Merge branch 'master' of https://github.com/jaredgk/ppp

andrew/andrew_tests.py

@@ -1,56 +1,56 @@
 import unittest, filecmp, sys, os
 import vcftools
-import vcftools_calc
+import vcf_calc
 
 def compare_to_expected(test_output, expected_output):
     return filecmp.cmp(test_output, expected_output)
 
 class vcftools_tests (unittest.TestCase):
-       
+
     def test_vcf_argument_parser (self):
         input_arg = vcftools.assign_vcftools_input_arg('example/locus8.vcf.gz')
         self.assertEqual(input_arg, ['--gzvcf', 'example/locus8.vcf.gz'])
-    
+
     def test_produce_vcftools_log (self):
         vcftools.produce_vcftools_log('Log Test:\n1\n2\n3\n', 'out', '.logTest')
         self.assertTrue(compare_to_expected('out.logTest.log', 'example/locus8.logTest.log'))
         self.addCleanup(os.remove, 'out.logTest.log')
-        
+
     def test_check_vcftools_for_errors (self):
         self.assertTrue(vcftools.check_vcftools_for_errors('Log Test:\n1\n2\n3\nRun Time'))
         with self.assertRaises(SystemExit) as cm:
             vcftools.check_vcftools_for_errors('Log Test:\n1\n2\n3\nError: No Input')
         self.assertEqual(cm.exception.code, 'Error: No Input')
-    
+
     def test_Fst_window (self):
-        vcftools_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'windowed-weir-fst', '--pop-file', 'example/Paniscus.txt', '--pop-file', 'example/Troglodytes.txt', '--out', 'out'])
+        vcf_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'windowed-weir-fst', '--pop-file', 'example/Paniscus.txt', '--pop-file', 'example/Troglodytes.txt', '--out', 'out'])
         self.assertTrue(compare_to_expected('out.windowed.weir.fst', 'example/locus8.windowed.weir.fst'))
         self.addCleanup(os.remove, 'out.windowed.weir.fst')
         self.addCleanup(os.remove, 'out.windowed.weir.fst.log')
 
-    def test_tajimasD (self): 
-        vcftools_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'TajimaD', '--out', 'out'])
+    def test_tajimasD (self):
+        vcf_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'TajimaD', '--out', 'out'])
         self.assertTrue(compare_to_expected('out.Tajima.D', 'example/locus8.Tajima.D'))
         self.addCleanup(os.remove, 'out.Tajima.D')
         self.addCleanup(os.remove, 'out.Tajima.D.log')
-        
-    def test_window_pi (self): 
-        vcftools_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'pi', '--out', 'out'])
+
+    def test_window_pi (self):
+        vcf_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'pi', '--out', 'out'])
         self.assertTrue(compare_to_expected('out.windowed.pi', 'example/locus8.windowed.pi'))
         self.addCleanup(os.remove, 'out.windowed.pi')
         self.addCleanup(os.remove, 'out.windowed.pi.log')
-        
-    def test_freq (self):    
-        vcftools_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'freq', '--out', 'out'])
+
+    def test_freq (self):
+        vcf_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'freq', '--out', 'out'])
         self.assertTrue(compare_to_expected('out.frq', 'example/locus8.frq'))
         self.addCleanup(os.remove, 'out.frq')
         self.addCleanup(os.remove, 'out.frq.log')
-    
-    def test_het (self):    
-        vcftools_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'het', '--out', 'out'])
+
+    def test_het (self):
+        vcf_calc.run(['example/locus8.vcf.gz', '--calc-statistic', 'het', '--out', 'out'])
         self.assertTrue(compare_to_expected('out.het', 'example/locus8.het'))
         self.addCleanup(os.remove, 'out.het')
         self.addCleanup(os.remove, 'out.het.log')
 
 if __name__ == "__main__":
-    unittest.main()
+    unittest.main()

andrew/vcftools_calc.py → andrew/vcf_calc.py

@@ -7,12 +7,12 @@
 from vcftools import *
 
 # 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,'jared')))
+sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, 'jared')))
 
 #from logging_module import initLogger
 
 def vcf_calc_parser(passed_arguments):
-    '''VCF Argument Parser'''
+    '''VCF Argument Parser - Assigns arguments from command line'''
 
     def parser_confirm_file ():
         '''Custom action to confirm file exists'''
@@ -70,14 +70,14 @@ def run (passed_arguments = []):
     '''
     Statistic calculation using VCFTools.
 
-    Automates the calculation of specific statistics (Fst (site/windowed),
-    Tajima's D, Pi, allele frequency, and heterozygosity) using VCFTools. If no
-    statistic is specified, windowed Fst is used by default.
+    Automates the calculation of site/windowed fixation index (Fst), Tajima's D,
+    nucleotide diversity (Pi), allele frequency, and heterozygosity using
+    VCFTools. If no statistic is specified, windowed Fst is used by default.
 
     Parameters
     ----------
     VCF_Input : str
-        Input VCF filename
+        Specifies the input VCF filename
     --out : str
         Specifies the output filename
     --pop-file : str
@@ -87,14 +87,23 @@ def run (passed_arguments = []):
         Specifies the statistic to calculate. Choices: weir-fst,
         windowed-weir-fst (Default), TajimaD, pi, freq, het
     --statistic-window-size : int
-        Specifies the window size of the statistic
+        Specifies the window size for window-based statistics
     --statistic-window-step : int
-        Specifies step size between windows
+        Specifies step size between windows for spcific window-based statistics
 
     Returns
     -------
-    file
+    output : file
         Statistic file output
+    log : file
+        Log file output
+
+    Raises
+    ------
+    IOError
+        Input VCF file does not exist
+    IOError
+        Output file already exists
 
 
     '''

andrew/vcf_filter.py

@@ -0,0 +1,179 @@
+import os
+import sys
+import subprocess
+import argparse
+import logging
+
+from vcftools import *
+
+# 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, 'jared')))
+
+#from logging_module import initLogger
+
+def vcf_filter_parser(passed_arguments):
+    '''VCF Argument Parser - Assigns arguments from command line'''
+
+    def parser_confirm_file ():
+        '''Custom action to confirm 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
+                setattr(args, self.dest, value)
+        return customAction
+
+    def parser_confirm_no_file ():
+        '''Custom action to confirm file does not exist'''
+        class customAction(argparse.Action):
+            def __call__(self, parser, args, value, option_string=None):
+                if os.path.isfile(value):
+                    raise IOError # File found
+                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
+
+    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())
+
+    # 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())
+
+    ### Site Filters
+
+    # Chromosome Filters
+    vcf_parser.add_argument('--filter-chr', help = 'Specifies the chromosome(s) to include', nargs = '+', type = str)
+    vcf_parser.add_argument('--filter-not-chr', help = 'Specifies the chromosome(s) to exclude', nargs = '+', type = str)
+
+    # Position Filters
+    vcf_parser.add_argument('--filter-from-bp', help = 'Specifies the lower bound of sites to include (May only be used with a single chromosome)', type = int)
+    vcf_parser.add_argument('--filter-to-bp', help = 'Specifies the upper bound of sites to include (May only be used with a single chromosome)', type = int)
+
+
+    if passed_arguments:
+        return vcf_parser.parse_args(passed_arguments)
+    else:
+        return vcf_parser.parse_args()
+
+def run (passed_arguments = []):
+    '''
+    Filter VCF files using VCFTools.
+
+    Automates various filters to VCF files using VCFtools.
+
+    Parameters
+    ----------
+    VCF_Input : str
+        Specifies the input VCF filename
+    --out : str
+        Specifies the output filename
+
+    Returns
+    -------
+    output : file
+        Filtered VCF file output
+    log : file
+        Log file output
+
+    Raises
+    ------
+    IOError
+        Input VCF file does not exist
+    IOError
+        Output file already exists
+
+
+    '''
+
+    # Grab VCF arguments from command line
+    vcf_args = vcf_filter_parser(passed_arguments)
+
+    print vcf_args
+
+    '''
+    # Argument container for vcftools
+    vcftools_call_args = ['--out', vcf_args.out]
+
+    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')
+
+        # 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]]
+        vcftools_window_args = ['--fst-window-size', vcf_args.statistic_window_size, '--fst-window-step', vcf_args.statistic_window_step]
+
+        # Assigns all the vcftools arguments for calculating windowed fst
+        vcftools_call_args.extend(vcftools_pop_args + vcftools_window_args)
+
+        # Assigns the suffix for the vcftools log file
+        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')
+
+        # 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'
+
+    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'
+
+    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'
+
+    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'
+
+    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'
+
+    # Assigns the file argument for vcftools
+    vcfname_arg = assign_vcftools_input_arg(vcf_args.vcfname)
+
+    # vcftools subprocess call
+    vcftools_call = subprocess.Popen(['vcftools'] + vcfname_arg + list(map(str, vcftools_call_args)), stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    vcftools_out, vcftools_err = vcftools_call.communicate()
+
+    # 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, vcftools_log_suffix)
+    '''
+
+if __name__ == "__main__":
+    #initLogger()
+    run()

andrew/vcf_phase.py

@@ -4,7 +4,7 @@
 import argparse
 import glob
 
-sys.path.insert(0, os.path.abspath(os.path.join(os.pardir,'jared')))
+sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, 'jared')))
 
 import vcf_reader_func
 
@@ -41,7 +41,7 @@ def __call__(self, parser, args, value, option_string=None):
 
     # Other basic arguments. Expand as needed
     phase_parser.add_argument('--out', help = 'Defines the output filename', default = 'out', action = parser_confirm_no_file())
-    phase_parser.add_argument('--estimate-file', help = 'Defines the estimated genotype frequency filename. For used if using beagle', default = 'estimated_gt', action = parser_confirm_no_file())
+    phase_parser.add_argument('--estimate-file', help = 'Defines the estimated genotype frequency filename. Required for the beagle algorithm', default = 'estimated_gt', action = parser_confirm_no_file())
 
     if passed_arguments:
         return phase_parser.parse_args(passed_arguments)
@@ -68,7 +68,39 @@ def assign_vcf_extension (filename):
         sys.exit('Unknown file format')
 
 def run (passed_arguments = []):
-    ''' Wrapper code for Phasing. Commands are assigned using argparse.'''
+    '''
+        Phaser for VCF files.
+
+        Automates the phasing process for a specified VCF file. The function
+        allows users to select between multiple phasing algorithms: beagle
+        (default) and shapit.
+
+        Parameters
+        ----------
+        VCF_Input : str
+            Specifies the input VCF filename
+        --phase-algorithm : str
+            Specifies the algorithm to be used. Choices: beagle (default) and
+            shapit
+        --out : str
+            Specifies the output filename
+        --estimate-file : str
+            Defines the estimated genotype frequency filename. Required for the
+            beagle algorithm
+
+        Returns
+        -------
+        output : file
+            Phased VCF file
+
+        Raises
+        ------
+        IOError
+            Input VCF file does not exist
+        IOError
+            Output file already exists
+
+    '''
 
     # Grab VCF arguments from command line
     phase_args = phase_argument_parser(passed_arguments)