paths_and_settings.py

from os.path import join
from os import getcwd
from utils import create_folder_if_not_existent
import time

# IF LOGGING SET TO TRUE, COMMUNICATES WILL BE REDIRECTED TO A FILE, OTHERWISE THEY WILL BE PRINTED IN CONSOLE

# MAIN PATH FROM WHICH THE PIPELINE WILL BE WORKING
PROJECT_HOME = getcwd()
MAIN_CSV_NAME = join(PROJECT_HOME, 'master_table.csv')

LOGGING = False
LOG_FILE = ''
if LOGGING:
    t = time.localtime()
    timestamp = time.strftime('%b-%d-%Y', t)
    LOG_FILE = join(PROJECT_HOME, f'log_{timestamp}.txt')
    with open(LOG_FILE, 'w') as _:
        pass

# REQUIRED INPUT DATA
RAW_DATA_FOLDER = join(PROJECT_HOME, 'raw_data')
RAW_TARGETS = join(RAW_DATA_FOLDER, 'CHEMBL25-targets.csv')                 # CSV from with targets will be taken
RAW_IC50_ACTIVITY = join(RAW_DATA_FOLDER, 'CHEMBL25-single_ic50.csv')       # CSV with IC50 activities for targets
RAW_KI_ACTIVITY = join(RAW_DATA_FOLDER, 'CHEMBL25-single_ki.csv')           # CSV with Ki activities for targets
RAW_KD_ACTVITY = join(RAW_DATA_FOLDER, 'CHEMBL25-single_kd.csv')            # CSV with Kd activities for targets
RAW_UNIPROT_3D_IDS = join(RAW_DATA_FOLDER, 'has_3d_pdb_acc.csv')            # file with uniprot ID's for which any X-ray PDB structure exists

DEKOIS_PATH = join(PROJECT_HOME, 'DEKOIS2.0_library')                       # path to folder containing DEKOIS database
DUDE_PATH = join(PROJECT_HOME, 'DUDE')                                      # path to folder containing DUD-E database


# OUTPUTS DURING PIPELINE
STANDARDS_FOLDER = join(PROJECT_HOME, 'standards_csv')                      # processed activities CSV's will be stored here
create_folder_if_not_existent(STANDARDS_FOLDER)

COMPOUND_SAMPLING_FOLDER = join(PROJECT_HOME, 'actives_number_sampling')    # folder in which results of filtering by actives number are stored
create_folder_if_not_existent(COMPOUND_SAMPLING_FOLDER)

BLAST_MAIN_FOLDER = join(PROJECT_HOME, 'blast_similarities')
create_folder_if_not_existent(BLAST_MAIN_FOLDER)

CHEMBL_SMILES_FOLDER = join(PROJECT_HOME, 'chembl_smiles')
create_folder_if_not_existent(CHEMBL_SMILES_FOLDER)

SIMILARITY_MATRICES_FOLDER = join(PROJECT_HOME, 'similarity_matrices')
create_folder_if_not_existent(SIMILARITY_MATRICES_FOLDER)

# FILTERING ARGUMENTS
INITIAL_FILTER = False
ACTIVITY_VALUE_THRESHOLD = 10  # in nM

SAMPLING_FILTER = False
PDB_LIGANDS_WEIGHT_LOWER_THRESHOLD = 100  # int
LOWEST_TC_SIMILARITY_BETWEEN_LIGANDS_THRESHOLD = (0.95,)    # a tuple
LOWER_LIMIT_OF_LIGANDS = (50,)  # a tuple

# BLAST ARGUMENTS
BLAST = False
CHOSEN_TC_THRESHOLD = 0.95  # float
CHOSEN_LIGAND_LIMIT = 50    # int
E_VALUE_THRESHOLD = 0.00001   # float
MAX_BLAST_SIMILARITY = 30  # sequences % similarity threshold between targets and best hits from DEKOIS/DUD-E

# CRATING SIMILARITY MATRICES
CREATE_ACTIVES_SIMILARITY_MATRICES = False

# REDUCING ANALOGUE BIAS AMONG ACTIVES
FILTER_ACTIVES = False
ACTIVES_TC_SIMILARITY_THRESHOLD = 0.9  # maximum Tanimoto similarity between compounds in active set
USE_JOBLIB = False
JOBLIB_WORKERS = 4


# DECOY SEARCHING
# Legend:
# 'HBD'           : max_difference,
# 'HBA'           : max_difference,
# 'rotates'       : max_difference,
# 'weight'        : max_query_weight_%_difference,
# 'logp'          : max_query_logp_%_difference,
# 'murco_scaffold': max_similarity_value
THRESHOLDS_DICT = {'HBD': 0, 'HBA': 0, 'rotates': 0, 'weight': 0.15, 'logp': 0.2, 'murco_scaffold': 0.7}  # sample values

SOFT_LOGP_THRESHOLDS = True
SOFT_LOGP_PARAMETER = 4
# if SOFT_LOGP_THRESHOLDS IS True
# logp acceptation is calculated by abs(difference)-(1-(abs(chembl_ligand_logp)/SOFT_LOGP_PARAMETER)), and if the result is <0 it is considered ok,
# so for ligands with very low logp any hits were still possible and it narrowed possible matches for ligands with very high logp
# ligands with logp lower between -SOFT_LOGP_PARAMETER, to SOFT_LOGP_PARAMETER will be benefiting from this rule (broadening the acceptance interval),
# and <-SOFT_LOGP_PARAMETER and >SOFT_LOGP_PARAMETER will be penalized (narrowing down their acceptance interval)
# soft threshold does not influence the logp difference value saved to a file, it is applied only during comparing difference with threshold


# POST-PROCESS FOUND DECOYS
FILTER_FOUND_DECOYS = False
FOUND_DECOYS_FOLDER = join(PROJECT_HOME, 'found_decoys')
FOUND_DECOYS_CSV = join(FOUND_DECOYS_FOLDER, 'all_decoys_found.csv')
DECOYS_PER_LIGAND = 100