Restructure project to not mix packages, scripts and executables (Fixes #21)

bin/psot

 #!/usr/bin/env python3
-import argparse
 import os
-from config import load_config
-import copy
-import shutil
-from nextflow import setup_execution_directory, execute_analysis
-import tempfile
+import sys
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
 
-def main():
-    parser = argparse.ArgumentParser(description='Make bioinformatic observations on aminoacid sequences')
-    parser.set_defaults(func=help)
-
-    subparsers = parser.add_subparsers()
-    info_parser = subparsers.add_parser('info')
-    info_parser.add_argument('--listanalyses', '-l', action='store_true', help='Show available analysis steps')
-    info_parser.set_defaults(func=info)
-
-    analyze_parser = subparsers.add_parser('analyze')
-
-    analyze_parser.add_argument('--fasta', '-f', required=True, help='A fasta file with aminoacid sequences')
-    analyze_parser.add_argument('--output', '-o', required=True, help='The output directory for the json documents')
-    analyze_parser.add_argument('--profile', '-p', default='fast', help='The profile to use')
-    analyze_parser.add_argument('--live', '-l', action='store_true', help='Report results as they are computed, not only at the end of the computation. The live results will be available in the $output/live.')
-    analyze_parser.add_argument('--config', '-c', help='The config to use')
-    analyze_parser.add_argument('--debug', '-d', action='store_true', help='Debug mode, computation directory will not be removed after computation')
-    analyze_parser.add_argument('--execution_dir', '-e', help='Use the specified execution directory and do not delete it after the computation')
-    analyze_parser.set_defaults(func=analyze)
-
-    args = parser.parse_args()
-    config = load_config()
-    args.parser = parser
-    args.func(args, config)
-
-def help(args, config):
-    args.parser.print_help()
-
-def info(args, config):
-    show_analyses(config)
-
-def analyze(args, config):
-    execution = generate_execution(config, args)
-    if args.debug:
-        print(execution)
-    setup_execution_directory(execution)
-    execute_analysis(execution)
-    cleanup(execution)
-
-def cleanup(execution):
-    if not execution['debug']:
-        shutil.rmtree(execution['directory'])
-
-def generate_execution(config, args):
-    execution = {}
-    execution['debug'] = args.debug
-    execution['mode'] = 'live' if args.live else 'complete'
-    execution['bin_path'] = config['app']['bin_path']
-    execution['fasta'] = os.path.abspath(args.fasta)
-    execution['output'] = os.path.abspath(args.output)
-    if args.execution_dir:
-        execution['directory'] = os.path.abspath(args.execution_dir)
-    else:
-        execution['directory'] = tempfile.mkdtemp()
-    execution['modules'] = generate_execution_modules_for_profile(config, args.profile)
-    return execution
-    
-def generate_execution_modules_for_profile(config, profile):
-    # find profile by name
-    p = [x for x in config['profiles'] if x['name'] == profile][0]
-    modules = copy.deepcopy(p['modules'])
-    # generate unique ids for each module
-    for module in modules:
-        module['id'] = module['name']
-        module['analysis']['params'] = generate_params_string(module['analysis']['parameters'])
-        module['converter']['params'] = generate_params_string(module['converter']['parameters'])
-    return modules
-
-def generate_params_string(options):
-    params = ''
-    if options:
-        l = []
-        for k in options:
-            if options[k]:
-                l.append('--' + k + " '" + options[k] + "'")
-            else:
-                l.append('--' + k)
-        params = ' '.join(l)
-    return params
-
-
-def show_analyses(config):
-    print('Profiles:')
-    for profile in config['profiles']:
-        print('   {0:<20} - {1}'.format(profile['name'], profile['info']))
-        if 'modules' in profile:
-            for module in profile['modules']:
-                print('      {0:<20}'.format(module['name']))
-
-    print()
-    print('Available modules for custom profile:')
-    for key in config['modules']:
-        module = config['modules'][key]
-        print('   {0:<20} - {1}'.format(module['name'], module['info']))
-
-if __name__ == "__main__":
-    main()
+from psot.main import main
+main()

psot/main.py

+#!/usr/bin/env python3
+import argparse
+import os
+from psot.config import load_config
+import copy
+import shutil
+from psot.nextflow import setup_execution_directory, execute_analysis
+import tempfile
+
+def main():
+    parser = argparse.ArgumentParser(description='Make bioinformatic observations on aminoacid sequences')
+    parser.set_defaults(func=help)
+
+    subparsers = parser.add_subparsers()
+    info_parser = subparsers.add_parser('info')
+    info_parser.add_argument('--listanalyses', '-l', action='store_true', help='Show available analysis steps')
+    info_parser.set_defaults(func=info)
+
+    analyze_parser = subparsers.add_parser('analyze')
+
+    analyze_parser.add_argument('--fasta', '-f', required=True, help='A fasta file with aminoacid sequences')
+    analyze_parser.add_argument('--output', '-o', required=True, help='The output directory for the json documents')
+    analyze_parser.add_argument('--profile', '-p', default='fast', help='The profile to use')
+    analyze_parser.add_argument('--live', '-l', action='store_true', help='Report results as they are computed, not only at the end of the computation. The live results will be available in the $output/live.')
+    analyze_parser.add_argument('--config', '-c', help='The config to use')
+    analyze_parser.add_argument('--debug', '-d', action='store_true', help='Debug mode, computation directory will not be removed after computation')
+    analyze_parser.add_argument('--execution_dir', '-e', help='Use the specified execution directory and do not delete it after the computation')
+    analyze_parser.set_defaults(func=analyze)
+
+    args = parser.parse_args()
+    config = load_config()
+    args.parser = parser
+    args.func(args, config)
+
+def help(args, config):
+    args.parser.print_help()
+
+def info(args, config):
+    show_analyses(config)
+
+def analyze(args, config):
+    execution = generate_execution(config, args)
+    if args.debug:
+        print(execution)
+    setup_execution_directory(execution)
+    execute_analysis(execution)
+    cleanup(execution)
+
+def cleanup(execution):
+    if not execution['debug']:
+        shutil.rmtree(execution['directory'])
+
+def generate_execution(config, args):
+    execution = {}
+    execution['debug'] = args.debug
+    execution['mode'] = 'live' if args.live else 'complete'
+    execution['bin_path'] = config['app']['bin_path']
+    execution['script_path'] = os.path.abspath(os.path.join(os.path.dirname(config['app']['bin_path']), 'scripts'))
+    execution['psot_path'] = os.path.abspath(os.path.join(os.path.dirname(config['app']['bin_path']), 'psot'))
+    execution['fasta'] = os.path.abspath(args.fasta)
+    execution['output'] = os.path.abspath(args.output)
+    if args.execution_dir:
+        execution['directory'] = os.path.abspath(args.execution_dir)
+    else:
+        execution['directory'] = tempfile.mkdtemp()
+    execution['modules'] = generate_execution_modules_for_profile(config, args.profile)
+    return execution
+    
+def generate_execution_modules_for_profile(config, profile):
+    # find profile by name
+    p = [x for x in config['profiles'] if x['name'] == profile][0]
+    modules = copy.deepcopy(p['modules'])
+    # generate unique ids for each module
+    for module in modules:
+        module['id'] = module['name']
+        module['analysis']['params'] = generate_params_string(module['analysis']['parameters'])
+        module['converter']['params'] = generate_params_string(module['converter']['parameters'])
+    return modules
+
+def generate_params_string(options):
+    params = ''
+    if options:
+        l = []
+        for k in options:
+            if options[k]:
+                l.append('--' + k + " '" + options[k] + "'")
+            else:
+                l.append('--' + k)
+        params = ' '.join(l)
+    return params
+
+
+def show_analyses(config):
+    print('Profiles:')
+    for profile in config['profiles']:
+        print('   {0:<20} - {1}'.format(profile['name'], profile['info']))
+        if 'modules' in profile:
+            for module in profile['modules']:
+                print('      {0:<20}'.format(module['name']))
+
+    print()
+    print('Available modules for custom profile:')
+    for key in config['modules']:
+        module = config['modules'][key]
+        print('   {0:<20} - {1}'.format(module['name'], module['info']))
+
+if __name__ == "__main__":
+    main()

bin/nextflow.py → psot/nextflow.py

@@ -119,8 +119,10 @@ def setup_execution_directory(execution):
     with open(directory + '/main.nf', 'w') as script_file:
         script_file.write(nextflow_script)
 
-    if not os.path.exists(directory+'/bin'):
-        os.symlink(execution['bin_path'], directory + '/bin')
+    if not os.path.exists(directory + '/bin'):
+        os.symlink(execution['script_path'], directory + '/bin')
+    if not os.path.exists(directory + '/psot'):
+        os.symlink(execution['psot_path'], directory + '/psot')
 
 def execute_analysis(execution):
     old_cwd = os.getcwd()

bin/run_blastp.py → scripts/run_blastp.py

 #!/usr/bin/env python3
+import os
+import sys
+# add psot libraries to module search path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
 import argparse
-import config
+from psot import config
 from os import system
 
 blastp_tool = config.load_config()['tools'].get('blastp', 'blastp')
@@ -19,7 +24,7 @@ if args.use_accession:
     format += ' saccver'
 else:
     format += ' sseqid'
-format += ' pident qstart qend sstart send evalue bitscore'
+format += ' pident qstart qend sstart send slen evalue bitscore'
 
 if args.alignment:
     format += ' btop'

bin/run_ghostx.py → scripts/run_ghostx.py

 #!/usr/bin/env python3
+import os
+import sys
+# add psot libraries to module search path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
 import argparse
-import config
 import re
 from os import system,makedirs
+from psot import config
 import subprocess
 import json
 

bin/run_hmmer.py → scripts/run_hmmer.py

 #!/usr/bin/env python3
+import os
+import sys
+# add psot libraries to module search path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
 import argparse
-import config
+from psot import config
 from os import system, makedirs
 
 hmmscan_tool = config.load_config()['tools'].get('hmmscan', 'hmmscan')

bin/run_identity.py → scripts/run_identity.py

 #!/usr/bin/env python3
 import argparse
-import config
-import re
-from os import system,makedirs
+from os import system
 import subprocess
-import json
 
 parser = argparse.ArgumentParser(description='Script that returns the fasta as it is inserted. Can be used for modules that have no actual analysis.')
 parser.add_argument('--fasta', '-f', required=True, help='A fasta file with aminoacid sequences')

bin/run_signalp.py → scripts/run_signalp.py

 #!/usr/bin/env python3
+import os
+import sys
+# add psot libraries to module search path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
 import argparse
-import config
+from psot import config
 from os import system
 
 signalp_tool = config.load_config()['tools'].get('signalp', 'signalp')