Coverage for biobb_flexserv/pcasuite/pcz_hinges.py: 55%
130 statements
« prev ^ index » next coverage.py v7.9.1, created at 2025-06-19 15:08 +0000
1#!/usr/bin/env python3
3"""Module containing the PCZhinges class and the command line interface."""
4import argparse
5from typing import Optional
6import shutil
7import json
8import re
9from pathlib import PurePath
10from biobb_common.tools import file_utils as fu
11from biobb_common.generic.biobb_object import BiobbObject
12from biobb_common.configuration import settings
13from biobb_common.tools.file_utils import launchlogger
16class PCZhinges(BiobbObject):
17 """
18 | biobb_flexserv PCZhinges
19 | Compute possible hinge regions (residues around which large protein movements are organized) of a molecule from a compressed PCZ file.
20 | Wrapper of the pczdump tool from the PCAsuite FlexServ module.
22 Args:
23 input_pcz_path (str): Input compressed trajectory file. File type: input. `Sample file <https://github.com/bioexcel/biobb_flexserv/raw/master/biobb_flexserv/test/data/pcasuite/pcazip.pcz>`_. Accepted formats: pcz (edam:format_3874).
24 output_json_path (str): Output hinge regions x PCA mode file. File type: output. `Sample file <https://github.com/bioexcel/biobb_flexserv/raw/master/biobb_flexserv/test/reference/pcasuite/hinges.json>`_. Accepted formats: json (edam:format_3464).
25 properties (dict - Python dictionary object containing the tool parameters, not input/output files):
26 * **binary_path** (*str*) - ("pczdump") pczdump binary path to be used.
27 * **eigenvector** (*int*) - (0) PCA mode (eigenvector) from which to extract bfactor values per residue (0 means average over all modes).
28 * **method** (*str*) - ("Dynamic_domain") Method to compute the hinge regions (Options: Bfactor_slope, Force_constant, Dynamic_domain)
29 * **remove_tmp** (*bool*) - (True) [WF property] Remove temporal files.
30 * **restart** (*bool*) - (False) [WF property] Do not execute if output files exist.
31 * **sandbox_path** (*str*) - ("./") [WF property] Parent path to the sandbox directory.
33 Examples:
34 This is a use example of how to use the building block from Python::
36 from biobb_flexserv.pcasuite.pcz_hinges import pcz_hinges
37 prop = {
38 'eigenvector': 1,
39 'pdb': True
40 }
41 pcz_hinges( input_pcz_path='/path/to/pcazip_input.pcz',
42 output_json_path='/path/to/hinges.json',
43 properties=prop)
45 Info:
46 * wrapped_software:
47 * name: FlexServ PCAsuite
48 * version: >=1.0
49 * license: Apache-2.0
50 * ontology:
51 * name: EDAM
52 * schema: http://edamontology.org/EDAM.owl
54 """
56 def __init__(self, input_pcz_path: str, output_json_path: str,
57 properties: Optional[dict] = None, **kwargs) -> None:
59 properties = properties or {}
61 # Call parent class constructor
62 super().__init__(properties)
63 self.locals_var_dict = locals().copy()
65 # Input/Output files
66 self.io_dict = {
67 'in': {'input_pcz_path': input_pcz_path},
68 'out': {'output_json_path': output_json_path}
69 }
71 # Properties specific for BB
72 self.properties = properties
73 self.binary_path = properties.get('binary_path', 'pczdump')
74 self.eigenvector = properties.get('eigenvector', 1)
75 self.method = properties.get('method', "Bfactor_slope")
77 # Check the properties
78 self.check_properties(properties)
79 self.check_arguments()
81 def parse_output(self, output_file):
82 """ Parses FlexServ hinges methods output file report """
84 method = ''
85 if self.method == "Bfactor_slope":
86 method = "#### Distance variation method"
87 elif self.method == "Force_constant":
88 method = "#### Force constant"
89 elif self.method == "Dynamic_domain":
90 method = "#### Lavery method"
91 else:
92 print("Method not recognised ({}), please check it and try again. ".format(self.method))
94 start = False
95 out_data = ''
96 with open(output_file, 'r') as file:
97 for line in file:
98 if method in line:
99 start = True
100 elif "####" in line:
101 start = False
102 if start:
103 out_data += line
105 dict_out = {}
106 dict_out["method"] = self.method
107 if self.method == "Force_constant":
108 dict_out["values_per_residue"] = []
109 for line in out_data.split("\n"):
110 if line and "#" not in line:
111 dict_out["values_per_residue"].append(float(line.strip()))
112 if "possible hinge" in line: # Peak constant (possible hinge): residue 64 (16.740)
113 residue = int(line.split(' ')[6])
114 dict_out["hinge_residues"] = residue
115 elif self.method == "Bfactor_slope":
116 dict_out["hinge_residues"] = []
117 for line in out_data.split("\n"):
118 if "Window" in line: # Window 28: residue 54 seems a downhill hinge point
119 residue = int(re.split(r'\s+', line)[3])
120 dict_out["hinge_residues"].append(residue)
121 if "Consensus" in line: # Consensus Downhill hinge point : 23.7 ( 64.965)
122 hinge_point = float(line.split(':')[1].split('(')[0])
123 dict_out["consensus_hinge"] = hinge_point
124 elif self.method == "Dynamic_domain":
125 start = 0
126 dict_out["clusters"] = []
127 for line in out_data.split("\n"):
128 if "threshold" not in line and "nClusters" in line: # nClusters: 2
129 nclusters = int(line.split(':')[1])
130 dict_out["nClusters"] = nclusters
131 if "Threshold" in line: # *** Threshold defined: 0.300000
132 threshold = float(line.split(':')[1])
133 dict_out["threshold"] = threshold
134 if "Min. drij" in line: # *** Min. drij: 0.000322
135 minValue = float(line.split(':')[1])
136 dict_out["minValue"] = minValue
137 if "Max. drij" in line: # *** Max. drij: 6.385425
138 maxValue = float(line.split(':')[1])
139 dict_out["maxValue"] = maxValue
140 if "threshold" in line: # nClusters: 2 threshold: 3.192873
141 final_threshold = float(line.split(':')[2])
142 dict_out["final_threshold"] = final_threshold
143 if "Cluster" in line and "elements" in line: # Cluster 0 (74 elements)
144 clusterLine = line.split()
145 clusterNum = int(clusterLine[1])
146 clusterElems = int(clusterLine[2].replace('(', ''))
147 cluster = {"clusterNum": clusterNum, "clusterElems": clusterElems}
148 dict_out["clusters"].append(cluster)
149 start = start + 1
150 if start and "[" in line:
151 # dict_out["clusters"][start-1]["residues"] = list(map(int,list(line.replace(", ]", "").replace(" [","").split(', '))))
152 dict_out["clusters"][start-1]["residues"] = eval(line)
153 # Interacting regions: 13 14 30 31 69 70 84 85 112 113 114 115 116 166 167 199 200
154 if "Interacting regions" in line:
155 nums = line.split(':')[1]
156 dict_out["interacting_regions"] = list(map(int, nums.split()))
157 # Hinge residues: 13 14 30 31 69 70 84 85 112 113 114 115 116 166 167 199 200
158 if "Hinge residues" in line:
159 nums = line.split(':')[1]
160 dict_out["hinge_residues"] = list(map(int, nums.split()))
162 return dict_out
164 @launchlogger
165 def launch(self):
166 """Launches the execution of the FlexServ pcz_hinges module."""
168 # Setup Biobb
169 if self.check_restart():
170 return 0
171 # self.stage_files()
173 # Internal file paths
174 # try:
175 # # Using rel paths to shorten the amount of characters due to fortran path length limitations
176 # input_pcz = str(Path(self.stage_io_dict["in"]["input_pcz_path"]).relative_to(Path.cwd()))
177 # output_json = str(Path(self.stage_io_dict["out"]["output_json_path"]).relative_to(Path.cwd()))
178 # except ValueError:
179 # # Container or remote case
180 # input_pcz = self.stage_io_dict["in"]["input_pcz_path"]
181 # output_json = self.stage_io_dict["out"]["output_json_path"]
183 # Manually creating a Sandbox to avoid issues with input parameters buffer overflow:
184 # Long strings defining a file path makes Fortran or C compiled programs crash if the string
185 # declared is shorter than the input parameter path (string) length.
186 # Generating a temporary folder and working inside this folder (sandbox) fixes this problem.
187 # The problem was found in Galaxy executions, launching Singularity containers (May 2023).
189 # Creating temporary folder
190 self.tmp_folder = fu.create_unique_dir()
191 fu.log('Creating %s temporary folder' % self.tmp_folder, self.out_log)
193 shutil.copy2(self.io_dict["in"]["input_pcz_path"], self.tmp_folder)
195 # Temporary output
196 # temp_out = str(Path(self.stage_io_dict.get("unique_dir", "")).joinpath("output.dat"))
197 temp_out = "output.dat"
198 temp_log = "output.log"
199 temp_json = "output.json"
201 # Command line (1: dat file)
202 # pczdump -i structure.ca.std.pcz --fluc=1 -o bfactor_1.dat
203 # self.cmd = [self.binary_path,
204 # "-i", input_pcz,
205 # "-o", temp_out,
206 # "-t", "0.3",
207 # "--hinge={}".format(self.eigenvector),
208 # ">&", "pcz_dump.hinges.log"
209 # ]
211 self.cmd = ['cd', self.tmp_folder, ';',
212 self.binary_path,
213 '-i', PurePath(self.io_dict["in"]["input_pcz_path"]).name,
214 '-o', temp_out,
215 "-t", "0.3",
216 "--hinge={}".format(self.eigenvector),
217 ">&", temp_log
218 ]
220 # Run Biobb block
221 self.run_biobb()
223 # Parsing output file and extracting results for the given method
224 dict_out = self.parse_output(PurePath(self.tmp_folder).joinpath(temp_out))
226 with open(PurePath(self.tmp_folder).joinpath(temp_json), 'w') as out_file:
227 out_file.write(json.dumps(dict_out, indent=4))
229 # Copy outputs from temporary folder to output path
230 shutil.copy2(PurePath(self.tmp_folder).joinpath(temp_json), PurePath(self.io_dict["out"]["output_json_path"]))
232 # Copy files to host
233 # self.copy_to_host()
235 # remove temporary folder(s)
236 self.tmp_files.extend([
237 # self.stage_io_dict.get("unique_dir", ""),
238 self.tmp_folder
239 ])
240 self.remove_tmp_files()
242 self.check_arguments(output_files_created=True, raise_exception=False)
244 return self.return_code
247def pcz_hinges(input_pcz_path: str, output_json_path: str,
248 properties: Optional[dict] = None, **kwargs) -> int:
249 """Create :class:`PCZhinges <flexserv.pcasuite.pcz_hinges>`flexserv.pcasuite.PCZhinges class and
250 execute :meth:`launch() <flexserv.pcasuite.pcz_hinges.launch>` method"""
252 return PCZhinges(input_pcz_path=input_pcz_path,
253 output_json_path=output_json_path,
254 properties=properties).launch()
256 pcz_hinges.__doc__ = PCZhinges.__doc__
259def main():
260 parser = argparse.ArgumentParser(description='Compute possible hinge regions (residues around which large protein movements are organized) of a molecule from a compressed PCZ file.', formatter_class=lambda prog: argparse.RawTextHelpFormatter(prog, width=99999))
261 parser.add_argument('--config', required=False, help='Configuration file')
263 # Specific args
264 required_args = parser.add_argument_group('required arguments')
265 required_args.add_argument('--input_pcz_path', required=True, help='Input compressed trajectory file. Accepted formats: pcz.')
266 required_args.add_argument('--output_json_path', required=True, help='Output hinge regions x PCA mode file. Accepted formats: json.')
268 args = parser.parse_args()
269 args.config = args.config or "{}"
270 properties = settings.ConfReader(config=args.config).get_prop_dic()
272 # Specific call
273 pcz_hinges(input_pcz_path=args.input_pcz_path,
274 output_json_path=args.output_json_path,
275 properties=properties)
278if __name__ == '__main__':
279 main()