Updated vcftools functions

andrew/vcf_calc.py

@@ -32,19 +32,14 @@ def __call__(self, parser, args, value, option_string=None):
                 setattr(args, self.dest, value)
         return customAction
 
-    def parser_confirm_files ():
-        '''Custom action to confirm multiple file exists'''
-        class customAction(argparse.Action):
-            def __call__(self, parser, args, value, option_string=None):
-                if not os.path.isfile(value):
-                    raise IOError # File not found
-                getattr(args, self.dest).append(value)
-        return customAction
+    def metavar_list (var_list):
+        '''Create a formmated metavar list for the help output'''
+        return '{' + ', '.join(var_list) + '}'
 
     vcf_parser = argparse.ArgumentParser(formatter_class = argparse.ArgumentDefaultsHelpFormatter)
 
     # Input arguments.
-    vcf_parser.add_argument("vcfname", metavar='VCF_Input', help = "Input VCF filename", type = str, action = parser_confirm_file())
+    vcf_parser.add_argument("vcfname", metavar = 'VCF_Input', help = "Input VCF filename", type = str, action = parser_confirm_file())
 
     # Other file arguments. Expand as needed
     vcf_parser.add_argument('--out', help = 'Specifies the output filename', type = str,  default = 'out', action = parser_confirm_no_file())
@@ -54,7 +49,7 @@ def __call__(self, parser, args, value, option_string=None):
     statistic_list = ['weir-fst', 'windowed-weir-fst', 'TajimaD', 'pi', 'freq', 'het']
     statistic_default = 'windowed-weir-fst'
 
-    vcf_parser.add_argument('--calc-statistic', metavar = '{' + ', '.join(statistic_list) + '}', help = 'Specifies the statistic to calculate', type = str, choices = statistic_list, default = statistic_default)
+    vcf_parser.add_argument('--calc-statistic', metavar = metavar_list(statistic_list), help = 'Specifies the statistic to calculate', type = str, choices = statistic_list, default = statistic_default)
 
     # Statistic window options
     vcf_parser.add_argument('--statistic-window-size', help = 'Specifies the size of window calculations', type = int, default = 10000)
@@ -124,7 +119,8 @@ def run (passed_arguments = []):
     if vcf_args.calc_statistic == 'windowed-weir-fst':
         # Confirms that at least two population files have been specified
         if not vcf_args.pop_file or len(vcf_args.pop_file) < 2:
-            sys.exit('Two or more population files requried. Please assign using --pop-file')
+            logging.error('Two or more population files requried. Please assign using --pop-file')
+            raise IOError('Two or more population files requried. Please assign using --pop-file')
 
         # Assigns specific vcftools arguments for calculating fst
         vcftools_pop_args = [population_args for population_file in vcf_args.pop_file for population_args in ['--weir-fst-pop', population_file]]
@@ -134,53 +130,57 @@ def run (passed_arguments = []):
         vcftools_call_args.extend(vcftools_pop_args + vcftools_window_args)
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.windowed.weir.fst'
+        vcftools_log_suffix = 'windowed.weir.fst'
 
     elif vcf_args.calc_statistic == 'weir-fst':
         # Confirms that at least two population files have been specified
         if not vcf_args.pop_file or len(vcf_args.pop_file) < 2:
-            sys.exit('Two or more population files requried. Please assign using --pop-file')
+            logging.error('Two or more population files requried. Please assign using --pop-file')
+            raise IOError('Two or more population files requried. Please assign using --pop-file')
 
         # Assigns specific vcftools arguments for calculating site-based fst
         vcftools_call_args.extend([population_args for population_file in vcf_args.pop_file for population_args in ['--weir-fst-pop', population_file]])
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.weir.fst'
+        vcftools_log_suffix = 'weir.fst'
 
     elif vcf_args.calc_statistic == 'TajimaD':
 
         # Assigns all the vcftools arguments for calculating TajimaD
         vcftools_call_args.extend(['--TajimaD', vcf_args.statistic_window_size])
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.Tajima.D'
+        vcftools_log_suffix = 'Tajima.D'
 
     elif vcf_args.calc_statistic == 'pi':
 
         # Assigns all the vcftools arguments for calculating pi
         vcftools_call_args.extend(['--window-pi', vcf_args.statistic_window_size, '--window-pi-step', vcf_args.statistic_window_step])
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.windowed.pi'
+        vcftools_log_suffix = 'windowed.pi'
 
     elif vcf_args.calc_statistic == 'freq':
 
         # Assigns all the vcftools arguments for the allele frequency
         vcftools_call_args.extend(['--freq'])
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.frq'
+        vcftools_log_suffix = 'frq'
 
     elif vcf_args.calc_statistic == 'het':
 
         # Assigns all the vcftools arguments for calculating heterozygosity
         vcftools_call_args.extend(['--het'])
 
         # Assigns the suffix for the vcftools log file
-        vcftools_log_suffix = '.het'
+        vcftools_log_suffix = 'het'
 
     logging.info('vcftools parameters assigned')
 
+    # Confirm the vcftools output and log file do not exist
+    check_for_vcftools_output (vcf_args.out, vcftools_log_suffix, vcftools_log_suffix)
+
     # Assigns the file argument for vcftools
     vcfname_arg = assign_vcftools_input_arg(vcf_args.vcfname)
     logging.info('Input file assigned')

andrew/vcf_filter.py

@@ -32,27 +32,22 @@ def __call__(self, parser, args, value, option_string=None):
                 setattr(args, self.dest, value)
         return customAction
 
-    def parser_confirm_files ():
-        '''Custom action to confirm multiple file exists'''
-        class customAction(argparse.Action):
-            def __call__(self, parser, args, value, option_string=None):
-                if not os.path.isfile(value):
-                    raise IOError # File not found
-                getattr(args, self.dest).append(value)
-        return customAction
+    def metavar_list (var_list):
+        '''Create a formmated metavar list for the help output'''
+        return '{' + ', '.join(var_list) + '}'
 
     vcf_parser = argparse.ArgumentParser(formatter_class = argparse.ArgumentDefaultsHelpFormatter)
 
     # Input arguments.
-    vcf_parser.add_argument("vcfname", metavar='VCF_Input', help = "Input VCF filename", type = str, action = parser_confirm_file())
+    vcf_parser.add_argument("vcfname", metavar = 'VCF_Input', help = "Input VCF filename", type = str, action = parser_confirm_file())
 
     # Other file arguments. Expand as needed
-    vcf_parser.add_argument('--out', help = 'Specifies the output filename', type = str,  default = 'out', action = parser_confirm_no_file())
+    vcf_parser.add_argument('--out', help = 'Specifies the filtered VCF output filename', type = str,  default = 'out', action = parser_confirm_no_file())
 
     out_format_list = ['vcf', 'bcf']
     out_format_default = 'bcf'
 
-    vcf_parser.add_argument('--out-format', metavar = '{' + ', '.join(out_format_list) + '}', help = 'Specifies the output format', type = str, choices = out_format_list, default = out_format_default)
+    vcf_parser.add_argument('--out-format', metavar = metavar_list(out_format_list), help = 'Specifies the output format.', type = str, choices = out_format_list, default = out_format_default)
     ### Filters
 
     # Chromosome filters
@@ -109,6 +104,34 @@ def run (passed_arguments = []):
         Specifies the input VCF filename
     --out : str
         Specifies the output filename
+    --out-format : str
+        Specifies the output format {vcf, bcf} (Default: bcf)
+    --filter-include-chr : list or str
+        Specifies the chromosome(s) to include
+    --filter-exclude-chr : list or str
+        Specifies the chromosome(s) to exclude
+    --filter-from-bp : int
+        Specifies the lower bound of sites to include. May only be used with a single chromosome
+    --filter-to-bp : int
+        Specifies the upper bound of sites to include. May only be used with a single chromosome
+    --filter-include-bed : str
+        Specifies a set of sites to include within a BED file
+    --filter-exclude-bed : str
+        Specifies a set of sites to exclude within a BED file
+    --filter-include-passed : bool
+        Specifies that only sites with the filter flag 'PASS' should be included (Default: False)
+    --filter-include-filtered : list or str
+        Specifies that all sites with the given filter flag should be included
+    --filter-exclude-filtered : list or str
+        Specifies that all sites with the given filter flag should be excluded
+    --filter-include-info : list or str
+        Specifies that all sites with the given info flag should be included
+    --filter-exclude-info : list or str
+        Specifies that all sites with the given info flag should be excluded
+    --filter-min-alleles : int
+        Specifies that only sites with a number of allele >= to the number given should be included
+    --filter-min-alleles : int
+        Specifies that only sites with a number of allele <= to the number given should be included
 
     Returns
     -------
@@ -124,7 +147,6 @@ def run (passed_arguments = []):
     IOError
         Output file already exists
 
-
     '''
 
     # Grab VCF arguments from command line
@@ -194,6 +216,9 @@ def run (passed_arguments = []):
 
     logging.info('vcftools parameters assigned')
 
+    # Confirm the vcftools output and log file do not exist
+    check_for_vcftools_output (vcf_args.out, 'recode.' + vcf_args.out_format, '.filter')
+
     # Assigns the file argument for vcftools
     vcfname_arg = assign_vcftools_input_arg(vcf_args.vcfname)
     logging.info('Input file assigned')
@@ -205,7 +230,7 @@ def run (passed_arguments = []):
 
     # Check that the log file was created correctly, get the suffix for the log file, and create the file
     if check_vcftools_for_errors(vcftools_err):
-        produce_vcftools_log(vcftools_err, vcf_args.out, '.filter')
+        produce_vcftools_log(vcftools_err, vcf_args.out, 'filter')
 
 
 if __name__ == "__main__":

andrew/vcf_phase.py

@@ -32,14 +32,18 @@ def __call__(self, parser, args, value, option_string=None):
                 setattr(args, self.dest, value)
         return customAction
 
+    def metavar_list (var_list):
+        '''Create a formmated metavar list for the help output'''
+        return '{' + ', '.join(var_list) + '}'
+
     phase_parser = argparse.ArgumentParser()
 
     # Input arguments.
     phase_parser.add_argument("vcfname", metavar='VCF_Input', help = "Input VCF filename", type = str, action = parser_confirm_file())
 
     phasing_list = ['beagle', 'shapeit']
     phasing_default = 'beagle'
-    phase_parser.add_argument('--phase-algorithm', metavar = '{' + ', '.join(phasing_list) + '}', help = 'Specifies the phase algorithm to be used', type = str, choices = phasing_list, default = phasing_default)
+    phase_parser.add_argument('--phase-algorithm', metavar = metavar_list(phasing_list), help = 'Specifies the phase algorithm to be used', type = str, choices = phasing_list, default = phasing_default)
 
     # Other basic arguments. Expand as needed
     phase_parser.add_argument('--out', help = 'Defines the output filename', default = 'out', action = parser_confirm_no_file())

andrew/vcf_sampler.py

@@ -29,6 +29,10 @@ def __call__(self, parser, args, value, option_string = None):
                 setattr(args, self.dest, value)
         return customAction
 
+    def metavar_list (var_list):
+        '''Create a formmated metavar list for the help output'''
+        return '{' + ', '.join(var_list) + '}'
+
     sampler_parser = argparse.ArgumentParser()
 
     # Input arguments
@@ -42,14 +46,14 @@ def __call__(self, parser, args, value, option_string = None):
     # Statistic based arguments.
     statistic_list = ['windowed-weir-fst', 'TajimaD']
     statistic_default = 'windowed-weir-fst'
-    sampler_parser.add_argument('--calc-statistic', metavar = '{' + ', '.join(statistic_list) + '}', help = 'Specifies the statistic calculated ', type=str, choices = statistic_list, default = statistic_default)
+    sampler_parser.add_argument('--calc-statistic', metavar = metavar_list(statistic_list), help = 'Specifies the statistic calculated ', type=str, choices = statistic_list, default = statistic_default)
 
     sampler_parser.add_argument('--statistic-window-size', help = 'Specifies the size of window calculations', type = int, default = 10000)
 
     # Sampling methods. Currently mutually exclusive to only allow a single sampling method
     sampling_list = ['uniform', 'random']
     sampling_default = 'random'
-    sampler_parser.add_argument('--sampling-scheme', metavar = '{' + ', '.join(sampling_list) + '}', help = 'Specifies the sampling scheme ', type=str, choices = sampling_list, default = sampling_default)
+    sampler_parser.add_argument('--sampling-scheme', metavar = metavar_list(sampling_list), help = 'Specifies the sampling scheme ', type=str, choices = sampling_list, default = sampling_default)
 
     # Sampling options
     sampler_parser.add_argument('--uniform-bins', help="Number of bins in uniform sampling", type = int, default = 10)
@@ -136,6 +140,10 @@ def run ():
         ----------
         VCF_Input : str
             Specifies the input VCF filename
+        --out : str
+            Specifies the VCF output filename
+        --sample-file : str
+            Specifies the sampled (statistic file) tsv output filename
         --statistic-file : str
             Specifies the statistic file for filtering
         --calc-statistic : str

andrew/vcftools.py

@@ -1,10 +1,49 @@
 import os
 import sys
+import logging
 
 sys.path.insert(0, os.path.abspath(os.path.join(os.pardir,'jared')))
 
 import vcf_reader_func
 
+def check_for_vcftools_output (output_prefix, output_suffix, log_suffix):
+    '''
+        Checks for the previous vcftools output
+
+        Confirms that neither a previous vcftools log or output file exists.
+
+        Parameters
+        ----------
+        output_prefix : str
+            Specifies the prefix used by vcftools, may be the default value or
+            user-defined
+        output_suffix : str
+            Specifies the standard output suffix used by vcftools.
+        log_suffix : str
+            Specifies the log suffix (PPP-specified) without the log extension
+
+        Raises
+        ------
+        IOError
+            If the vcftools standard output exists
+        IOError
+            If the vcftools log file exists
+
+    '''
+    # Check if output file already exists
+    if os.path.isfile(output_prefix + '.' + output_suffix):
+        logging.error('Output file already exists')
+        raise IOError('Output file already exists')
+
+    logging.info('Output file assigned')
+
+    # Check if log file already exists
+    if os.path.isfile(output_prefix + '.' + log_suffix + '.log'):
+        logging.error('Log file already exists')
+        raise IOError('Log file already exists')
+
+    logging.info('Log file assigned')
+
 def check_vcftools_for_errors (vcftools_output):
     '''
         Checks the vcftools stderr for errors
@@ -34,11 +73,13 @@ def check_vcftools_for_errors (vcftools_output):
         # Splits log into list of lines
         vcftools_output_lines = vcftools_output.splitlines()
         # Prints the error(s)
-        sys.exit('\n'.join((output_line for output_line in vcftools_output_lines if output_line.startswith('Error'))))
+        logging.error('\n'.join((output_line for output_line in vcftools_output_lines if output_line.startswith('Error'))))
+        raise Exception('\n'.join((output_line for output_line in vcftools_output_lines if output_line.startswith('Error'))))
 
     # Print output if not completed and no error found. Unlikely to be used, but included.
     else:
-        sys.exit(vcftools_output)
+        logging.error(vcftools_output)
+        raise Exception(vcftools_output)
 
 def produce_vcftools_log (output, filename, function):
     '''
@@ -68,12 +109,9 @@ def produce_vcftools_log (output, filename, function):
             Log file already exists
     '''
 
-    if not os.path.isfile(filename + function + '.log'):
-        vcftools_log_file = open(filename + function + '.log','w')
-        vcftools_log_file.write(str(output))
-        vcftools_log_file.close()
-    else:
-        sys.exit('Error: Log file already exits')
+    vcftools_log_file = open(filename + '.' + function + '.log','w')
+    vcftools_log_file.write(str(output))
+    vcftools_log_file.close()
 
 
 def assign_vcftools_input_arg (filename):
@@ -118,4 +156,5 @@ def assign_vcftools_input_arg (filename):
         elif vcfname_format == 'bgzip':
             return ['--bcf', filename]
         else:
-            sys.exit('Unknown file format')
+            logging.error('Unknown VCF file format')
+            raise Exception('Unknown VCF file format')