Mercurial > repos > tduigou > sculpt_sequences

annotate sculpt_sequences.py @ 1:a0cd867780ec draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
author tduigou
date Thu, 17 Jul 2025 13:24:49 +0000
parents 0c7f75a2338b
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
1 import argparse
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
2 import json
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
3 import sys
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
4 import os
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
5 import re
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
6 import dnacauldron
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
7 import dnachisel
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
8 from Bio import SeqIO
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
9 import proglog
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
10
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
11
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
12 def smart_number(val):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
13 try:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
14 float_val = float(val)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
15 if float_val.is_integer():
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
16 return int(float_val)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
17 else:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
18 return float_val
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
19 except ValueError:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
20 raise ValueError(f"Invalid number: {val}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
21
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
22
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
23 def sculpt_sequances(files_to_sculpt, file_name_mapping, outdir_scul, outdir_unscul, use_file_names_as_id,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
24 avoid_patterns, enforce_gc_content, DnaOptimizationProblemClass,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
25 kmer_size,hairpin_constraints):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
26
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
27 files_to_sculpt = files_to_sculpt.split(',')
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
28
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
29 records_to_sculpt = dnacauldron.biotools.load_records_from_files(
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
30 files=files_to_sculpt,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
31 folder=None,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
32 use_file_names_as_ids=use_file_names_as_id
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
33 )
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
34
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
35 constraint_list = []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
36
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
37 # avoid_patterns pearsing
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
38 for pattern in avoid_patterns:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
39 constraint_list.append(dnachisel.AvoidPattern(pattern))
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
40 print(f"AvoidPattern constraint: {pattern}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
41
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
42 # gc_constraints pearsing
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
43 for constraint in enforce_gc_content:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
44 constraints = [c.strip() for c in constraint.split(' ') if c.strip()]
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
45 for line in constraints:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
46 if not line:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
47 continue
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
48 print(f"GC constraint: {line}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
49
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
50 try:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
51 pairs = [kv.strip() for kv in line.split(',')]
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
52 params = {}
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
53 for pair in pairs:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
54 key, val = pair.split('=')
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
55 key = key.strip()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
56 val = val.strip()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
57 params[key] = smart_number(val)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
58
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
59 constraint_list.append(dnachisel.EnforceGCContent(**params))
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
60
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
61 except Exception as e:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
62 print(f"Skipping invalid gc_constraints: {line} ({e})")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
63
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
64 # hairpin_constraints pearsing
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
65 for constraint in hairpin_constraints:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
66 constraints = [c.strip() for c in constraint.split(' ') if c.strip()]
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
67 for line in constraints:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
68 if not line:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
69 continue
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
70 print(f"Hairpin constraint: {line}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
71
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
72 try:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
73 pairs = [kv.strip() for kv in line.split(',')]
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
74 params = {}
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
75 for pair in pairs:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
76 key, val = pair.split('=')
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
77 key = key.strip()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
78 val = val.strip()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
79 params[key] = smart_number(val)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
80
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
81 constraint_list.append(dnachisel.AvoidHairpins(**params))
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
82
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
83 except Exception as e:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
84 print(f"Skipping invalid hairpin_constraints: {line} ({e})")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
85
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
86 # k_size pearsing
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
87 for k_size in kmer_size:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
88 try:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
89 constraint_list.append(dnachisel.UniquifyAllKmers(k=int(k_size)))
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
90 print(f"k-mer size is: {k_size}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
91 except ValueError:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
92 print(f"Skipping invalid k-mer size: {k_size}")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
93
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
94 # refine the real record name dict
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
95 if isinstance(file_name_mapping, str):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
96 file_name_mapping = dict(
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
97 item.split(":") for item in file_name_mapping.split(",")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
98 )
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
99 real_names = {
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
100 os.path.splitext(os.path.basename(k))[0]: v.replace(".gb", "")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
101 for k, v in file_name_mapping.items()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
102 }
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
103
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
104 counter = 0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
105 # not_divisible_seqs = []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
106 for record in records_to_sculpt:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
107
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
108 # Look up the "real name" from the simplified dict (file_name_mapping)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
109 real_name = real_names.get(record.id, record.id)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
110 if real_name is None:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
111 raise ValueError(f"Error: {record.id} not found in the fragment names dictionary!")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
112
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
113 if DnaOptimizationProblemClass == 'DnaOptimizationProblem':
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
114 problem = dnachisel.DnaOptimizationProblem.from_record(record, extra_constraints=constraint_list)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
115 elif DnaOptimizationProblemClass == 'CircularDnaOptimizationProblem':
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
116 problem = dnachisel.CircularDnaOptimizationProblem.from_record(record, extra_constraints=constraint_list)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
117
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
118 if not problem.all_constraints_pass():
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
119 counter += 1
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
120 print(real_name)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
121 len_seq = len(problem.sequence)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
122 print("Length:", len_seq)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
123 target_path = os.path.join(outdir_scul, real_name + ".zip")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
124 problem.optimize_with_report(target=target_path)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
125 print()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
126 print(problem.constraints_text_summary())
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
127 print()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
128 else: # just save the genbank
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
129 genbank_filename = real_name + ".gb"
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
130 with open(os.path.join(outdir_unscul, genbank_filename), "w") as output_handle:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
131 SeqIO.write(record, output_handle, "genbank")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
132 print()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
133 print("Counter:", counter)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
134 # print("Contents of outdir_scul:")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
135 # for f in os.listdir(outdir_scul):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
136 # print(" -", f)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
137
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
138 # Unzip all files in scul
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
139 import zipfile
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
140 for file in os.listdir(outdir_scul):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
141 if file.endswith(".zip"):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
142 zip_path = os.path.join(outdir_scul, file)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
143 extract_dir = os.path.join(outdir_scul, os.path.splitext(file)[0])
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
144 os.makedirs(extract_dir, exist_ok=True)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
145 with zipfile.ZipFile(zip_path, 'r') as zip_ref:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
146 zip_ref.extractall(extract_dir)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
147 # print("Contents of outdir_scul:")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
148 # for f in os.listdir(outdir_scul):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
149 # print(" -", f)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
150 reports = os.listdir(outdir_scul)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
151 # print (f'report is {reports}')
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
152 from shutil import copyfile
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
153 for report in reports:
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
154 if report.endswith(".zip"):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
155 continue
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
156 gb_file = os.path.join(outdir_scul, report, "final_sequence.gb")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
157
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
158 target_filename = report + ".gb"
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
159 target_file = os.path.join(outdir_unscul, target_filename)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
160 print(gb_file, target_file)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
161 copyfile(gb_file, target_file)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
162 # print("Contents of outdir_unscul:")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
163 # for f in os.listdir(outdir_unscul):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
164 # print(" -", f)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
165
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
166 return outdir_scul, outdir_unscul
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
167
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
168
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
169 def parse_command_line_args():
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
170 parser = argparse.ArgumentParser(description="Evaluate manufacturability of DNA sequences.")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
171
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
172 parser.add_argument("--files_to_sculpt", required=True,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
173 help="List of GenBank files (Comma-separated)")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
174 parser.add_argument('--file_name_mapping', type=str,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
175 help='Mapping of Galaxy filenames to original filenames')
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
176 parser.add_argument("--outdir_scul", required=True, help="scul file dir")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
177 parser.add_argument("--outdir_unscul", required=True, help="unscul file dir")
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
178 parser.add_argument("--use_file_names_as_id", type=lambda x: x.lower() == 'true', default=True,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
179 help="Use file names as IDs (True/False)")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
180 parser.add_argument("--avoid_patterns", required=False,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
181 help="List of patterns to avoid (comma-separated, e.g., 'BsaI_site,BsmBI_site')")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
182 parser.add_argument("--gc_constraints", required=False,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
183 help="GC content constraints as 'min;max;window' (space-separated, e.g., '0.3;0.7;100 0.1;0.3;100')")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
184 parser.add_argument("--DnaOptimizationProblemClass", required=False,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
185 help="the class to use for DnaOptimizationProblem")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
186 parser.add_argument("--hairpin_constraints", required=False,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
187 help="Hairpin constraints as 'stem_size;window_size' (space-separated, e.g., '20;200 30;250')")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
188 parser.add_argument("--kmer_size", required = False,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
189 help="K-mer uniqueness size (e.g., '15')")
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
190 parser.add_argument("--json_params", required=False,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
191 help="JSON params for the tool")
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
192 parser.add_argument("--use_json_param", required=True,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
193 help="If use JSON as param source")
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
194
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
195 return parser.parse_args()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
196
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
197
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
198 def extract_constraints_from_args(args):
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
199 """Extract constraints directly from the command-line arguments."""
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
200
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
201 split_pattern = r'(?:__cn__|\s{2,})'
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
202
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
203 # 1. Avoid patterns (split by any whitespace)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
204 avoid_patterns = re.split(split_pattern, args.avoid_patterns.strip()) if args.avoid_patterns.strip() else []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
205
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
206 # 2. Hairpin constraint: one dictionary (print as string later)
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
207 hairpin_constraints = re.split(split_pattern,args.hairpin_constraints.strip()) if args.hairpin_constraints.strip() else []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
208
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
209 # 3. GC constraints: split by 2+ spaces or newlines
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
210 gc_constraints = re.split(split_pattern, args.gc_constraints.strip()) if args.gc_constraints.strip() else []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
211
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
212 # 4. k-mer size: single value or list
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
213 kmer_size = [int(k.strip()) for k in args.kmer_size.strip().split(',') if k.strip()] if args.kmer_size.strip() else []
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
214
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
215 # 5. DnaOptimizationProblemClass (as string)
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
216 DnaOptimizationProblemClass = args.DnaOptimizationProblemClass if args.DnaOptimizationProblemClass else None
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
217
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
218 return avoid_patterns, hairpin_constraints, gc_constraints, kmer_size, DnaOptimizationProblemClass
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
219
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
220
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
221 def load_constraints_from_json(json_path):
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
222 with open(json_path, 'r') as f:
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
223 params = json.load(f)
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
224
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
225 def split_lines(val):
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
226 if isinstance(val, str):
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
227 return [line.strip() for line in val.strip().split('\n') if line.strip()]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
228 return val
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
229
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
230 avoid_patterns = split_lines(params.get("avoid_patterns", ""))
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
231 hairpin_constraints = split_lines(params.get("hairpin_constraints", ""))
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
232 gc_constraints = split_lines(params.get("gc_constraints", ""))
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
233 kmer_size = [int(k.strip()) for k in str(params.get("kmer_size", "")).split(',') if k.strip()]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
234 DnaOptimizationProblemClass = params.get("DnaOptimizationProblemClass", None)
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
235
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
236 return {
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
237 "avoid_patterns": avoid_patterns,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
238 "hairpin_constraints": hairpin_constraints,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
239 "gc_constraints": gc_constraints,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
240 "kmer_size": kmer_size,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
241 "DnaOptimizationProblemClass": DnaOptimizationProblemClass
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
242 }
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
243
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
244 if __name__ == "__main__":
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
245 args = parse_command_line_args()
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
246
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
247 avoid_patterns, hairpin_constraints, gc_constraints, kmer_size, DnaOptimizationProblemClass = extract_constraints_from_args(args)
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
248
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
249 # Check if the flag --use_json_param is present and set to true
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
250 if "--use_json_param" in sys.argv:
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
251 use_json_index = sys.argv.index("--use_json_param") + 1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
252 use_json = sys.argv[use_json_index].lower() == "true"
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
253 else:
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
254 use_json = False
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
255
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
256 # Now only check --json_params if use_json is True
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
257 if use_json:
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
258 if "--json_params" in sys.argv:
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
259 json_index = sys.argv.index("--json_params") + 1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
260 json_file = sys.argv[json_index]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
261 if json_file.lower() != "none":
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
262 json_constraints = load_constraints_from_json(json_file)
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
263 avoid_patterns = json_constraints["avoid_patterns"]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
264 hairpin_constraints = json_constraints["hairpin_constraints"]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
265 gc_constraints = json_constraints["gc_constraints"]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
266 kmer_size = json_constraints["kmer_size"]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
267 DnaOptimizationProblemClass = json_constraints["DnaOptimizationProblemClass"]
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
268
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
269 params = {
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
270 "files_to_sculpt": args.files_to_sculpt,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
271 "file_name_mapping": args.file_name_mapping,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
272 "outdir_unscul": args.outdir_unscul,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
273 "outdir_scul": args.outdir_scul,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
274 "use_file_names_as_id": args.use_file_names_as_id,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
275 "avoid_patterns": avoid_patterns,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
276 "hairpin_constraints": hairpin_constraints,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
277 "gc_constraints": gc_constraints,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
278 "kmer_size": kmer_size,
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
279 "DnaOptimizationProblemClass": DnaOptimizationProblemClass
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
280 }
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
281
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
282
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
283 sculpt_sequances(
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
284 args.files_to_sculpt, args.file_name_mapping,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
285 args.outdir_scul, args.outdir_unscul,
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
286 args.use_file_names_as_id, avoid_patterns,
1
a0cd867780ec planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6b6ce806b5d016b3c7f20318180eff2dbe64395a-dirty
tduigou
parents: 0
diff changeset
287 gc_constraints, DnaOptimizationProblemClass,
0
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
288 kmer_size, hairpin_constraints
0c7f75a2338b planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit fa4c85dd6ad48d404a28e21667f18b628bbdc702-dirty
tduigou
parents:
diff changeset
289 )