alphafold2: scripts/outputs.py annotate

annotate scripts/outputs.py @ 23:2891385d6ace draft default tip

planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty

author	galaxy-australia
date	Wed, 16 Apr 2025 05:46:58 +0000
parents	3f188450ca4f
children

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 """Generate additional output files not produced by AlphaFold.
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 Currently this is includes:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	4 - model confidence scores
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	5 - per-residue confidence scores (pLDDTs - optional output)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	6 - model_*.pkl files renamed with rank order
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 N.B. There have been issues with this script breaking between AlphaFold
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	9 versions due to minor changes in the output directory structure across minor
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	10 versions. It will likely need updating with future releases of AlphaFold.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	11
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	12 This code is more complex than you might expect due to the output files
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	13 'moving around' considerably, depending on run parameters. You will see that
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	14 several output paths are determined dynamically.
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 import argparse
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	18 import json
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	19 import numpy as np
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	20 import os
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	21 import pickle as pk
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	22 import shutil
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	23 import zipfile
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	24 from matplotlib import pyplot as plt
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	25 from pathlib import Path
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	26 from typing import Dict, List
16 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 OUTPUT_DIR = 'extra'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	29 OUTPUTS = {
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	30 'model_pkl': OUTPUT_DIR + '/ranked_{rank}.pkl',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	31 'model_pae': OUTPUT_DIR + '/pae_ranked_{rank}.csv',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	32 'model_plot': OUTPUT_DIR + '/ranked_{rank}.png',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	33 'model_confidence_scores': OUTPUT_DIR + '/model_confidence_scores.tsv',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	34 'plddts': OUTPUT_DIR + '/plddts.tsv',
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	35 'relax': OUTPUT_DIR + '/relax_metrics_ranked.json',
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	36 'msa': OUTPUT_DIR + '/msa_coverage.png',
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	37 }
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	38
20 6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	39 HTML_PATH = Path(__file__).parent / "alphafold.html"
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	40 HTML_OUTPUT_FILENAME = 'alphafold.html'
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	41 HTML_BUTTON_ATTR = 'class="btn" id="btn-ranked_{rank}"'
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	42 HTML_BUTTON_ATTR_DISABLED = (
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	43 'class="btn disabled" id="btn-ranked_{rank}" disabled')
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	44
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	45
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	46 class PLDDT_KEY:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	47 """Dict keys for accessing confidence data from JSON/pkl files."
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	48 Changes depending on which model PRESET was used.
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	49 """
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	50 monomer = 'plddts'
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	51 multimer = 'iptm+ptm'
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	52
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	53
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	54 class PRESETS:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	55 monomer = 'monomer'
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	56 monomer_ptm = 'monomer_ptm'
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	57 multimer = 'multimer'
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	58
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	59
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	60 class Settings:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	61 """Parse and store settings/config."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	62 def __init__(self):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	63 self.workdir = None
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	64 self.output_confidence_scores = True
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	65 self.output_residue_scores = False
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	66 self.parse()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	67
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	68 def parse(self) -> None:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	69 parser = argparse.ArgumentParser()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	70 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	71 "workdir",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	72 help="alphafold output directory",
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	73 type=str,
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	74 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	75 parser.add_argument(
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	76 "-s",
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	77 "--confidence-scores",
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	78 help="output per-residue confidence scores (pLDDTs)",
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	79 action="store_true",
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	80 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	81 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	82 "--pkl",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	83 help="rename model pkl outputs with rank order",
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	84 action="store_true",
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	85 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	86 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	87 "--pae",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	88 help="extract PAE from pkl files to CSV format",
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	89 action="store_true",
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	90 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	91 parser.add_argument(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	92 "--plot",
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	93 help="Plot pLDDT and PAE for each model",
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	94 action="store_true",
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	95 )
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	96 parser.add_argument(
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	97 "--plot-msa",
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	98 help="Plot multiple-sequence alignment coverage as a heatmap",
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	99 action="store_true",
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	100 )
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	101 parser.add_argument(
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	102 "--msa",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	103 help="Collect multiple-sequence alignments as ZIP archives",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	104 action="store_true",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	105 )
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	106 parser.add_argument(
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	107 "--msa_only",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	108 help="Alphafold generated MSA files only - skip all other outputs",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	109 action="store_true",
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	110 )
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	111 args = parser.parse_args()
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	112 self.workdir = Path(args.workdir.rstrip('/'))
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	113 self.output_residue_scores = args.confidence_scores
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	114 self.output_model_pkls = args.pkl
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	115 self.output_model_plots = args.plot
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	116 self.output_pae = args.pae
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	117 self.plot_msa = args.plot_msa
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	118 self.collect_msas = args.msa
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	119 self.model_preset = self._sniff_model_preset()
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	120 self.is_multimer = self.model_preset == PRESETS.multimer
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	121 self.output_dir = self.workdir / OUTPUT_DIR
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	122 self.msa_only = args.msa_only
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	123 os.makedirs(self.output_dir, exist_ok=True)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	124
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	125 def _sniff_model_preset(self) -> bool:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	126 """Check if the run was multimer or monomer."""
22 3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	127 for path in self.workdir.glob('*.pkl'):
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	128 if 'feature' not in path.name:
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	129 if '_multimer_' in path.name:
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	130 return PRESETS.multimer
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	131 if '_ptm_' in path.name:
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	132 return PRESETS.monomer_ptm
3f188450ca4f planemo upload for repository https://github.com/usegalaxy-au/tools-au commit d626bb28203543a70d3fc60d662cb054bc3cef7c galaxy-australia parents: 21 diff changeset	133 return PRESETS.monomer
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	134 return PRESETS.monomer
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	135
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	136
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	137 class ExecutionContext:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	138 """Collect file paths etc."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	139 def __init__(self, settings: Settings):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	140 self.settings = settings
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	141 if settings.is_multimer:
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	142 self.plddt_key = PLDDT_KEY.multimer
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	143 else:
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	144 self.plddt_key = PLDDT_KEY.monomer
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	145
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	146 def get_model_key(self, ix: int) -> str:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	147 """Return json key for model index.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	148
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	149 The key format changed between minor AlphaFold versions so this
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	150 function determines the correct key.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	151 """
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	152 with open(self.ranking_debug) as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	153 data = json.load(f)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	154 model_keys = list(data[self.plddt_key].keys())
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	155 for k in model_keys:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	156 if k.startswith(f"model_{ix}_"):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	157 return k
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	158 return KeyError(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	159 f'Could not find key for index={ix} in'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	160 ' ranking_debug.json')
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 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	163 def ranking_debug(self) -> str:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	164 return self.settings.workdir / 'ranking_debug.json'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	165
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	166 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	167 def relax_metrics(self) -> str:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	168 return self.settings.workdir / 'relax_metrics.json'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	169
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	170 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	171 def relax_metrics_ranked(self) -> str:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	172 return self.settings.workdir / 'relax_metrics_ranked.json'
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	173
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	174 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	175 def model_pkl_paths(self) -> List[str]:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	176 return sorted([
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	177 self.settings.workdir / f
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	178 for f in os.listdir(self.settings.workdir)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	179 if f.startswith('result_model_') and f.endswith('.pkl')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	180 ])
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
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	183 class ResultModelPrediction:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	184 """Load and manipulate data from result_model_*.pkl files."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	185 def __init__(self, path: str, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	186 self.context = context
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	187 self.path = path
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	188 self.name = os.path.basename(path).replace('result_', '').split('.')[0]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	189 with open(path, 'rb') as path:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	190 self.data = pk.load(path)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	191
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	192 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	193 def plddts(self) -> List[float]:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	194 """Return pLDDT scores for each residue."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	195 return list(self.data['plddt'])
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	196
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	197
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	198 class ResultRanking:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	199 """Load and manipulate data from ranking_debug.json file."""
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 def __init__(self, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	202 self.path = context.ranking_debug
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	203 self.context = context
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	204 with open(self.path, 'r') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	205 self.data = json.load(f)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	206
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	207 @property
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	208 def order(self) -> List[str]:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	209 """Return ordered list of model indexes."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	210 return self.data['order']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	211
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	212 def get_plddt_for_rank(self, rank: int) -> List[float]:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	213 """Get pLDDT score for model instance."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	214 return self.data[self.context.plddt_key][self.data['order'][rank - 1]]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	215
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	216 def get_rank_for_model(self, model_name: str) -> int:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	217 """Return 0-indexed rank for given model name.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	218
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	219 Model names are expressed in result_model_*.pkl file names.
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 return self.data['order'].index(model_name)
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 write_confidence_scores(ranking: ResultRanking, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	225 """Write per-model confidence scores."""
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	226 outfile = context.settings.workdir / OUTPUTS['model_confidence_scores']
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	227 scores: Dict[str, list] = {}
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	228 header = ['model', context.plddt_key]
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	229
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	230 for i, path in enumerate(context.model_pkl_paths):
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	231 rank = int(path.name.split('model_')[-1][0])
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	232 scores_ls = [ranking.get_plddt_for_rank(rank)]
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	233 with open(path, 'rb') as f:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	234 data = pk.load(f)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	235 if 'ptm' in data:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	236 scores_ls.append(data['ptm'])
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	237 if i == 0:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	238 header += ['ptm']
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	239 if 'iptm' in data:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	240 scores_ls.append(data['iptm'])
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	241 if i == 0:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	242 header += ['iptm']
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	243 scores[rank] = scores_ls
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	244
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	245 with open(outfile, 'w') as f:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	246 f.write('\t'.join(header) + '\n')
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	247 for rank, score_ls in scores.items():
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	248 row = [f"ranked_{rank - 1}"] + [str(x) for x in score_ls]
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	249 f.write('\t'.join(row) + '\n')
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	250
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 def write_per_residue_scores(
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	253 ranking: ResultRanking,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	254 context: ExecutionContext,
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	255 ):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	256 """Write per-residue plddts for each model.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	257
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	258 A row of plddt values is written for each model in tabular format.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	259 """
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	260 model_plddts = {}
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	261 for i, path in enumerate(context.model_pkl_paths):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	262 model = ResultModelPrediction(path, context)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	263 rank = ranking.get_rank_for_model(model.name)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	264 model_plddts[rank] = model.plddts
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	265
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	266 path = context.settings.workdir / OUTPUTS['plddts']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	267 with open(path, 'w') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	268 for i in sorted(list(model_plddts.keys())):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	269 row = [f'ranked_{i}'] + [
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	270 str(x) for x in model_plddts[i]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	271 ]
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	272 f.write('\t'.join(row) + '\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	273
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	274
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	275 def rename_model_pkls(ranking: ResultRanking, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	276 """Rename model.pkl files so the rank order is implicit."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	277 for path in context.model_pkl_paths:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	278 model = ResultModelPrediction(path, context)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	279 rank = ranking.get_rank_for_model(model.name)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	280 new_path = (
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	281 context.settings.workdir
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	282 / OUTPUTS['model_pkl'].format(rank=rank)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	283 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	284 shutil.copyfile(path, new_path)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	285
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	286
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	287 def extract_pae_to_csv(ranking: ResultRanking, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	288 """Extract predicted alignment error matrix from pickle files.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	289
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	290 Creates a CSV file for each of five ranked models.
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	291 """
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	292 for path in context.model_pkl_paths:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	293 model = ResultModelPrediction(path, context)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	294 rank = ranking.get_rank_for_model(model.name)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	295 with open(path, 'rb') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	296 data = pk.load(f)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	297 if 'predicted_aligned_error' not in data:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	298 print("Skipping PAE output"
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	299 f" - not found in {path}."
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	300 " Running with model_preset=monomer?")
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	301 return
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	302 pae = data['predicted_aligned_error']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	303 out_path = (
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	304 context.settings.workdir
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	305 / OUTPUTS['model_pae'].format(rank=rank)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	306 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	307 with open(out_path, 'w') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	308 for row in pae:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	309 f.write(','.join([str(x) for x in row]) + '\n')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	310
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	311
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	312 def rekey_relax_metrics(ranking: ResultRanking, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	313 """Replace keys in relax_metrics.json with 0-indexed rank."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	314 with open(context.relax_metrics) as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	315 data = json.load(f)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	316 for k in list(data.keys()):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	317 rank = ranking.get_rank_for_model(k)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	318 data[f'ranked_{rank}'] = data.pop(k)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	319 new_path = context.settings.workdir / OUTPUTS['relax']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	320 with open(new_path, 'w') as f:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	321 json.dump(data, f)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	322
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	323
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	324 def plddt_pae_plots(ranking: ResultRanking, context: ExecutionContext):
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	325 """Generate a pLDDT + PAE plot for each model."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	326 for path in context.model_pkl_paths:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	327 num_plots = 2
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	328 model = ResultModelPrediction(path, context)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	329 rank = ranking.get_rank_for_model(model.name)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	330 png_path = (
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	331 context.settings.workdir
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	332 / OUTPUTS['model_plot'].format(rank=rank)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	333 )
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	334 plddts = model.data['plddt']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	335 if 'predicted_aligned_error' in model.data:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	336 pae = model.data['predicted_aligned_error']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	337 max_pae = model.data['max_predicted_aligned_error']
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	338 else:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	339 num_plots = 1
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	340
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	341 plt.figure(figsize=[8 * num_plots, 6])
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	342 plt.subplot(1, num_plots, 1)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	343 plt.plot(plddts)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	344 plt.title('Predicted LDDT')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	345 plt.xlabel('Residue')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	346 plt.ylabel('pLDDT')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	347
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	348 if num_plots == 2:
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	349 plt.subplot(1, 2, 2)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	350 plt.imshow(pae, vmin=0., vmax=max_pae, cmap='Greens_r')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	351 plt.colorbar(fraction=0.046, pad=0.04)
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	352 plt.title('Predicted Aligned Error')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	353 plt.xlabel('Scored residue')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	354 plt.ylabel('Aligned residue')
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	355
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	356 plt.savefig(png_path)
21 e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	357 plt.close()
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	358
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	359
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	360 def plot_msa(wdir: Path, dpi: int = 150):
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	361 """Plot MSA as a heatmap."""
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	362 with open(wdir / 'features.pkl', 'rb') as f:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	363 features = pk.load(f)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	364
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	365 msa = features.get('msa')
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	366 if msa is None:
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	367 print("Could not plot MSA coverage - 'msa' key not found in"
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	368 " features.pkl")
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	369 return
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	370 seqid = (np.array(msa[0] == msa).mean(-1))
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	371 seqid_sort = seqid.argsort()
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	372 non_gaps = (msa != 21).astype(float)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	373 non_gaps[non_gaps == 0] = np.nan
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	374 final = non_gaps[seqid_sort] * seqid[seqid_sort, None]
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	375
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	376 plt.figure(figsize=(6, 4))
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	377 # plt.subplot(111)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	378 plt.title("Sequence coverage")
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	379 plt.imshow(final,
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	380 interpolation='nearest', aspect='auto',
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	381 cmap="rainbow_r", vmin=0, vmax=1, origin='lower')
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	382 plt.plot((msa != 21).sum(0), color='black')
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	383 plt.xlim(-0.5, msa.shape[1] - 0.5)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	384 plt.ylim(-0.5, msa.shape[0] - 0.5)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	385 plt.colorbar(label="Sequence identity to query", )
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	386 plt.xlabel("Positions")
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	387 plt.ylabel("Sequences")
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	388 plt.tight_layout()
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	389 plt.savefig(wdir / OUTPUTS['msa'], dpi=dpi)
e7f1b552a695 planemo upload for repository https://github.com/usegalaxy-au/tools-au commit 628c9fdcb77489063145a2307b6bb6a450416dd6-dirty galaxy-australia parents: 20 diff changeset	390 plt.close()
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	391
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	392
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	393 def collect_msas(settings: Settings):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	394 """Collect MSA files into ZIP archive(s)."""
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	395
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	396 def zip_dir(directory: Path, is_multimer: bool, name: str):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	397 chain_id = directory.with_suffix('.zip').stem
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	398 msa_dir = settings.output_dir / 'msas'
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	399 msa_dir.mkdir(exist_ok=True)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	400 zip_name = (
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	401 f"MSA-{chain_id}-{name}.zip"
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	402 if is_multimer
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	403 else f"MSA-{name}.zip")
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	404 zip_path = msa_dir / zip_name
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	405 with zipfile.ZipFile(zip_path, 'w') as z:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	406 for path in directory.glob('*'):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	407 z.write(path, path.name)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	408
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	409 print("Collecting MSA archives...")
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	410 chain_names = get_input_sequence_ids(
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	411 settings.workdir.parent.parent / 'alphafold.fasta')
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	412 msa_dir = settings.workdir / 'msas'
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	413 is_multimer = (msa_dir / 'A').exists()
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	414 if is_multimer:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	415 msa_dirs = sorted([
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	416 path for path in msa_dir.glob('*')
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	417 if path.is_dir()
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	418 ])
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	419 for i, path in enumerate(msa_dirs):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	420 zip_dir(path, is_multimer, chain_names[i])
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	421 else:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	422 zip_dir(msa_dir, is_multimer, chain_names[0])
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	423
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	424
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	425 def get_input_sequence_ids(fasta_file: Path) -> List[str]:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	426 """Read headers from the input FASTA file.
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	427 Split them to get a sequence ID and truncate to 20 chars max.
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	428 """
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	429 headers = []
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	430 for line in fasta_file.read_text().split('\n'):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	431 if line.startswith('>'):
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	432 seq_id = line[1:].split(' ')[0]
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	433 seq_id_trunc = seq_id[:20].strip()
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	434 if len(seq_id) > 20:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	435 seq_id_trunc += '...'
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	436 headers.append(seq_id_trunc)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	437 return headers
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	438
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	439
20 6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	440 def template_html(context: ExecutionContext):
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	441 """Template HTML file.
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	442
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	443 Remove buttons that are redundant with limited model outputs.
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	444 """
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	445 print("Templating HTML file...")
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	446 with open(HTML_PATH) as f:
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	447 html = f.read()
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	448 for i in range(len(context.model_pkl_paths), 5):
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	449 btn_id = HTML_BUTTON_ATTR.format(rank=i)
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	450 btn_attr_disabled = HTML_BUTTON_ATTR_DISABLED.format(rank=i)
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	451 html = html.replace(btn_id, btn_attr_disabled)
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	452 with open(context.settings.output_dir / HTML_OUTPUT_FILENAME, 'w') as f:
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	453 f.write(html)
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	454
6ab1a261520a planemo upload for repository https://github.com/usegalaxy-au/tools-au commit c3a90eb12ada44d477541baa4dd6182be29cd554-dirty galaxy-australia parents: 16 diff changeset	455
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	456 def main():
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	457 """Parse output files and generate additional output files."""
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	458 settings = Settings()
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	459 if not settings.msa_only:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	460 context = ExecutionContext(settings)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	461 ranking = ResultRanking(context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	462 write_confidence_scores(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	463 rekey_relax_metrics(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	464 template_html(context)
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	465
23 2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	466 # Optional outputs
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	467 if settings.output_model_pkls:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	468 rename_model_pkls(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	469 if settings.output_model_plots:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	470 plddt_pae_plots(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	471 if settings.output_pae:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	472 # Only created by monomer_ptm and multimer models
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	473 extract_pae_to_csv(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	474 if settings.output_residue_scores:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	475 write_per_residue_scores(ranking, context)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	476 if settings.plot_msa:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	477 plot_msa(settings.workdir)
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	478 if settings.collect_msas or settings.msa_only:
2891385d6ace planemo upload for repository https://github.com/usegalaxy-au/tools-au commit b347c6ccc82b14fcbff360b3357050d1d43e3ef5-dirty galaxy-australia parents: 22 diff changeset	479 collect_msas(settings)
16 f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	480
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	481
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	482 if __name__ == '__main__':
f9eb041c518c planemo upload for repository https://github.com/usegalaxy-au/tools-au commit ee77734f1800350fa2a6ef28b2b8eade304a456f-dirty galaxy-australia parents: diff changeset	483 main()

Mercurial > repos > galaxy-australia > alphafold2

annotate scripts/outputs.py @ 23:2891385d6ace draft default tip