annotate validate_fasta.py @ 7:eb085b3dbaf8 draft

"planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
author galaxy-australia
date Tue, 19 Apr 2022 00:39:29 +0000
parents 04e95886cf24
children ca90d17ff51b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
1 """Validate input FASTA sequence."""
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
2
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
3 import re
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
4 import sys
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
5 import argparse
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
6 from typing import List, TextIO
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
7
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
8
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
9 class Fasta:
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
10 def __init__(self, header_str: str, seq_str: str):
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
11 self.header = header_str
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
12 self.aa_seq = seq_str
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
13
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
14
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
15 class FastaLoader:
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
16 def __init__(self, fasta_path: str):
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
17 """Initialize from FASTA file."""
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
18 self.fastas = []
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
19 self.load(fasta_path)
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
20
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
21 def load(self, fasta_path: str):
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
22 """Load bare or FASTA formatted sequence."""
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
23 with open(fasta_path, 'r') as f:
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
24 self.content = f.read()
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
25
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
26 if "__cn__" in self.content:
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
27 # Pasted content with escaped characters
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
28 self.newline = '__cn__'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
29 self.read_caret = '__gt__'
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
30 else:
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
31 # Uploaded file with normal content
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
32 self.newline = '\n'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
33 self.read_caret = '>'
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
34
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
35 self.lines = self.content.split(self.newline)
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
36
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
37 if not self.lines[0].startswith(self.read_caret):
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
38 # Fasta is headless, load as single sequence
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
39 self.update_fastas(
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
40 '', ''.join(self.lines)
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
41 )
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
42
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
43 else:
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
44 header = None
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
45 sequence = None
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
46 for line in self.lines:
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
47 if line.startswith(self.read_caret):
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
48 if header:
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
49 self.update_fastas(header, sequence)
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
50 header = '>' + self.strip_header(line)
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
51 sequence = ''
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
52 else:
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
53 sequence += line.strip('\n ')
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
54 self.update_fastas(header, sequence)
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
55
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
56 def strip_header(self, line):
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
57 """Strip characters escaped with underscores from pasted text."""
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
58 return re.sub(r'\_\_.{2}\_\_', '', line).strip('>')
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
59
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
60 def update_fastas(self, header: str, sequence: str):
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
61 # if we have a sequence
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
62 if sequence:
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
63 # create generic header if not exists
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
64 if not header:
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
65 fasta_count = len(self.fastas)
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
66 header = f'>sequence_{fasta_count}'
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
67
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
68 # Create new Fasta
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
69 self.fastas.append(Fasta(header, sequence))
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
70
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
71
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
72 class FastaValidator:
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
73 def __init__(
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
74 self,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
75 fasta_list: List[Fasta],
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
76 min_length=None,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
77 max_length=None):
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
78 self.min_length = min_length
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
79 self.max_length = max_length
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
80 self.fasta_list = fasta_list
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
81 self.iupac_characters = {
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
82 'A', 'B', 'C', 'D', 'E', 'F', 'G',
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
83 'H', 'I', 'K', 'L', 'M', 'N', 'P',
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
84 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
85 'Y', 'Z', '-'
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
86 }
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
87
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
88 def validate(self):
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
89 """Perform FASTA validation."""
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
90 self.validate_num_seqs()
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
91 self.validate_length()
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
92 self.validate_alphabet()
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
93
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
94 # not checking for 'X' nucleotides at the moment.
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
95 # alphafold can throw an error if it doesn't like it.
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
96 # self.validate_x()
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
97
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
98 def validate_num_seqs(self) -> None:
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
99 """Assert that only one sequence has been provided."""
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
100 if len(self.fasta_list) > 1:
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
101 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
102 'Error encountered validating FASTA:\n'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
103 f' More than 1 sequence detected ({len(self.fasta_list)}).'
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
104 ' Please use single FASTA sequence as input.')
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
105 elif len(self.fasta_list) == 0:
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
106 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
107 'Error encountered validating FASTA:\n'
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
108 ' input file has no FASTA sequences')
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
109
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
110 def validate_length(self):
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
111 """Confirm whether sequence length is valid."""
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
112 fasta = self.fasta_list[0]
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
113 if self.min_length:
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
114 if len(fasta.aa_seq) < self.min_length:
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
115 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
116 'Error encountered validating FASTA:\n Sequence too short'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
117 f' ({len(fasta.aa_seq)}AA).'
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
118 f' Minimum length is {self.min_length}AA.')
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
119 if self.max_length:
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
120 if len(fasta.aa_seq) > self.max_length:
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
121 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
122 'Error encountered validating FASTA:\n'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
123 f' Sequence too long ({len(fasta.aa_seq)}AA).'
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
124 f' Maximum length is {self.max_length}AA.')
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
125
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
126 def validate_alphabet(self):
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
127 """Confirm whether the sequence conforms to IUPAC codes.
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
128
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
129 If not, report the offending character and its position.
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
130 """
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
131 fasta = self.fasta_list[0]
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
132 for i, char in enumerate(fasta.aa_seq.upper()):
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
133 if char not in self.iupac_characters:
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
134 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
135 'Error encountered validating FASTA:\n Invalid amino acid'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
136 f' found at pos {i}: "{char}"')
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
137
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
138 def validate_x(self):
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
139 """Check for X bases."""
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
140 fasta = self.fasta_list[0]
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
141 for i, char in enumerate(fasta.aa_seq.upper()):
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
142 if char == 'X':
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
143 raise ValueError(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
144 'Error encountered validating FASTA:\n Unsupported AA code'
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
145 f' "X" found at pos {i}')
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
146
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
147
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
148 class FastaWriter:
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
149 def __init__(self) -> None:
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
150 self.line_wrap = 60
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
151
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
152 def write(self, fasta: Fasta):
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
153 header = fasta.header
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
154 seq = self.format_sequence(fasta.aa_seq)
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
155 sys.stdout.write(header + '\n')
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
156 sys.stdout.write(seq)
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
157
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
158 def format_sequence(self, aa_seq: str):
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
159 formatted_seq = ''
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
160 for i in range(0, len(aa_seq), self.line_wrap):
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
161 formatted_seq += aa_seq[i: i + self.line_wrap] + '\n'
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
162 return formatted_seq
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
163
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
164
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
165 def main():
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
166 # load fasta file
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
167 try:
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
168 args = parse_args()
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
169 fas = FastaLoader(args.input)
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
170
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
171 # validate
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
172 fv = FastaValidator(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
173 fas.fastas,
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
174 min_length=args.min_length,
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
175 max_length=args.max_length,
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
176 )
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
177 fv.validate()
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
178
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
179 # write cleaned version
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
180 fw = FastaWriter()
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
181 fw.write(fas.fastas[0])
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
182
7
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
183 except ValueError as exc:
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
184 sys.stderr.write(f"{exc}\n\n")
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
185 raise exc
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
186
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
187 except Exception as exc:
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
188 sys.stderr.write(
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
189 "Input error: FASTA input is invalid. Please check your input.\n\n"
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
190 )
eb085b3dbaf8 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 8d9f0ae6af9e8d9313c6cdcc551b24c6c44ae341"
galaxy-australia
parents: 6
diff changeset
191 raise exc
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
192
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
193
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
194 def parse_args() -> argparse.Namespace:
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
195 parser = argparse.ArgumentParser()
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
196 parser.add_argument(
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
197 "input",
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
198 help="input fasta file",
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
199 type=str
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
200 )
6
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
201 parser.add_argument(
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
202 "--min_length",
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
203 dest='min_length',
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
204 help="Minimum length of input protein sequence (AA)",
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
205 default=None,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
206 type=int,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
207 )
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
208 parser.add_argument(
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
209 "--max_length",
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
210 dest='max_length',
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
211 help="Maximum length of input protein sequence (AA)",
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
212 default=None,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
213 type=int,
04e95886cf24 "planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 724a7a389c878dded1c0332f3b6e507e0c4cd52a-dirty"
galaxy-australia
parents: 1
diff changeset
214 )
0
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
215 return parser.parse_args()
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
216
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
217
7ae9d78b06f5 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit 7b79778448363aa8c9b14604337e81009e461bd2-dirty"
galaxy-australia
parents:
diff changeset
218 if __name__ == '__main__':
1
6c92e000d684 "planemo upload for repository https://github.com/usegalaxy-au/galaxy-local-tools commit a510e97ebd604a5e30b1f16e5031f62074f23e86"
galaxy-australia
parents: 0
diff changeset
219 main()