alphafold2: scripts/validate

annotate scripts/validate_fasta.py @ 16:f9eb041c518c draft

planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty

author	galaxy-australia
date	Mon, 03 Apr 2023 01:00:42 +0000
parents
children	e4a053d67e24

rev	line source
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	1 """Validate input FASTA sequence."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	2
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	3 import argparse
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	4 import re
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	5 import sys
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	6 from typing import List
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	7
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	8 MULTIMER_MAX_SEQUENCE_COUNT = 10
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	9
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	10
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	11 class Fasta:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	12 def __init__(self, header_str: str, seq_str: str):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	13 self.header = header_str
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	14 self.aa_seq = seq_str
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	15
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	16
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	17 class FastaLoader:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	18 def __init__(self, fasta_path: str):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	19 """Initialize from FASTA file."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	20 self.fastas = []
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	21 self.load(fasta_path)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	22
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	23 def load(self, fasta_path: str):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	24 """Load bare or FASTA formatted sequence."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	25 with open(fasta_path, 'r') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	26 self.content = f.read()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	27
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	28 if "__cn__" in self.content:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	29 # Pasted content with escaped characters
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	30 self.newline = '__cn__'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	31 self.read_caret = '__gt__'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	32 else:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	33 # Uploaded file with normal content
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	34 self.newline = '\n'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	35 self.read_caret = '>'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	36
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	37 self.lines = self.content.split(self.newline)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	38
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	39 if not self.lines[0].startswith(self.read_caret):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	40 # Fasta is headless, load as single sequence
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	41 self.update_fastas(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	42 '', ''.join(self.lines)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	43 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	44
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	45 else:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	46 header = None
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	47 sequence = None
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	48 for line in self.lines:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	49 if line.startswith(self.read_caret):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	50 if header:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	51 self.update_fastas(header, sequence)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	52 header = '>' + self.strip_header(line)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	53 sequence = ''
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	54 else:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	55 sequence += line.strip('\n ')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	56 self.update_fastas(header, sequence)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	57
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	58 def strip_header(self, line):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	59 """Strip characters escaped with underscores from pasted text."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	60 return re.sub(r'\_\_.{2}\_\_', '', line).strip('>')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	61
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	62 def update_fastas(self, header: str, sequence: str):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	63 # if we have a sequence
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	64 if sequence:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	65 # create generic header if not exists
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	66 if not header:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	67 fasta_count = len(self.fastas)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	68 header = f'>sequence_{fasta_count}'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	69
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	70 # Create new Fasta
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	71 self.fastas.append(Fasta(header, sequence))
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	72
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	73
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	74 class FastaValidator:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	75 def __init__(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	76 self,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	77 min_length=None,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	78 max_length=None,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	79 multiple=False):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	80 self.multiple = multiple
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	81 self.min_length = min_length
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	82 self.max_length = max_length
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	83 self.iupac_characters = {
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	84 'A', 'B', 'C', 'D', 'E', 'F', 'G',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	85 'H', 'I', 'K', 'L', 'M', 'N', 'P',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	86 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	87 'Y', 'Z', '-'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	88 }
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	89
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	90 def validate(self, fasta_list: List[Fasta]):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	91 """Perform FASTA validation."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	92 self.fasta_list = fasta_list
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	93 self.validate_num_seqs()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	94 self.validate_length()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	95 self.validate_alphabet()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	96 # not checking for 'X' nucleotides at the moment.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	97 # alphafold can throw an error if it doesn't like it.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	98 # self.validate_x()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	99 return self.fasta_list
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	100
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	101 def validate_num_seqs(self) -> None:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	102 """Assert that only one sequence has been provided."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	103 fasta_count = len(self.fasta_list)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	104
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	105 if self.multiple:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	106 if fasta_count < 2:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	107 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	108 'Error encountered validating FASTA:\n'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	109 'Multimer mode requires multiple input sequence.'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	110 f' Only {fasta_count} sequences were detected in'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	111 ' the provided file.')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	112 self.fasta_list = self.fasta_list
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	113
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	114 elif fasta_count > MULTIMER_MAX_SEQUENCE_COUNT:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	115 sys.stderr.write(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	116 f'WARNING: detected {fasta_count} sequences but the'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	117 f' maximum allowed is {MULTIMER_MAX_SEQUENCE_COUNT}'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	118 ' sequences. The last'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	119 f' {fasta_count - MULTIMER_MAX_SEQUENCE_COUNT} sequence(s)'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	120 ' have been discarded.\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	121 self.fasta_list = self.fasta_list[:MULTIMER_MAX_SEQUENCE_COUNT]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	122 else:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	123 if fasta_count > 1:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	124 sys.stderr.write(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	125 'WARNING: More than 1 sequence detected.'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	126 ' Using first FASTA sequence as input.\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	127 self.fasta_list = self.fasta_list[:1]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	128
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	129 elif len(self.fasta_list) == 0:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	130 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	131 'Error encountered validating FASTA:\n'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	132 ' no FASTA sequences detected in input file.')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	133
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	134 def validate_length(self):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	135 """Confirm whether sequence length is valid."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	136 fasta = self.fasta_list[0]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	137 if self.min_length:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	138 if len(fasta.aa_seq) < self.min_length:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	139 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	140 'Error encountered validating FASTA:\n Sequence too short'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	141 f' ({len(fasta.aa_seq)}AA).'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	142 f' Minimum length is {self.min_length}AA.')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	143 if self.max_length:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	144 if len(fasta.aa_seq) > self.max_length:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	145 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	146 'Error encountered validating FASTA:\n'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	147 f' Sequence too long ({len(fasta.aa_seq)}AA).'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	148 f' Maximum length is {self.max_length}AA.')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	149
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	150 def validate_alphabet(self):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	151 """Confirm whether the sequence conforms to IUPAC codes.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	152
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	153 If not, report the offending character and its position.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	154 """
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	155 fasta = self.fasta_list[0]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	156 for i, char in enumerate(fasta.aa_seq.upper()):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	157 if char not in self.iupac_characters:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	158 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	159 'Error encountered validating FASTA:\n Invalid amino acid'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	160 f' found at pos {i}: "{char}"')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	161
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	162 def validate_x(self):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	163 """Check for X bases."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	164 fasta = self.fasta_list[0]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	165 for i, char in enumerate(fasta.aa_seq.upper()):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	166 if char == 'X':
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	167 raise ValueError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	168 'Error encountered validating FASTA:\n Unsupported AA code'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	169 f' "X" found at pos {i}')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	170
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	171
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	172 class FastaWriter:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	173 def __init__(self) -> None:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	174 self.line_wrap = 60
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	175
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	176 def write(self, fasta: Fasta):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	177 header = fasta.header
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	178 seq = self.format_sequence(fasta.aa_seq)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	179 sys.stdout.write(header + '\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	180 sys.stdout.write(seq)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	181
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	182 def format_sequence(self, aa_seq: str):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	183 formatted_seq = ''
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	184 for i in range(0, len(aa_seq), self.line_wrap):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	185 formatted_seq += aa_seq[i: i + self.line_wrap] + '\n'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	186 return formatted_seq.upper()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	187
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	188
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	189 def main():
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	190 # load fasta file
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	191 try:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	192 args = parse_args()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	193 fas = FastaLoader(args.input)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	194
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	195 # validate
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	196 fv = FastaValidator(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	197 min_length=args.min_length,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	198 max_length=args.max_length,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	199 multiple=args.multimer,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	200 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	201 clean_fastas = fv.validate(fas.fastas)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	202
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	203 # write clean data
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	204 fw = FastaWriter()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	205 for fas in clean_fastas:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	206 fw.write(fas)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	207
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	208 sys.stderr.write("Validated FASTA sequence(s):\n\n")
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	209 for fas in clean_fastas:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	210 sys.stderr.write(fas.header + '\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	211 sys.stderr.write(fas.aa_seq + '\n\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	212
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	213 except ValueError as exc:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	214 sys.stderr.write(f"{exc}\n\n")
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	215 raise exc
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	216
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	217 except Exception as exc:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	218 sys.stderr.write(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	219 "Input error: FASTA input is invalid. Please check your input.\n\n"
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	220 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	221 raise exc
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	222
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	223
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	224 def parse_args() -> argparse.Namespace:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	225 parser = argparse.ArgumentParser()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	226 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	227 "input",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	228 help="input fasta file",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	229 type=str
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	230 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	231 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	232 "--min_length",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	233 dest='min_length',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	234 help="Minimum length of input protein sequence (AA)",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	235 default=None,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	236 type=int,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	237 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	238 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	239 "--max_length",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	240 dest='max_length',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	241 help="Maximum length of input protein sequence (AA)",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	242 default=None,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	243 type=int,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	244 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	245 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	246 "--multimer",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	247 action='store_true',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	248 help="Require multiple input sequences",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	249 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	250 return parser.parse_args()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	251
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	252
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	253 if __name__ == '__main__':
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	254 main()

Mercurial > repos > galaxy-australia > alphafold2

annotate scripts/validate_fasta.py @ 16:f9eb041c518c draft