add exclusion of steps for run

Author: bbremer

ggsr

@@ -42,40 +42,46 @@ if sys.argv[1] == '--help':
 with open(sys.argv[1]) as f:
     config_map = yaml.safe_load(f)
 
-sys.argv = [sys.argv[0], config_map['seq_fn'], config_map['pdb_fn'],
-            config_map['out_prefix']]
-run_args = [str(c)+' ' for c in sys.argv]
-run_args[0] = 'step1.py '
-print('Running:', ''.join(run_args))
-# exec(open(dirname + '/step1.py').read())
-print('Step1 done\n')
+excluded_steps = config_map['excluded_steps']
 
-sys.argv = [sys.argv[0],
-            config_map['num_bl'], config_map['minBL'], config_map['maxBL'],
-            config_map['start_oh_seq'], config_map['start_oh_pos'],
-            config_map['end_oh_seq'], config_map['end_oh_pos'],
-            config_map['out_prefix']+'_AA.fasta',
-            config_map['out_prefix']+'_contacts.json',
-            config_map['libraries_fn']]
-run_args = [str(c)+' ' for c in sys.argv]
-run_args[0] = 'step2.py '
-print('Running:', ''.join(run_args))
-exec(open(dirname + '/step2.py').read())
-print('Step2 done\n')
+if 1 not in excluded_steps:
+    sys.argv = [sys.argv[0], config_map['seq_fn'], config_map['pdb_fn'],
+                config_map['out_prefix']]
+    run_args = [str(c)+' ' for c in sys.argv]
+    run_args[0] = 'step1.py '
+    print('Running:', ''.join(run_args))
+    exec(open(dirname + '/step1.py').read())
+    print('Step1 done\n')
 
-sys.argv = [sys.argv[0], config_map['libraries_fn'],
-            config_map['gg_prob_threshold'], config_map['chosen_lib_fn']]
-run_args = [str(c)+' ' for c in sys.argv]
-run_args[0] = 'step3.py '
-print('Running:', ''.join(run_args))
-exec(open(dirname + '/step3.py').read())
-print('Step3 done\n')
+if 2 not in excluded_steps:
+    sys.argv = [sys.argv[0],
+                config_map['num_bl'], config_map['minBL'], config_map['maxBL'],
+                config_map['start_oh_seq'], config_map['start_oh_pos'],
+                config_map['end_oh_seq'], config_map['end_oh_pos'],
+                config_map['out_prefix']+'_AA.fasta',
+                config_map['out_prefix']+'_contacts.json',
+                config_map['libraries_fn']]
+    run_args = [str(c)+' ' for c in sys.argv]
+    run_args[0] = 'step2.py '
+    print('Running:', ''.join(run_args))
+    exec(open(dirname + '/step2.py').read())
+    print('Step2 done\n')
 
-sys.argv = [sys.argv[0], config_map['out_prefix']+'_AA.fasta',
-            config_map['out_prefix']+'_CDN.fasta', config_map['chosen_lib_fn'],
-            config_map['frags_order_fn']]
-run_args = [str(c)+' ' for c in sys.argv]
-run_args[0] = 'step4.py '
-print('Running:', ''.join(run_args))
-exec(open(dirname + '/step4.py').read())
-print('Step4 done')
+if 3 not in excluded_steps:
+    sys.argv = [sys.argv[0], config_map['libraries_fn'],
+                config_map['gg_prob_threshold'], config_map['chosen_lib_fn']]
+    run_args = [str(c)+' ' for c in sys.argv]
+    run_args[0] = 'step3.py '
+    print('Running:', ''.join(run_args))
+    exec(open(dirname + '/step3.py').read())
+    print('Step3 done\n')
+
+if 4 not in excluded_steps:
+    sys.argv = [sys.argv[0], config_map['out_prefix']+'_AA.fasta',
+                config_map['out_prefix']+'_CDN.fasta',
+                config_map['chosen_lib_fn'], config_map['frags_order_fn']]
+    run_args = [str(c)+' ' for c in sys.argv]
+    run_args[0] = 'step4.py '
+    print('Running:', ''.join(run_args))
+    exec(open(dirname + '/step4.py').read())
+    print('Step4 done')

run_config.yaml

@@ -1,3 +1,7 @@
+# which steps to exclude during run.
+# example: [1, 2] to only run steps 3 and 4
+excluded_steps: []
+
 # names of step1 input files
 # can make pdb_fn "search" to automatically search
 # the PDB database and download a PDB structure

tests/bgl3_sample/run_config.yaml

@@ -1,9 +1,13 @@
+# which steps to exclude during run.
+# example: [1, 2] to only run steps 3 and 4
+excluded_steps: [1, 2]
+
 # names of step1 input files
 # can make pdb_fn "search" to automatically search
 # the PDB database and download a PDB structure
 # close to the first parent sequence
 seq_fn: bgl3_sequences.fasta
-pdb_fn: 1GNX.pdb
+pdb_fn: search
 
 # run parameters
 num_bl: 4  # number of blocks in library
@@ -34,7 +38,7 @@ out_prefix: bgl3
 # name of step 2 output file
 libraries_fn: bgl3_libraries.json
 
-# gg_prob threshold for acceptable libraries
+# gg_prob threshold for acceptable libraries (needs to be str for some reason)
 gg_prob_threshold: "0.95"
 
 # name of step 3 output file