Mercurial > repos > tduigou > domestication

annotate domestication.py @ 10:c52967fb95da 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 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
author tduigou
date Tue, 27 May 2025 08:08:21 +0000
parents f78651af72e4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
1 import os
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
2 import argparse
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
3 import pandas
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
4 import zipfile
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
5 import genedom
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
6 import dnacauldron
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
7 import proglog
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
8 import shutil
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
9 #proglog.notebook()
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
10 from Bio.SeqRecord import SeqRecord
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
11 from Bio.Seq import Seq
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
12 from Bio import SeqIO
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
13
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
14 def domestication(files_to_domestication, csv_file,file_name_mapping, use_file_names_as_id,
4
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
15 allow_edits, output_dom, gb_dom):
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
16
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
17 file_to_domestication = files_to_domestication.split(',')
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
18
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
19 records_to_domesticate = dnacauldron.biotools.load_records_from_files(
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
20 files=file_to_domestication,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
21 folder=None,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
22 use_file_names_as_ids=use_file_names_as_id
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
23 )
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
24
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
25 #refine the real record name dict
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
26 if isinstance(file_name_mapping, str):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
27 file_name_mapping = dict(
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
28 item.split(":") for item in file_name_mapping.split(",")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
29 )
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
30 real_names = {
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
31 os.path.splitext(os.path.basename(k))[0]: v.replace(".gb", "")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
32 for k, v in file_name_mapping.items()
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
33 }
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
34 updated_records = []
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
35 for record in records_to_domesticate:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
36 original_id = record.id
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
37 if original_id in real_names:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
38 new_id = real_names[original_id]
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
39 record.id = new_id
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
40 updated_records.append(record)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
41 ########################################################
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
42
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
43 df=pandas.read_csv(csv_file)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
44 EMMA_PLUS = genedom.GoldenGateDomesticator.standard_from_spreadsheet(dataframe=df)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
45 genedom.batch_domestication(
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
46 records=updated_records,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
47 standard=EMMA_PLUS,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
48 allow_edits=allow_edits,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
49 target=output_dom)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
50
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
51 # Check if any names were truncated:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
52 if isinstance(output_dom, str) and output_dom.endswith(".zip"):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
53 with zipfile.ZipFile(output_dom, "r") as zipf:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
54 with zipf.open("order_ids.csv") as f:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
55 order_ids = pandas.read_csv(f)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
56 else:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
57 order_ids = pandas.read_csv(os.path.join(output_dom, "order_ids.csv"))
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
58
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
59 any_truncated = False
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
60 for index, row in order_ids.iterrows():
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
61 if row["sequence"] != row["order_id"]:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
62 any_truncated = True
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
63 print("Changed names:", end=" ")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
64 print(" --> ".join(row))
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
65 if not any_truncated:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
66 print("Part names were not truncated")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
67
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
68 #zip compressing
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
69 if os.path.isdir(output_dom):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
70 zip_path = output_dom.rstrip("/\\") + ".zip"
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
71 shutil.make_archive(output_dom, 'zip', output_dom)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
72 shutil.move(zip_path, output_dom)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
73
4
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
74 temp_extract_dir = 'temp_extracted'
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
75 os.makedirs(temp_extract_dir, exist_ok=True)
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
76
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
77 with zipfile.ZipFile(output_dom, 'r') as zip_ref:
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
78 zip_ref.extractall(temp_extract_dir)
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
79
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
80 # Navigate into domesticated_genbanks and copy .gb files
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
81 dom_gb_dir = os.path.join(temp_extract_dir, 'domesticated_genbanks')
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
82 if os.path.isdir(dom_gb_dir):
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
83 for file in os.listdir(dom_gb_dir):
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
84 if file.endswith('.gb'):
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
85 shutil.copy(os.path.join(dom_gb_dir, file), gb_dom)
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
86 #print(f"Copied: {file}")
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
87 else:
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
88 print("Folder 'domesticated_genbanks' not found inside zip.")
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
89
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
90 # Optional: clean up
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
91 shutil.rmtree(temp_extract_dir)
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
92
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
93
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
94 return output_dom
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
95
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
96 def methylation_protection(domestication_target, output_methprot):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
97
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
98 if domestication_target.endswith(".zip"):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
99 extracted_dir = "extracted_genbanks"
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
100 with zipfile.ZipFile(domestication_target, "r") as zipf:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
101 os.makedirs(extracted_dir, exist_ok=True)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
102 for member in zipf.namelist():
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
103 if member.startswith("domesticated_genbanks/") and member.endswith(".gb"):
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
104 zipf.extract(member, path=extracted_dir)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
105 gb_folder = os.path.join(extracted_dir, "domesticated_genbanks")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
106 else:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
107 gb_folder = os.path.join(domestication_target, "domesticated_genbanks")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
108
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
109 records_to_protect = dnacauldron.biotools.load_records_from_files(folder=gb_folder, use_file_names_as_ids=True)
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
110
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
111 for record in records_to_protect:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
112 new_seqrecord = SeqRecord("TTC") + record + SeqRecord("GAA") # these sequences are designed to prevent Dcm recognition site formation
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
113 new_seqrecord.id = record.id
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
114 new_seqrecord.name = record.name
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
115 new_seqrecord.annotations = {"molecule_type": "DNA"}
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
116
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
117 target_file = os.path.join(output_methprot, new_seqrecord.id + ".gb")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
118 with open(target_file, "w") as output_handle:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
119 SeqIO.write(new_seqrecord, output_handle, "genbank")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
120
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
121 def parse_command_line_args():
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
122 parser = argparse.ArgumentParser(description="Domestication")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
123
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
124 parser.add_argument("--files_to_domestication", required=True,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
125 help="List of GenBank files (Comma-separated)")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
126 parser.add_argument("--csv_file", required=True,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
127 help="csv file")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
128 parser.add_argument('--file_name_mapping', type=str,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
129 help='Mapping of Galaxy filenames to original filenames')
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
130 parser.add_argument("--use_file_names_as_id", type=lambda x: x.lower() == 'true', default=True,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
131 help="Use file names as IDs (True/False)")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
132 parser.add_argument("--allow_edits", type=lambda x: x.lower() == 'true', default=True,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
133 help="Allow sequence edits")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
134 parser.add_argument("--output_dom", required=True,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
135 help="zip output for domestication results")
4
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
136 parser.add_argument("--output_methprot", required=True,
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
137 help="gb output for methylation protection")
4
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
138 parser.add_argument("--output_gb_dom", required=True,
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
139 help="gb output for mdomesticated gb files")
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
140 parser.add_argument("--methylation_protection", type=lambda x: x.lower() == 'true', default=False,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
141 help="Enable methyl protection (true/false)")
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
142
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
143 return parser.parse_args()
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
144
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
145 if __name__ == "__main__":
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
146 args = parse_command_line_args()
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
147
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
148 domestication(
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
149 args.files_to_domestication, args.csv_file,
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
150 args.file_name_mapping, args.use_file_names_as_id,
4
f78651af72e4 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents: 0
diff changeset
151 args.allow_edits, args.output_dom, args.output_gb_dom
0
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
152 )
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
153
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
154 if args.methylation_protection:
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
155 methylation_protection(
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
156 args.output_dom, args.output_methprot
8fd9af4cb080 planemo upload for repository https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb commit 6ae809b563b40bcdb6be2e74fe2a84ddad5484ae
tduigou
parents:
diff changeset
157 )