ref_fn'][0],
                    self.files['target_ref_fn'][0])
                utils.log(
                    self.loggingName, 'info',
                    'Creating blast db files for target %s with reference file %s'
                    % (self.name, self.files['target_ref_fn'][0]))
                p = subprocess.Popen(cmd,
                                     stdout=subprocess.PIPE,
                                     stderr=subprocess.PIPE,
                                     shell=True)
                output, errors = p.communicate()
                if errors != '':
                    utils.log(
                        self.loggingName, 'debug',
                        'Failed to make blast db files using reference file %s'
                        % self.files['target_ref_fn'][0])