proteusPy.data

Global declarations for the proteusPy package

 1"""
 2Global declarations for the proteusPy package
 3"""
 4
 5# init for proteusPy data module
 6# Copyright (c) 2023 Eric G. Suchanek, PhD., all rights reserved
 7# Subject to the GNU public license.
 8
 9import os
10
11HOME_DIR = os.path.expanduser("~")
12PDB_BASE = os.getenv("PDB")
13
14if PDB_BASE is None:
15    print(f"The PDB environment variable is not set. Defaulting to {HOME_DIR}.")
16    PDB_BASE = HOME_DIR
17
18# absolute location for the disulfide .pkl files
19_abspath = os.path.dirname(os.path.abspath(__file__))
20
21DATA_DIR = f"{_abspath}/"
22SS_PICKLE_FILE = "PDB_all_ss.pkl"
23SS_DICT_PICKLE_FILE = "PDB_all_ss_dict.pkl"
24SS_DICT_PICKLE_FILE2 = "PDB_all_ss_dict_ind.pkl"
25SS_TORSIONS_FILE = "PDB_all_ss_torsions.csv"
26PROBLEM_ID_FILE = "PDB_all_SS_problems.csv"
27SS_ID_FILE = "ss_ids.txt"
28
29SS_SUBSET_PICKLE_FILE = "PDB_subset_ss.pkl"
30SS_SUBSET_DICT_PICKLE_FILE = "PDB_subset_ss_dict.pkl"
31SS_SUBSET_DICT_PICKLE_FILE_IND = "PDB_subset_ss_dict_ind.pkl"
32SS_SUBSET_TORSIONS_FILE = "PDB_subset_torsions.csv"
33SS_PROBLEM_SUBSET_ID_FILE = "PDB_subset_problems.csv"
34
35# contains the dihedral classes and their members
36SS_CLASS_DICT_FILE = "PDB_ss_classes_dict.pkl"
37SS_CONSENSUS_FILE = "SS_consensus_class_sext.pkl"
38
39LOADER_FNAME = "PDB_SS_ALL_LOADER.pkl"
40LOADER_SUBSET_FNAME = "PDB_SS_SUBSET_LOADER.pkl"
41
42# "https://github.com/suchanek/proteusPy/raw/master/data/PDB_SS_ALL_LOADER.pkl
43LOADER_FNAME_URL = "https://raw.githubusercontent.com/suchanek/proteusPy/master/proteusPy/data/PDB_SS_ALL_LOADER.pkl"
44LOADER_SUBSET_FNAME_URL = "https://raw.githubusercontent.com/suchanek/proteusPy/blob/master/proteusPy/data/PDB_SS_SUBSET_LOADER.pkl"
45
46
47LOADER_ALL_URL = "https://drive.google.com/uc?id=1igF-sppLPaNsBaUS7nkb13vtOGZZmsFp"
48LOADER_SUBSET_URL = "https://drive.google.com/uc?id=1puy9pxrClFks0KN9q5PPV_ONKvL-hg33"
49
50SS_CLASS_DEFINITIONS = """
51Idx,chi1_s,chi2_s,chi3_s,chi4_s,chi5_s,class_id,SS_Classname,FXN
520,-1,-1,-1,-1,-1,00000,-LHSpiral,UNK
531,-1,-1,-1,-1,1,00002,00002,UNK
542,-1,-1,-1,1,-1,00020,-LHHook,UNK
553,-1,-1,-1,1,1,00022,00022,UNK
564,-1,-1,1,-1,-1,00200,-RHStaple,Allosteric
575,-1,-1,1,-1,1,00202,00202,UNK
586,-1,-1,1,1,-1,00220,00220,UNK
597,-1,-1,1,1,1,00222,00222,UNK
608,-1,1,-1,-1,-1,02000,02000,UNK
619,-1,1,-1,-1,1,02002,02002,UNK
6210,-1,1,-1,1,-1,02020,-LHStaple,UNK
6311,-1,1,-1,1,1,02022,02022,UNK
6412,-1,1,1,-1,-1,02200,-RHHook,UNK
6513,-1,1,1,-1,1,02202,02202,UNK
6614,-1,1,1,1,-1,02220,-RHSpiral,UNK
6715,-1,1,1,1,1,02222,02222,UNK
6816,1,-1,-1,-1,-1,20000,±LHSpiral,UNK
6917,1,-1,-1,-1,1,20002,+LHSpiral,UNK
7018,1,-1,-1,1,-1,20020,±LHHook,UNK
7119,1,-1,-1,1,1,20022,+LHHook,UNK
7220,1,-1,1,-1,-1,20200,±RHStaple,UNK
7321,1,-1,1,-1,1,20202,+RHStaple,UNK
7422,1,-1,1,1,-1,20220,±RHHook,Catalytic
7523,1,-1,1,1,1,20222,20222,UNK
7624,1,1,-1,-1,-1,22000,-/+LHHook,UNK
7725,1,1,-1,-1,1,22002,22002,UNK
7826,1,1,-1,1,-1,22020,±LHStaple,UNK
7927,1,1,-1,1,1,22022,+LHStaple,UNK
8028,1,1,1,-1,-1,22200,-/+RHHook,UNK
8129,1,1,1,-1,1,22202,+RHHook,UNK
8230,1,1,1,1,-1,22220,±RHSpiral,UNK
8331,1,1,1,1,1,22222,+RHSpiral,UNK
84"""
85
86CLASSOBJ_FNAME = "PDB_CLASS_OBJ.pkl"
87# end of file
HOME_DIR = $HOME
PDB_BASE = $PDB
DATA_DIR = '/Users/egs/repos/proteusPy/proteusPy/data/'
SS_PICKLE_FILE = 'PDB_all_ss.pkl'
SS_DICT_PICKLE_FILE = 'PDB_all_ss_dict.pkl'
SS_DICT_PICKLE_FILE2 = 'PDB_all_ss_dict_ind.pkl'
SS_TORSIONS_FILE = 'PDB_all_ss_torsions.csv'
PROBLEM_ID_FILE = 'PDB_all_SS_problems.csv'
SS_ID_FILE = 'ss_ids.txt'
SS_SUBSET_PICKLE_FILE = 'PDB_subset_ss.pkl'
SS_SUBSET_DICT_PICKLE_FILE = 'PDB_subset_ss_dict.pkl'
SS_SUBSET_DICT_PICKLE_FILE_IND = 'PDB_subset_ss_dict_ind.pkl'
SS_SUBSET_TORSIONS_FILE = 'PDB_subset_torsions.csv'
SS_PROBLEM_SUBSET_ID_FILE = 'PDB_subset_problems.csv'
SS_CLASS_DICT_FILE = 'PDB_ss_classes_dict.pkl'
SS_CONSENSUS_FILE = 'SS_consensus_class_sext.pkl'
LOADER_FNAME = 'PDB_SS_ALL_LOADER.pkl'
LOADER_SUBSET_FNAME = 'PDB_SS_SUBSET_LOADER.pkl'
LOADER_FNAME_URL = 'https://raw.githubusercontent.com/suchanek/proteusPy/master/proteusPy/data/PDB_SS_ALL_LOADER.pkl'
LOADER_SUBSET_FNAME_URL = 'https://raw.githubusercontent.com/suchanek/proteusPy/blob/master/proteusPy/data/PDB_SS_SUBSET_LOADER.pkl'
LOADER_ALL_URL = 'https://drive.google.com/uc?id=1igF-sppLPaNsBaUS7nkb13vtOGZZmsFp'
LOADER_SUBSET_URL = 'https://drive.google.com/uc?id=1puy9pxrClFks0KN9q5PPV_ONKvL-hg33'
SS_CLASS_DEFINITIONS = '\nIdx,chi1_s,chi2_s,chi3_s,chi4_s,chi5_s,class_id,SS_Classname,FXN\n0,-1,-1,-1,-1,-1,00000,-LHSpiral,UNK\n1,-1,-1,-1,-1,1,00002,00002,UNK\n2,-1,-1,-1,1,-1,00020,-LHHook,UNK\n3,-1,-1,-1,1,1,00022,00022,UNK\n4,-1,-1,1,-1,-1,00200,-RHStaple,Allosteric\n5,-1,-1,1,-1,1,00202,00202,UNK\n6,-1,-1,1,1,-1,00220,00220,UNK\n7,-1,-1,1,1,1,00222,00222,UNK\n8,-1,1,-1,-1,-1,02000,02000,UNK\n9,-1,1,-1,-1,1,02002,02002,UNK\n10,-1,1,-1,1,-1,02020,-LHStaple,UNK\n11,-1,1,-1,1,1,02022,02022,UNK\n12,-1,1,1,-1,-1,02200,-RHHook,UNK\n13,-1,1,1,-1,1,02202,02202,UNK\n14,-1,1,1,1,-1,02220,-RHSpiral,UNK\n15,-1,1,1,1,1,02222,02222,UNK\n16,1,-1,-1,-1,-1,20000,±LHSpiral,UNK\n17,1,-1,-1,-1,1,20002,+LHSpiral,UNK\n18,1,-1,-1,1,-1,20020,±LHHook,UNK\n19,1,-1,-1,1,1,20022,+LHHook,UNK\n20,1,-1,1,-1,-1,20200,±RHStaple,UNK\n21,1,-1,1,-1,1,20202,+RHStaple,UNK\n22,1,-1,1,1,-1,20220,±RHHook,Catalytic\n23,1,-1,1,1,1,20222,20222,UNK\n24,1,1,-1,-1,-1,22000,-/+LHHook,UNK\n25,1,1,-1,-1,1,22002,22002,UNK\n26,1,1,-1,1,-1,22020,±LHStaple,UNK\n27,1,1,-1,1,1,22022,+LHStaple,UNK\n28,1,1,1,-1,-1,22200,-/+RHHook,UNK\n29,1,1,1,-1,1,22202,+RHHook,UNK\n30,1,1,1,1,-1,22220,±RHSpiral,UNK\n31,1,1,1,1,1,22222,+RHSpiral,UNK\n'
CLASSOBJ_FNAME = 'PDB_CLASS_OBJ.pkl'