external config

config.yml

+ltr:
+  path: '/home/radovan/lexa/LTR_finder/ltr_finder'
+  prosite_path: &prosite_path /home/radovan/lexa/prosite
+  tRNAdb_path: &tRNAdb_path /home/radovan/lexa/LTR_finder/tRNAdb/Athal-tRNAs.fa
+  args:
+    l: 32
+    S: 3
+    o: 2
+    p: 15
+    g: 80
+    G: 5
+    T: 2
+    w: 2 #don't change, parsing format required!
+    a: *prosite_path
+    s: *tRNAdb_path
+    
+gt:
+  path: 'gt'
+  command: 'sketch'
+  style_path: &style_path /etc/nested/gt.style
+  args: 
+    width: 1400
+    style: *style_path
+
+blastx:
+  db_path: &db_path /home/radovan/lexa/blastDB/data/gydb_proteins.fa
+  args:
+    db: *db_path
+    outfmt: 5 #don't change, parsing format required!
+    num_threads: 3
+    dbsize: 90000
+    word_size: 2
+    evalue: 1
+
+logdir: /tmp/nested/logs

nested/cli/nester.py

@@ -34,9 +34,9 @@ def main(input_fasta, sketch_only, data_folder):
             print('Processing {a}: DONE [{b}]'.format(a=sequence.id[:strlen], b=seq_end_time - seq_start_time)) 
         except KeyboardInterrupt:
             raise
-        except CalledProcessError:
-            number_of_errors += 1
-            print('Processing {}: SUBPROCESS ERROR'.format(sequence.id[:strlen]))
+        #except CalledProcessError:
+        #    number_of_errors += 1
+        #    print('Processing {}: SUBPROCESS ERROR'.format(sequence.id[:strlen]))
         #except:
         #    number_of_errors += 1
         #    print('Processing {}: UNEXPECTED ERROR:'.format(sequence.id[:strlen]), sys.exc_info()[0]) 

nested/config/config.py

-ltr_finder_path = '/home/radovan/lexa/LTR_finder/ltr_finder'
-ltr_finder_prosite_path = '/home/radovan/lexa/prosite'
-ltr_finder_tRNAdb_path = '/home/radovan/lexa/LTR_finder/tRNAdb/Athal-tRNAs.fa'
+import yaml
 
-ltr_finder_args = ['-l', '32']
-ltr_finder_args += ['-S', '3']
-ltr_finder_args += ['-o', '2']
-ltr_finder_args += ['-p', '15']
-ltr_finder_args += ['-g', '80']
-ltr_finder_args += ['-G', '5']
-ltr_finder_args += ['-T', '2']
-ltr_finder_args += ['-w', '2'] #don't change, parsing format required!
-ltr_finder_args += ['-a', ltr_finder_prosite_path]
-ltr_finder_args += ['-s', ltr_finder_tRNAdb_path]
+with open('/etc/nested/config.yml') as stream:
+    config = yaml.load(stream)
 
-gt_sketch_path = 'gt'
-gt_sketch_args = ['sketch']
-gt_sketch_args += ['-width', '1400']
-gt_sketch_args += ['-style', '/home/radovan/git/nested/nested/config/gt.style']
-
-blastx_args = {
+def args_dict_to_list(args):
+    args_list = []
+    for arg in args:
+        args_list += ['-{}'.format(arg), str(args[arg])]
+    return args_list
\ No newline at end of file
-    'db': '/home/radovan/lexa/blastDB/data/gydb_proteins.fa',
-    'outfmt': 5, #don't change, parsing format required!
-    'num_threads': 3,
-    'dbsize': 90000,
-    'word_size': 2,
-    'evalue': 1
-}
-
-logdir = '/tmp/nested/logs'
-#ADD BLAST PARAMETERS

nested/core/domain.py

@@ -5,7 +5,7 @@ from io import StringIO
 from Bio.Blast import NCBIXML
 from Bio.Blast.Applications import NcbiblastxCommandline
 
-from nested.config import config
+from nested.config.config import config
 
 class Domain(object):
 	"""Class representing domain parsed from blastx output.
@@ -46,7 +46,7 @@ def run_blastx(sequence):
         list[Domain]: list of matched domains
     """
     domains = []
-    blastx_cline = NcbiblastxCommandline(**config.blastx_args)
+    blastx_cline = NcbiblastxCommandline(**config['blastx']['args'])
     xml_out, stderr = blastx_cline(stdin=str(sequence))
     blast_records = NCBIXML.parse(StringIO(xml_out))
     try:

nested/core/nester.py

@@ -4,7 +4,7 @@ from nested.utils import intervals
 from nested.core.gene import Gene
 from nested.core.nested_element import NestedElement
 from nested.logging.logger import NesterLogger
-from nested.config import config
+from nested.config.config import config
 
 class Nester(object):
     """Class represent nesting in sequence, recursivelly find best evaluated transposon and crop it until no new transposons found
@@ -19,7 +19,7 @@ class Nester(object):
         self.sequence = sequence.seq
         self.nested_element = None
         self._iteration = 0
-        self._logger = NesterLogger('{}/{}'.format(config.logdir, self.seqid))
+        self._logger = NesterLogger('{}/{}'.format(config['logdir'], self.seqid))
         self._find_nesting()                    
 
     def _find_nesting(self):

nested/core/te.py

@@ -7,7 +7,7 @@ import re
 from Bio import SeqIO
 from Bio.SeqRecord import SeqRecord
 
-from nested.config import config
+from nested.config.config import config, args_dict_to_list
 
 class TE(object):
     """Class representing TE. Every location is in format [from, to].
@@ -65,7 +65,7 @@ def run_ltr_finder(seqid, sequence):
                     'fasta')
 
     #call LTR finder and feed stdin to it
-    process = subprocess.Popen([config.ltr_finder_path] + config.ltr_finder_args + ['/tmp/nested/ltr/{}.fa'.format(seqid)], 
+    process = subprocess.Popen([config['ltr']['path']] + args_dict_to_list(config['ltr']['args']) + ['/tmp/nested/ltr/{}.fa'.format(seqid)], 
                                 stdout=subprocess.PIPE, 
                                 stderr=subprocess.PIPE) 
 

nested/output/sketcher.py

@@ -9,7 +9,7 @@ from Bio.Seq import Seq
 from Bio.SeqRecord import SeqRecord
 from Bio.SeqFeature import SeqFeature, FeatureLocation
 
-from nested.config import config
+from nested.config.config import config, args_dict_to_list
 from nested.utils import intervals
 
 DEFAULT_DIRPATH = 'data'
@@ -30,7 +30,7 @@ class Sketcher(object):
 
         self._create_dirs(nested_element.id, dirpath)
 
-        #find closes parent
+        #find closest parent
         nl = nested_element.nested_list
         parents = [-1] * len(nl)
         for i in range(len(parents) - 1):
@@ -116,8 +116,8 @@ class Sketcher(object):
         #self._create_dirs(id, dirpath)
         null = open(os.devnull, 'w')
         process = subprocess.check_output(
-            [config.gt_sketch_path] + 
-            config.gt_sketch_args + 
+            [config['gt']['path'], config['gt']['command']] + 
+            args_dict_to_list(config['gt']['args']) + 
             ['{}/{}/{}.png'.format(dirpath, id, id)] + 
             ['{}/{}/{}.gff'.format(dirpath, id, id)], stderr=null)
         

setup.sh

+#!/bin/sh
+
+#TODO ADD CONFIG PATH
+
+python3 setup.py install
+
+#remove unnecessary dirs
+rm -rf build dist nested.egg-info
+
+#move config file
+mkdir -p /etc/nested
+cp config.yml /etc/nested/.
+cp nested/config/gt.style /etc/nested/.
\ No newline at end of file