Update functions to work with args. Changed tests to args

pgpipe/bed_utilities.py

@@ -237,12 +237,11 @@
 import csv
 import numpy as np
 
-# Import basic bedtools functions
 from pgpipe.bedtools_wrapper import merge_bed_files, standard_bedtools_call, log_bedtools_reference
-
 from pgpipe.logging_module import initLogger, logArgs
+from pgpipe.misc import argprase_kwargs
 
-def bed_argument_parser(passed_arguments):
+def bed_argument_parser(passed_arguments = []):
     '''Phase Argument Parser - Assigns arguments for vcftools from command line.
     Depending on the argument in question, a default value may be specified'''
 
@@ -314,12 +313,10 @@ def metavar_list (var_list):
     bed_parser.add_argument('--out', help = 'Defines the output filename', default = 'out.bed')
     bed_parser.add_argument('--overwrite', help = "Defines that previous output files should be overwritten", action = 'store_true')
 
-
-
     if passed_arguments:
-        return bed_parser.parse_args(passed_arguments)
+        return vars(bed_parser.parse_args(passed_arguments))
     else:
-        return bed_parser.parse_args()
+        return vars(bed_parser.parse_args())
 
 def random_bed_sampler (bed_filename, out_filename, sample_size, with_replacements = False):
     '''
@@ -364,7 +361,7 @@ def random_bed_sampler (bed_filename, out_filename, sample_size, with_replacemen
         # Write the output
         out_writer.writerows(random_samples)
 
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
         Utilites for BED-formatted files
 
@@ -425,8 +422,12 @@ def run (passed_arguments = []):
 
     '''
 
-    # Grab BED arguments from command line
-    bed_args = bed_argument_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, bed_argument_parser)
+
+    # Assign arguments
+    bed_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(bed_args, 'bed_utilities')
@@ -597,4 +598,4 @@ def run (passed_arguments = []):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**bed_argument_parser())

pgpipe/misc.py

@@ -36,13 +36,17 @@ def confirm_executable (executable):
     return None
 
 
-def argprase_kwargs (kwarg_dict, argparse_function):
+def argprase_kwargs (kwarg_dict, argparse_function, bad = False):
 
     # Convert the kwargs
     kwargs_list = []
     for arg, value in kwarg_dict.items():
         if isinstance(value, (bool)):
             kwargs_list.append('--%s' % arg.replace('_','-'))
+        elif isinstance(value, (list)) and isinstance(value[0], (list)):
+            for sub_list in value: kwargs_list.extend(['--%s' % arg.replace('_','-')] + [str(sub_value) for sub_value in sub_list])
+        elif isinstance(value, (list)):
+            kwargs_list.extend(['--%s' % arg.replace('_','-')] + [str(sub_value) for sub_value in value])
         else:
             kwargs_list.extend(['--%s' % arg.replace('_','-'), str(value)])
     return argparse_function(kwargs_list)

pgpipe/model_creator.py

@@ -209,8 +209,9 @@
 
 from pgpipe.model import Model, ModelFile, write_model_file, read_model_file
 from pgpipe.logging_module import initLogger, logArgs
+from pgpipe.misc import argprase_kwargs
 
-def model_creator_parser (passed_arguments):
+def model_creator_parser (passed_arguments = []):
     '''VCF Argument Parser - Assigns arguments from command line'''
 
     def parser_confirm_file ():
@@ -351,9 +352,9 @@ def metavar_list (var_list):
     model_parser.add_argument('--overwrite', help = "Specifies if previous output files should be overwritten", action = 'store_true')
 
     if passed_arguments:
-        return model_parser.parse_args(passed_arguments)
+        return vars(model_parser.parse_args(passed_arguments))
     else:
-        return model_parser.parse_args()
+        return vars(model_parser.parse_args())
 
 def incompatible_duplicates_check (term, *arguments_to_test):
 
@@ -472,7 +473,7 @@ def file_dict_argument_to_str (term, file_argument):
     # Return data as str
     return data_str
 
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
 
     Parameters
@@ -529,8 +530,12 @@ def run (passed_arguments = []):
         No individuals assigned to population
     '''
 
-    # Grab VCF arguments from command line
-    creator_args = model_creator_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, model_creator_parser)
+
+    # Assign arguments
+    creator_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(creator_args, 'model_creator')
@@ -751,4 +756,4 @@ def run (passed_arguments = []):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**model_creator_parser())

pgpipe/stat_sampler.py

@@ -95,22 +95,13 @@
 import pandas as pd
 from collections import defaultdict
 
-# Insert Jared's directory path, required for calling Jared's functions. Change when directory structure changes.
-#sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, 'pppipe')))
-
-# Import log initializer
 from pgpipe.logging_module import initLogger, logArgs
-
-# Import basic vcftools functions
 from pgpipe.vcftools import *
-
-# Import basic vcf
 from pgpipe.bcftools import check_for_index, create_index
-
-# Model file related functions
 from pgpipe.model import read_model_file
+from pgpipe.misc import argprase_kwargs
 
-def sampler_parser(passed_arguments):
+def sampler_parser(passed_arguments = []):
     '''
     Stat Sampler Phase Argument Parser
 
@@ -172,9 +163,9 @@ def metavar_list (var_list):
 
 
     if passed_arguments:
-        return sampler_parser.parse_args(passed_arguments)
+        return vars(sampler_parser.parse_args(passed_arguments))
     else:
-        return sampler_parser.parse_args()
+        return vars(sampler_parser.parse_args())
 
 def random_vcftools_sampler (stat_file_data, sample_size, with_replacements = False):
     '''
@@ -331,7 +322,7 @@ def assign_statistic_column (sample_headers, statistic):
     # Return the converted statistic
     return converted_statistic
 
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
     Statistics sampler.
 
@@ -379,8 +370,12 @@ def run (passed_arguments = []):
         Window size argument not defined (if necessary)
     '''
 
-    # Get arguments from command line
-    sampler_args = sampler_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, sampler_parser)
+
+    # Assign arguments
+    sampler_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(sampler_args, func_name = 'stat_sampler')
@@ -468,4 +463,4 @@ def run (passed_arguments = []):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**sampler_parser())

pgpipe/vcf_calc.py

@@ -154,16 +154,12 @@
 import shutil
 import logging
 
-# Import basic vcftools functions
 from pgpipe.vcftools import *
-
-# Model file related functions
 from pgpipe.model import read_model_file
-
-# Import logging functions
 from pgpipe.logging_module import initLogger, logArgs
+from pgpipe.misc import argprase_kwargs
 
-def vcf_calc_parser(passed_arguments):
+def vcf_calc_parser(passed_arguments = []):
     '''
     VCF Calc Argument Parser
 
@@ -270,12 +266,11 @@ def metavar_list (var_list):
     vcf_parser.add_argument('--filter-exclude-indv-file', help = 'Defines a file of individuals to exclude', action = parser_confirm_file())
 
     if passed_arguments:
-        return vcf_parser.parse_args(passed_arguments)
+        return vars(vcf_parser.parse_args(passed_arguments))
     else:
-        return vcf_parser.parse_args()
+        return vars(vcf_parser.parse_args())
 
-
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
     Statistic calculation using VCFTools.
 
@@ -343,8 +338,12 @@ def calc_exception (selected_model, exc_type, exc_value, exc_traceback):
         # Report the original error    
         sys.__excepthook__(exc_type, exc_value, exc_traceback)
 
-    # Grab VCF arguments from command line
-    vcf_args = vcf_calc_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, vcf_calc_parser)
+
+    # Assign arguments
+    vcf_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(vcf_args, func_name = 'vcf_calc')
@@ -712,4 +711,4 @@ def return_filename (filepath):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**vcf_calc_parser())

pgpipe/vcf_filter.py

@@ -303,16 +303,6 @@ def metavar_list (var_list):
     snps_filters.add_argument('--filter-include-snps', help = 'Include variants if they contain a SNP', action = 'store_true')
     snps_filters.add_argument('--filter-exclude-snps', help = 'Exclude variants if they contain a SNP', action = 'store_true')
 
-    def str_to_bool(value):
-        if value.lower() in {'false', 'f', '0', 'no', 'n'}:
-            return False
-        elif value.lower() in {'true', 't', '1', 'yes', 'y'}:
-            return True
-        raise ValueError(f'{value} is not a valid boolean value')
-
-    vcf_parser.add_argument('--foo', type=str_to_bool, nargs='?', const=True, default=False)
-
-
     # Position filters
     vcf_parser.add_argument('--filter-include-pos', help = 'Defines comma seperated positions (i.e. CHROM:START-END) to include. START and END are optional. May be used multiple times', nargs = '+', type = str, action = parser_add_to_list())
     vcf_parser.add_argument('--filter-exclude-pos', help = 'Defines comma seperated positions (i.e. CHROM:START-END) to exclude. START and END are optional. May be used multiple times', nargs = '+', type = str, action = parser_add_to_list())

pgpipe/vcf_format_conversions.py

@@ -66,17 +66,15 @@
 import glob
 import logging
 
-#sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, 'pppipe')))
-
-# Import PPP modules and scripts
 from pgpipe.eigenstrat_wrapper import *
 from pgpipe.bcftools import convert_vcf, log_bcftools_reference
 from pgpipe.plink import convert_ped, convert_bed, convert_vcf_to_plink, log_plink_reference
 from pgpipe.vcf_reader_func import checkFormat
 from pgpipe.logging_module import initLogger, logArgs
+from pgpipe.misc import argprase_kwargs
 
 
-def convert_argument_parser(passed_arguments):
+def convert_argument_parser(passed_arguments = []):
     '''
     Convert Argument Parser
 
@@ -148,9 +146,9 @@ def metavar_list (var_list):
     convert_parser.add_argument('--threads', help = "Set the number of threads. Only supported with ped", type = int, default = 1)
 
     if passed_arguments:
-        return convert_parser.parse_args(passed_arguments)
+        return vars(convert_parser.parse_args(passed_arguments))
     else:
-        return convert_parser.parse_args()
+        return vars(convert_parser.parse_args())
 
 def check_if_conversion (input_format, out_format):
     '''
@@ -196,7 +194,7 @@ def check_conversion_support (out_format, supported_formats):
     if out_format not in supported_formats:
         raise Exception('Input cannot be converted into the format specified')
 
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
     File conversion suite
 
@@ -236,8 +234,12 @@ def run (passed_arguments = []):
         Conversion to same format given
     '''
 
-    # Grab plink arguments from command line
-    convert_args = convert_argument_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, convert_argument_parser)
+
+    # Assign arguments
+    convert_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(convert_args, func_name = 'convert')
@@ -384,4 +386,4 @@ def run (passed_arguments = []):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**convert_argument_parser())

pgpipe/vcf_phase.py

@@ -157,8 +157,9 @@
 from pgpipe.shapeit import call_shapeit, remove_shapeit_intermediate_files, check_for_shapeit_intermediate_files
 from pgpipe.bcftools import get_unique_chrs, get_samples, chr_subset_file, concatenate, check_for_index, create_index
 from pgpipe.plink import convert_haps_to_vcf
+from pgpipe.misc import argprase_kwargs
 
-def phase_argument_parser(passed_arguments):
+def phase_argument_parser(passed_arguments = []):
     '''
     VCF Phase Argument Parser
 
@@ -287,9 +288,9 @@ def metavar_list (var_list):
     phase_parser.add_argument('--beagle-nsteps', help = 'Number of consecutive --beagle-steps used for identifying long IBS segments (beagle)', type = int)
 
     if passed_arguments:
-        return phase_parser.parse_args(passed_arguments)
+        return vars(phase_parser.parse_args(passed_arguments))
     else:
-        return phase_parser.parse_args()
+        return vars(phase_parser.parse_args())
 
 def log_to_stdout (log_filename):
     '''
@@ -437,7 +438,7 @@ def assign_filename_prefix (output_filename, output_format):
     # Return the updated prefix
     return updated_prefix
 
-def run (passed_arguments = []):
+def run (**kwargs):
     '''
     Phaser for VCF files.
 
@@ -516,8 +517,12 @@ def run (passed_arguments = []):
         Incompatible arguments
     '''
 
-    # Grab VCF arguments from command line
-    phase_args = phase_argument_parser(passed_arguments)
+    # Update kwargs with defaults
+    if __name__ != "__main__":
+        kwargs = argprase_kwargs(kwargs, phase_argument_parser)
+
+    # Assign arguments
+    phase_args = argparse.Namespace(**kwargs)
 
     # Adds the arguments (i.e. parameters) to the log file
     logArgs(phase_args, func_name = 'vcf_phase')
@@ -555,8 +560,8 @@ def run (passed_arguments = []):
     filter_indv_filename = ''
 
     # Check if the has specified the output filename, without a prefix
-    if phase_args.out and '--out-prefix' not in passed_arguments and '--out-prefix' not in sys.argv:
-
+    if phase_args.out:
+        
         # Assign a prefix based on the output filename
         phase_args.out_prefix = assign_filename_prefix(phase_args.out, phase_args.out_format)
 
@@ -997,4 +1002,4 @@ def run (passed_arguments = []):
 
 if __name__ == "__main__":
     initLogger()
-    run()
+    run(**phase_argument_parser())