annotate qsar1.py @ 0:ce46f2008024 draft default tip

planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
author mbernt
date Tue, 09 Apr 2024 07:51:18 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
1 import argparse
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
2 import re
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
3
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
4 import pandas as pd
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
5
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
6 parser = argparse.ArgumentParser(description='Calculate baseline toxicity for different aquatic species')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
7 parser.add_argument('--function', type=str, choices=['calculate_baseline', 'apply_linear_functions'],
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
8 help='Function to execute')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
9 parser.add_argument('--csv_input', type=argparse.FileType('r'), help='Path to the input csv file')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
10 parser.add_argument('--functions_csv', type=argparse.FileType('r'), default=None,
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
11 help='Path to the csv file containing functions (only for apply_linear_functions)')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
12 parser.add_argument('--output', type=argparse.FileType('w'), help='Path for the output csv file')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
13 args = parser.parse_args()
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
14
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
15 if args.function == 'calculate_baseline':
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
16 df = pd.read_csv(args.csv_input)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
17 df.iloc[:, 0] = df.iloc[:, 0].astype(int)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
18 df['Caenorhabditis elegans [mol/L]'] = 10 ** (-(0.81 * df.iloc[:, 0] + 1.15))
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
19 df['Daphia magna [mol/L]'] = 10 ** (-(0.82 * df.iloc[:, 0] + 1.48))
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
20 df['Danio rerio [mol/L]'] = 10 ** (-(0.99 * df.iloc[:, 0] + 0.78))
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
21 df['Generic Human Cells [mol/L]'] = 0.026 / (10 ** df.iloc[:, 0]) * (1 + 10 ** (0.7 * df.iloc[:, 0] + 0.34) * 3 * 0.001 + 10 ** 3 * 0.07 * 0.001)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
22 df.to_csv(args.output, index=False)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
23
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
24 elif args.function == 'apply_linear_functions':
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
25 df = pd.read_csv(args.csv_input)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
26 functions_df = pd.read_csv(args.functions_csv)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
27
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
28 def parse_and_apply_equation(equation, x_values):
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
29 # Extract 'a' and 'b' from the equation (assuming the format 'ax+b' or 'ax-b')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
30 pattern = re.compile(r'([+-]?\d*\.?\d*)x([+-]\d+)?')
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
31 match = pattern.search(equation)
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
32 a = float(match.group(1)) if match.group(1) not in ('', '+', '-') else 1.0
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
33 b = float(match.group(2)) if match.group(2) else 0
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
34 return a * x_values + b
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
35
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
36 for i, row in functions_df.iterrows():
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
37 func = row['function']
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
38 df[f'result_{i}'] = parse_and_apply_equation(func, df['logD'])
ce46f2008024 planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tools/tox_tools/baseline_calculator commit 008f820fb9b8ec547e00205f809f982b8f4b8318
mbernt
parents:
diff changeset
39 df.to_csv(args.output, index=False)