This is git repo for XlinkNet.

Crosslinks dataset in the file: crosslinks.csv

• Change path inside the codes:

  1. in cross_link.py: OBJ_DIR - directory that contains the xl_objects pickle files OBJ_DIR_PREFIX - prefix of OBJ_DIR PROCESSED_OBJ_DIR - directory that contains xl_objects pivkle with distance info PROCESSED_OBJ_DIR_PREFIX - prefix of PROCESSED_OBJ_DIR
  2. in pdb_files_manger.py: PDB_FILES_DIR - directory that will contain pdb files INTER_AND_INTRA_LYS_XL_FILES_PATH - path to directory that needed in data generation process INTER_INTRA_LYS_XL_NEIGHBORS_FILES_PATH - path to directory that needed in data generation process XL_NEIGHBORS_FEATURE_DICT_INTER_INTRA_LYS - path to feature dictionary. needed in data generation process. XL_NEIGHBORS_EXE - path to the xl_neighbors executable (C++ program that extracts the features
  3. in general_utils.py: OBJ_DIR - directory that contains the pickle files PROJ_DIR - the project dir
  4. in alpha_fold_files.py: AF_PDB_DIR - directory for alpha fold pdb files AF_PRED_ERROR_DIR - directory for alpha fold pae files

• Generate the dataset on your own from csv file:

  1. In order to create objects of crosslinks from csv: a. If your csv file is from XlinkNet github / in the same format: If the csv file contains distances (it does in github) run: python3 cross_link.py --run_mode import --input_file_path <path_to_csv> --final_save_name <name_of_res_file> --download_files <True/False> Else: python3 cross_link.py --run_mode import --is_processed False --out_name <path_of_saving_processed> --input_file_path <path_to_csv> --final_save_name <name_of_res_file> --download_files <True/False> b. If your input file is in other format, please check the code in general_xl_parser.py and run: python3 cross_link.py --run_mode parse --parse_cfg <path_to_config_file> --is_processed False --out_name <path_of_saving_processed> --input_file_path <path_to_csv> --final_save_name <name_of_res_file> --download_files <True/False>

  2. Extract features from cross link objects (takes few hours, you can download features / dataset from github) python3 cross_link.py --run_mode extract --final_save_name <name_of_saved_object_list>

  3. Generate pytorch dataset for NN - run: python3 train.py --create_dataset True --xl_objects_file <name_of_saved_object_list> --dataset_name <name_for_the_new_dataset>

• Now that the dataset is set, we can train the model by running the train script along with the config file that we want: python3 train.py --cfg <path_to_config_file>