Skip to content

Commit 6228b54

Browse files
harryswift01harryswift01
authored
Merge pull request #65 from CCPBioSim/32-implement-dual-input-interface
Introduction of a YAML configuration file for CodeEntropy:
2 parents 1c1dd5a + ad0af42 commit 6228b54

File tree

4 files changed

+459
-204
lines changed

4 files changed

+459
-204
lines changed

‎CodeEntropy/main_mcc.py

+154-204
Original file line numberDiff line numberDiff line change
@@ -1,234 +1,184 @@
11
import argparse
22
import math
3+
import os
34

45
import MDAnalysis as mda
56

67
# import numpy as np
78
import pandas as pd
9+
import yaml
810

911
from CodeEntropy import EntropyFunctions as EF
1012
from CodeEntropy import LevelFunctions as LF
1113
from CodeEntropy import MDAUniverseHelper as MDAHelper
1214

1315
# from datetime import datetime
1416

17+
arg_map = {
18+
"top_traj_file": {
19+
"type": str,
20+
"nargs": "+",
21+
"help": "Path to Structure/topology file followed by Trajectory file(s)",
22+
"default": [],
23+
},
24+
"selection_string": {
25+
"type": str,
26+
"help": "Selection string for CodeEntropy",
27+
"default": "all",
28+
},
29+
"start": {
30+
"type": int,
31+
"help": "Start analysing the trajectory from this frame index",
32+
"default": 0,
33+
},
34+
"end": {
35+
"type": int,
36+
"help": "Stop analysing the trajectory at this frame index",
37+
"default": -1,
38+
},
39+
"step": {
40+
"type": int,
41+
"help": "Interval between two consecutive frames to be read index",
42+
"default": 1,
43+
},
44+
"bin_width": {
45+
"type": int,
46+
"help": "Bin width in degrees for making the histogram",
47+
"default": 30,
48+
},
49+
"tempra": {
50+
"type": float,
51+
"help": "Temperature for entropy calculation (K)",
52+
"default": 298.0,
53+
},
54+
"verbose": {
55+
"type": bool,
56+
"help": "True/False flag for noisy or quiet output",
57+
"default": False,
58+
},
59+
"thread": {"type": int, "help": "How many multiprocess to use", "default": 1},
60+
"outfile": {
61+
"type": str,
62+
"help": "Name of the file where the output will be written",
63+
"default": "outfile.out",
64+
},
65+
"resfile": {
66+
"type": str,
67+
"help": "Name of the file where the residue entropy output will be written",
68+
"default": "res_outfile.out",
69+
},
70+
"mout": {
71+
"type": str,
72+
"help": "Name of the file where certain matrices will be written",
73+
"default": None,
74+
},
75+
"force_partitioning": {"type": float, "help": "Force partitioning", "default": 0.5},
76+
"waterEntropy": {"type": bool, "help": "Calculate water entropy", "default": False},
77+
}
78+
79+
80+
def load_config(file_path):
81+
"""Load YAML configuration file."""
82+
if not os.path.exists(file_path):
83+
raise FileNotFoundError(f"Configuration file '{file_path}' not found.")
84+
85+
with open(file_path, "r") as file:
86+
config = yaml.safe_load(file)
87+
88+
# If YAML content is empty, return an empty dictionary
89+
if config is None:
90+
config = {}
91+
92+
return config
93+
94+
95+
def setup_argparse():
96+
"""Setup argument parsing dynamically based on arg_map."""
97+
parser = argparse.ArgumentParser(
98+
description="CodeEntropy: Entropy calculation with MCC method."
99+
)
100+
101+
for arg, properties in arg_map.items():
102+
kwargs = {key: properties[key] for key in properties if key != "help"}
103+
parser.add_argument(f"--{arg}", **kwargs, help=properties.get("help"))
104+
105+
return parser
106+
107+
108+
def merge_configs(args, run_config):
109+
"""Merge CLI arguments with YAML configuration."""
110+
if run_config is None:
111+
run_config = {}
112+
113+
if not isinstance(run_config, dict):
114+
raise TypeError("run_config must be a dictionary or None.")
115+
116+
# Step 1: Merge YAML configuration into args
117+
for key, value in run_config.items():
118+
if getattr(args, key, None) is None:
119+
setattr(args, key, value)
120+
121+
# Step 2: Set default values for any missing arguments from `arg_map`
122+
for key, params in arg_map.items():
123+
if getattr(args, key, None) is None:
124+
setattr(args, key, params.get("default"))
125+
126+
# Step 3: Override with CLI values if provided
127+
for key in arg_map.keys():
128+
cli_value = getattr(args, key, None)
129+
if cli_value is not None:
130+
run_config[key] = cli_value
131+
132+
return args
133+
15134

16135
def main():
17136
"""
18137
Main function for calculating the entropy of a system using the multiscale cell
19138
correlation method.
20139
"""
21-
22140
try:
23-
parser = argparse.ArgumentParser(
24-
description="""
25-
CodeEntropy-POSEIDON is a tool to compute entropy using the
26-
multiscale-cell-correlation (MCC) theory and force/torque covariance
27-
methods with the ablity to compute solvent entropy.
28-
Version:
29-
0.3.1;
30-
31-
Authors:
32-
Arghya Chakravorty (arghya90),
33-
Jas Kalayan (jkalayan),
34-
Donald Chang,
35-
Sarah Fegan
36-
Ioana Papa;
37-
38-
Output:
39-
*.csv = results from different calculateion,
40-
*.pkl - Pickled reduced universe for further analysis,
41-
*.out - detailed output such as matrix and spectra"""
42-
)
43-
44-
parser.add_argument(
45-
"-f",
46-
"--top_traj_file",
47-
required=True,
48-
dest="filePath",
49-
action="store",
50-
nargs="+",
51-
help="Path to Structure/topology file (AMBER PRMTOP, GROMACS TPR which "
52-
"contains topology and dihedral information) followed by Trajectory "
53-
"file(s) (AMBER NETCDF or GROMACS TRR) you will need to output the "
54-
"coordinates and forces to the same file. Required.",
55-
)
56-
parser.add_argument(
57-
"-l",
58-
"--selectString",
59-
action="store",
60-
dest="selection_string",
61-
type=str,
62-
default="all",
63-
help="Selection string for CodeEntropy such as protein or resid, refer to "
64-
"MDAnalysis.select_atoms for more information.",
65-
)
66-
parser.add_argument(
67-
"-b",
68-
"--begin",
69-
action="store",
70-
dest="start",
71-
help="Start analysing the trajectory from this frame index. Defaults to 0",
72-
default=0,
73-
type=int,
74-
)
75-
parser.add_argument(
76-
"-e",
77-
"--end",
78-
action="store",
79-
dest="end",
80-
help="Stop analysing the trajectory at this frame index. Defaults to -1 "
81-
"(end of trajectory file)",
82-
default=-1,
83-
type=int,
84-
)
85-
parser.add_argument(
86-
"-d",
87-
"--step",
88-
action="store",
89-
dest="step",
90-
help="interval between two consecutive frames to be read index. "
91-
"Defaults to 1",
92-
default=1,
93-
type=int,
94-
)
95-
parser.add_argument(
96-
"-n",
97-
"--bin_width",
98-
action="store",
99-
dest="bin_width",
100-
default=30,
101-
type=int,
102-
help="Bin width in degrees for making the histogram of the dihedral angles "
103-
"for the conformational entropy. Default: 30",
104-
)
105-
parser.add_argument(
106-
"-k",
107-
"--tempra",
108-
action="store",
109-
dest="temp",
110-
help="Temperature for entropy calculation (K). Default to 298.0 K",
111-
default=298.0,
112-
type=float,
113-
)
114-
parser.add_argument(
115-
"-v",
116-
"--verbose",
117-
action="store",
118-
dest="verbose",
119-
default=False,
120-
type=bool,
121-
help="True/False flag for noisy or quiet output. Default: False",
122-
)
123-
parser.add_argument(
124-
"-t",
125-
"--thread",
126-
action="store",
127-
dest="thread",
128-
help="How many multiprocess to use. Default 1 for single core execution.",
129-
default=1,
130-
type=int,
131-
)
132-
parser.add_argument(
133-
"-o",
134-
"--out",
135-
action="store",
136-
dest="outfile",
137-
default="outfile.out",
138-
help="Name of the file where the output will be written. "
139-
"Default: outfile.out",
140-
)
141-
parser.add_argument(
142-
"-r",
143-
"--resout",
144-
action="store",
145-
dest="resfile",
146-
default="res_outfile.out",
147-
help="Name of the file where the residue entropy output will be written. "
148-
"Default: res_outfile.out",
149-
)
150-
parser.add_argument(
151-
"-m",
152-
"--mout",
153-
action="store",
154-
dest="moutfile",
155-
default=None,
156-
help="Name of the file where certain matrices will be written "
157-
"(default: None).",
158-
)
159-
160-
parser.add_argument(
161-
"-c",
162-
"--cutShell",
163-
action="store",
164-
dest="cutShell",
165-
default=None,
166-
type=float,
167-
help="include cutoff shell analysis, add cutoff distance in angstrom "
168-
"Default None will ust the RAD Algorithm",
169-
)
170-
parser.add_argument(
171-
"-p",
172-
"--pureAtomNum",
173-
action="store",
174-
dest="puteAtomNum",
175-
default=1,
176-
type=int,
177-
help="Reference molecule resid for system of pure liquid. " "Default to 1",
178-
)
179-
parser.add_argument(
180-
"-x",
181-
"--excludedResnames",
182-
dest="excludedResnames",
183-
action="store",
184-
nargs="+",
185-
default=None,
186-
help="exclude a list of molecule names from nearest non-like analysis. "
187-
"Default: None. Multiples are gathered into list.",
188-
)
189-
parser.add_argument(
190-
"-w",
191-
"--water",
192-
dest="waterResnames",
193-
action="store",
194-
default="WAT",
195-
nargs="+",
196-
help="resname for water molecules. "
197-
"Default: WAT. Multiples are gathered into list.",
198-
)
199-
parser.add_argument(
200-
"-s",
201-
"--solvent",
202-
dest="solventResnames",
203-
action="store",
204-
nargs="+",
205-
default=None,
206-
help="include resname of solvent molecules (case-sensitive) "
207-
"Default: None. Multiples are gathered into list.",
208-
)
209-
parser.add_argument(
210-
"--solContact",
211-
action="store_true",
212-
dest="doSolContact",
213-
default=False,
214-
help="Do solute contact calculation",
215-
)
141+
config = load_config("config.yaml")
142+
143+
if config is None:
144+
raise ValueError(
145+
"No configuration file found, and no CLI arguments were provided."
146+
)
147+
148+
parser = setup_argparse()
149+
args, unknown = parser.parse_known_args()
150+
151+
# Process each run in the YAML configuration
152+
for run_name, run_config in config.items():
153+
if isinstance(run_config, dict):
154+
# Merging CLI arguments with YAML configuration
155+
args = merge_configs(args, run_config)
156+
157+
# Ensure necessary arguments are provided
158+
if not getattr(args, "top_traj_file"):
159+
raise ValueError(
160+
"The 'top_traj_file' argument is required but not provided."
161+
)
162+
if not getattr(args, "selection_string"):
163+
raise ValueError(
164+
"The 'selection_string' argument is required but not provided."
165+
)
216166

217-
args = parser.parse_args()
218-
except argparse.ArgumentError:
219-
print("Command line arguments are ill-defined, please check the arguments")
167+
# REPLACE INPUTS
168+
print(f"Printing all input for {run_name}")
169+
for arg in vars(args):
170+
print(f" {arg}: {getattr(args, arg) or ''}")
171+
else:
172+
print(f"Run configuration for {run_name} is not a dictionary.")
173+
except ValueError as e:
174+
print(e)
220175
raise
221176

222-
# REPLACE INPUTS
223-
print("printing all input")
224-
for arg in vars(args):
225-
print(" {} {}".format(arg, getattr(args, arg) or ""))
226-
227177
# startTime = datetime.now()
228178

229179
# Get topology and trajectory file names and make universe
230-
tprfile = args.filePath[0]
231-
trrfile = args.filePath[1:]
180+
tprfile = args.top_traj_file[0]
181+
trrfile = args.top_traj_file[1:]
232182
u = mda.Universe(tprfile, trrfile)
233183

234184
# Define bin_width for histogram from inputs

0 commit comments

Comments
 (0)