evaluate.py

import copy
import os
import torch
from datasets.moad import MOAD
from utils.gnina_utils import get_gnina_poses
from utils.molecules_utils import get_symmetry_rmsd

torch.multiprocessing.set_sharing_strategy('file_system')

import resource
rlimit = resource.getrlimit(resource.RLIMIT_NOFILE)
resource.setrlimit(resource.RLIMIT_NOFILE, (64000, rlimit[1]))

import time
from argparse import ArgumentParser, Namespace, FileType
from datetime import datetime
from functools import partial
import numpy as np
import wandb
from rdkit import RDLogger
from torch_geometric.loader import DataLoader
from rdkit.Chem import RemoveAllHs

from datasets.pdbbind import PDBBind
from utils.diffusion_utils import t_to_sigma as t_to_sigma_compl, get_t_schedule
from utils.sampling import randomize_position, sampling
from utils.utils import get_model, ExponentialMovingAverage
from utils.visualise import PDBFile
from tqdm import tqdm

RDLogger.DisableLog('rdApp.*')
import yaml
import pickle


def get_dataset(args, model_args, confidence=False):
    if args.dataset != 'moad':
        dataset = PDBBind(transform=None, root=args.data_dir, limit_complexes=args.limit_complexes, dataset=args.dataset,
                        chain_cutoff=args.chain_cutoff,
                        receptor_radius=model_args.receptor_radius,
                        cache_path=args.cache_path, split_path=args.split_path,
                        remove_hs=model_args.remove_hs, max_lig_size=None,
                        c_alpha_max_neighbors=model_args.c_alpha_max_neighbors,
                        matching=not model_args.no_torsion, keep_original=True,
                        popsize=args.matching_popsize,
                        maxiter=args.matching_maxiter,
                        all_atoms=model_args.all_atoms if 'all_atoms' in model_args else False,
                        atom_radius=model_args.atom_radius if 'all_atoms' in model_args else None,
                        atom_max_neighbors=model_args.atom_max_neighbors if 'all_atoms' in model_args else None,
                        esm_embeddings_path=args.esm_embeddings_path,
                        require_ligand=True,
                        num_workers=args.num_workers,
                        protein_file=args.protein_file,
                        ligand_file=args.ligand_file,
                        knn_only_graph=True if not hasattr(args, 'not_knn_only_graph') else not args.not_knn_only_graph,
                        include_miscellaneous_atoms=False if not hasattr(args,
                                                                         'include_miscellaneous_atoms') else args.include_miscellaneous_atoms,
                        num_conformers=args.samples_per_complex if args.resample_rdkit and not confidence else 1)

    else:
        dataset = MOAD(transform=None, root=args.data_dir, limit_complexes=args.limit_complexes,
                       chain_cutoff=args.chain_cutoff,
                       receptor_radius=model_args.receptor_radius,
                       cache_path=args.cache_path, split=args.split,
                       remove_hs=model_args.remove_hs, max_lig_size=None,
                       c_alpha_max_neighbors=model_args.c_alpha_max_neighbors,
                       matching=not model_args.no_torsion, keep_original=True,
                       popsize=args.matching_popsize,
                       maxiter=args.matching_maxiter,
                       all_atoms=model_args.all_atoms if 'all_atoms' in model_args else False,
                       atom_radius=model_args.atom_radius if 'all_atoms' in model_args else None,
                       atom_max_neighbors=model_args.atom_max_neighbors if 'all_atoms' in model_args else None,
                       esm_embeddings_path=args.esm_embeddings_path,
                       esm_embeddings_sequences_path=args.moad_esm_embeddings_sequences_path,
                       require_ligand=True,
                       num_workers=args.num_workers,
                       knn_only_graph=True if not hasattr(args, 'not_knn_only_graph') else not args.not_knn_only_graph,
                       include_miscellaneous_atoms=False if not hasattr(args,
                                                                        'include_miscellaneous_atoms') else args.include_miscellaneous_atoms,
                       num_conformers=args.samples_per_complex if args.resample_rdkit and not confidence else 1,
                       unroll_clusters=args.unroll_clusters, remove_pdbbind=args.remove_pdbbind,
                       min_ligand_size=args.min_ligand_size,
                       max_receptor_size=args.max_receptor_size,
                       remove_promiscuous_targets=args.remove_promiscuous_targets,
                       no_randomness=True,
                       skip_matching=args.skip_matching)
    return dataset



if __name__ == '__main__':
    cache_name = datetime.now().strftime('date%d-%m_time%H-%M-%S.%f')
    parser = ArgumentParser()
    parser.add_argument('--config', type=FileType(mode='r'), default=None)
    parser.add_argument('--model_dir', type=str, default='workdir/test_score', help='Path to folder with trained score model and hyperparameters')
    parser.add_argument('--ckpt', type=str, default='best_ema_inference_epoch_model.pt', help='Checkpoint to use inside the folder')
    parser.add_argument('--confidence_model_dir', type=str, default=None, help='Path to folder with trained confidence model and hyperparameters')
    parser.add_argument('--confidence_ckpt', type=str, default='best_model_epoch75.pt', help='Checkpoint to use inside the folder')
    parser.add_argument('--num_cpu', type=int, default=None, help='if this is a number instead of none, the max number of cpus used by torch will be set to this.')
    parser.add_argument('--run_name', type=str, default='test', help='')
    parser.add_argument('--project', type=str, default='ligbind_inf', help='')
    parser.add_argument('--out_dir', type=str, default=None, help='Where to save results to')
    parser.add_argument('--batch_size', type=int, default=40, help='Number of poses to sample in parallel')

    parser.add_argument('--old_score_model', action='store_true', default=False, help='')
    parser.add_argument('--old_confidence_model', action='store_true', default=True, help='')
    parser.add_argument('--matching_popsize', type=int, default=40, help='Differential evolution popsize parameter in matching')
    parser.add_argument('--matching_maxiter', type=int, default=40, help='Differential evolution maxiter parameter in matching')

    parser.add_argument('--esm_embeddings_path', type=str, default=None, help='If this is set then the LM embeddings at that path will be used for the receptor features')
    parser.add_argument('--moad_esm_embeddings_sequences_path', type=str, default=None, help='')
    parser.add_argument('--chain_cutoff', type=float, default=None, help='Cutoff of the chains from the ligand') # TODO remove
    parser.add_argument('--save_complexes', action='store_true', default=False, help='Save generated complex graphs')
    parser.add_argument('--complexes_save_path', type=str, default=None, help='')

    parser.add_argument('--dataset', type=str, default='moad', help='')
    parser.add_argument('--cache_path', type=str, default='data/cache', help='Folder from where to load/restore cached dataset')
    parser.add_argument('--data_dir', type=str, default='../../ligbind/data/BindingMOAD_2020_ab_processed_biounit/', help='Folder containing original structures')
    parser.add_argument('--split_path', type=str, default='data/BindingMOAD_2020_ab_processed/splits/val.txt', help='Path of file defining the split')

    parser.add_argument('--no_model', action='store_true', default=False, help='Whether to return seed conformer without running model')
    parser.add_argument('--no_random', action='store_true', default=False, help='Whether to add randomness in diffusion steps')
    parser.add_argument('--no_final_step_noise', action='store_true', default=False, help='Whether to add noise after the final step')
    parser.add_argument('--ode', action='store_true', default=False, help='Whether to run the probability flow ODE')
    parser.add_argument('--wandb', action='store_true', default=False, help='') # TODO remove
    parser.add_argument('--inference_steps', type=int, default=40, help='Number of denoising steps')
    parser.add_argument('--limit_complexes', type=int, default=0, help='Limit to the number of complexes')
    parser.add_argument('--num_workers', type=int, default=1, help='Number of workers for dataset creation')
    parser.add_argument('--tqdm', action='store_true', default=False, help='Whether to show progress bar')
    parser.add_argument('--save_visualisation', action='store_true', default=True, help='Whether to save visualizations')
    parser.add_argument('--samples_per_complex', type=int, default=4, help='Number of poses to sample for each complex')
    parser.add_argument('--resample_rdkit', action='store_true', default=False, help='')
    parser.add_argument('--skip_matching', action='store_true', default=False, help='')
    parser.add_argument('--sigma_schedule', type=str, default='expbeta', help='Schedule type, no other options')
    parser.add_argument('--inf_sched_alpha', type=float, default=1, help='Alpha parameter of beta distribution for t sched')
    parser.add_argument('--inf_sched_beta', type=float, default=1, help='Beta parameter of beta distribution for t sched')
    parser.add_argument('--pocket_knowledge', action='store_true', default=False, help='')
    parser.add_argument('--no_random_pocket', action='store_true', default=False, help='')
    parser.add_argument('--pocket_tr_max', type=float, default=3, help='')
    parser.add_argument('--pocket_cutoff', type=float, default=5, help='')
    parser.add_argument('--actual_steps', type=int, default=None, help='')
    parser.add_argument('--restrict_cpu', action='store_true', default=False, help='')
    parser.add_argument('--force_fixed_center_conv', action='store_true', default=False, help='')
    parser.add_argument('--protein_file', type=str, default='protein_processed', help='')
    parser.add_argument('--unroll_clusters', action='store_true', default=True, help='')
    parser.add_argument('--ligand_file', type=str, default='ligand', help='')
    parser.add_argument('--remove_pdbbind', action='store_true', default=False, help='')
    parser.add_argument('--split', type=str, default='val', help='')
    parser.add_argument('--limit_failures', type=float, default=5, help='')
    parser.add_argument('--min_ligand_size', type=float, default=0, help='')
    parser.add_argument('--max_receptor_size', type=float, default=None, help='')
    parser.add_argument('--remove_promiscuous_targets', type=float, default=None, help='')
    parser.add_argument('--initial_noise_std_proportion', type=float, default=-1.0, help='Initial noise std proportion')
    parser.add_argument('--choose_residue', action='store_true', default=False, help='')

    parser.add_argument('--temp_sampling_tr', type=float, default=1.0)
    parser.add_argument('--temp_psi_tr', type=float, default=0.0)
    parser.add_argument('--temp_sigma_data_tr', type=float, default=0.5)
    parser.add_argument('--temp_sampling_rot', type=float, default=1.0)
    parser.add_argument('--temp_psi_rot', type=float, default=0.0)
    parser.add_argument('--temp_sigma_data_rot', type=float, default=0.5)
    parser.add_argument('--temp_sampling_tor', type=float, default=1.0)
    parser.add_argument('--temp_psi_tor', type=float, default=0.0)
    parser.add_argument('--temp_sigma_data_tor', type=float, default=0.5)

    parser.add_argument('--gnina_minimize', action='store_true', default=False, help='')
    parser.add_argument('--gnina_path', type=str, default='gnina', help='')
    parser.add_argument('--gnina_log_file', type=str, default='gnina_log.txt', help='') # To redirect gnina subprocesses stdouts from the terminal window
    parser.add_argument('--gnina_full_dock', action='store_true', default=False, help='')
    parser.add_argument('--save_gnina_metrics', action='store_true', default=False, help='')
    parser.add_argument('--gnina_autobox_add', type=float, default=4.0)
    parser.add_argument('--gnina_poses_to_optimize', type=int, default=1)

    args = parser.parse_args()

    if args.config:
        config_dict = yaml.load(args.config, Loader=yaml.FullLoader)
        arg_dict = args.__dict__
        for key, value in config_dict.items():
            if isinstance(value, list):
                for v in value:
                    arg_dict[key].append(v)
            else:
                arg_dict[key] = value

    if args.restrict_cpu:
        threads = 16
        os.environ["OMP_NUM_THREADS"] = str(threads)  # export OMP_NUM_THREADS=4
        os.environ["OPENBLAS_NUM_THREADS"] = str(threads)  # export OPENBLAS_NUM_THREADS=4
        os.environ["MKL_NUM_THREADS"] = str(threads)  # export MKL_NUM_THREADS=6
        os.environ["VECLIB_MAXIMUM_THREADS"] = str(threads)  # export VECLIB_MAXIMUM_THREADS=4
        os.environ["NUMEXPR_NUM_THREADS"] = str(threads)  # export NUMEXPR_NUM_THREADS=6
        os.environ["CUDA_VISIBLE_DEVICES"] = ""

        torch.set_num_threads(threads)

    if args.out_dir is None: args.out_dir = f'inference_out_dir_not_specified/{args.run_name}'
    os.makedirs(args.out_dir, exist_ok=True)
    with open(f'{args.model_dir}/model_parameters.yml') as f:
        score_model_args = Namespace(**yaml.full_load(f))
        if not hasattr(score_model_args, 'separate_noise_schedule'):  # exists for compatibility with old runs that did not have the attribute
            score_model_args.separate_noise_schedule = False
        if not hasattr(score_model_args, 'lm_embeddings_path'):
            score_model_args.lm_embeddings_path = None
        if not hasattr(score_model_args, 'tr_only_confidence'):
            score_model_args.tr_only_confidence = True
        if not hasattr(score_model_args, 'high_confidence_threshold'):
            score_model_args.high_confidence_threshold = 0.0
        if not hasattr(score_model_args, 'include_confidence_prediction'):
            score_model_args.include_confidence_prediction = False
        if not hasattr(score_model_args, 'confidence_weight'):
            score_model_args.confidence_weight = 1
        if not hasattr(score_model_args, 'asyncronous_noise_schedule'):
            score_model_args.asyncronous_noise_schedule = False
        if not hasattr(score_model_args, 'correct_torsion_sigmas'):
            score_model_args.correct_torsion_sigmas = False
        if not hasattr(score_model_args, 'esm_embeddings_path'):
            score_model_args.esm_embeddings_path = None
        if args.force_fixed_center_conv:
            score_model_args.not_fixed_center_conv = False
    if args.confidence_model_dir is not None:
        with open(f'{args.confidence_model_dir}/model_parameters.yml') as f:
            confidence_args = Namespace(**yaml.full_load(f))
        if not os.path.exists(confidence_args.original_model_dir):
            print("Path does not exist: ", confidence_args.original_model_dir)
            confidence_args.original_model_dir = os.path.join(*confidence_args.original_model_dir.split('/')[-2:])
            print('instead trying path: ', confidence_args.original_model_dir)
        if not hasattr(confidence_args, 'use_original_model_cache'):
            confidence_args.use_original_model_cache = True
        if not hasattr(confidence_args, 'esm_embeddings_path'):
            confidence_args.esm_embeddings_path = None
        if not hasattr(confidence_args, 'num_classification_bins'):
            confidence_args.num_classification_bins = 2

    if args.num_cpu is not None:
        torch.set_num_threads(args.num_cpu)
    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
    test_dataset = get_dataset(args, score_model_args)
    test_loader = DataLoader(dataset=test_dataset, batch_size=1, shuffle=False)
    if args.confidence_model_dir is not None:
        if not (confidence_args.use_original_model_cache or confidence_args.transfer_weights):
            # if the confidence model uses the same type of data as the original model then we do not need this dataset and can just use the complexes
            print('HAPPENING | confidence model uses different type of graphs than the score model. Loading (or creating if not existing) the data for the confidence model now.')
            confidence_test_dataset = get_dataset(args, confidence_args, confidence=True)
            confidence_complex_dict = {d.name: d for d in confidence_test_dataset}

    t_to_sigma = partial(t_to_sigma_compl, args=score_model_args)

    if not args.no_model:
        model = get_model(score_model_args, device, t_to_sigma=t_to_sigma, no_parallel=True, old=args.old_score_model)
        state_dict = torch.load(f'{args.model_dir}/{args.ckpt}', map_location=torch.device('cpu'))
        if args.ckpt == 'last_model.pt':
            model_state_dict = state_dict['model']
            ema_weights_state = state_dict['ema_weights']
            model.load_state_dict(model_state_dict, strict=True)
            ema_weights = ExponentialMovingAverage(model.parameters(), decay=score_model_args.ema_rate)
            ema_weights.load_state_dict(ema_weights_state, device=device)
            ema_weights.copy_to(model.parameters())
        else:
            model.load_state_dict(state_dict, strict=True)
            model = model.to(device)
            model.eval()
        if args.confidence_model_dir is not None:
            if confidence_args.transfer_weights:
                with open(f'{confidence_args.original_model_dir}/model_parameters.yml') as f:
                    confidence_model_args = Namespace(**yaml.full_load(f))
                if not hasattr(confidence_model_args, 'separate_noise_schedule'):  # exists for compatibility with old runs that did not have the
                    # attribute
                    confidence_model_args.separate_noise_schedule = False
                if not hasattr(confidence_model_args, 'lm_embeddings_path'):
                    confidence_model_args.lm_embeddings_path = None
                if not hasattr(confidence_model_args, 'tr_only_confidence'):
                    confidence_model_args.tr_only_confidence = True
                if not hasattr(confidence_model_args, 'high_confidence_threshold'):
                    confidence_model_args.high_confidence_threshold = 0.0
                if not hasattr(confidence_model_args, 'include_confidence_prediction'):
                    confidence_model_args.include_confidence_prediction = False
                if not hasattr(confidence_model_args, 'confidence_dropout'):
                    confidence_model_args.confidence_dropout = confidence_model_args.dropout
                if not hasattr(confidence_model_args, 'confidence_no_batchnorm'):
                    confidence_model_args.confidence_no_batchnorm = False
                if not hasattr(confidence_model_args, 'confidence_weight'):
                    confidence_model_args.confidence_weight = 1
                if not hasattr(confidence_model_args, 'asyncronous_noise_schedule'):
                    confidence_model_args.asyncronous_noise_schedule = False
                if not hasattr(confidence_model_args, 'correct_torsion_sigmas'):
                    confidence_model_args.correct_torsion_sigmas = False
                if not hasattr(confidence_model_args, 'esm_embeddings_path'):
                    confidence_model_args.esm_embeddings_path = None
                if not hasattr(confidence_model_args, 'not_fixed_knn_radius_graph'):
                    confidence_model_args.not_fixed_knn_radius_graph = True
                if not hasattr(confidence_model_args, 'not_knn_only_graph'):
                    confidence_model_args.not_knn_only_graph = True
            else:
                confidence_model_args = confidence_args

            confidence_model = get_model(confidence_model_args, device, t_to_sigma=t_to_sigma, no_parallel=True,
                                        confidence_mode=True, old=args.old_confidence_model)
            state_dict = torch.load(f'{args.confidence_model_dir}/{args.confidence_ckpt}', map_location=torch.device('cpu'))
            confidence_model.load_state_dict(state_dict, strict=True)
            confidence_model = confidence_model.to(device)
            confidence_model.eval()
        else:
            confidence_model = None
            confidence_args = None
            confidence_model_args = None

    if args.wandb:
        run = wandb.init(
            entity='',
            settings=wandb.Settings(start_method="fork"),
            project=args.project,
            name=args.run_name,
            config=args
        )

    if args.pocket_knowledge and args.different_schedules:
        t_max = (np.log(args.pocket_tr_max) - np.log(score_model_args.tr_sigma_min)) / (
                    np.log(score_model_args.tr_sigma_max) - np.log(score_model_args.tr_sigma_min))
    else:
        t_max = 1

    tr_schedule = get_t_schedule(sigma_schedule=args.sigma_schedule, inference_steps=args.inference_steps,
                                 inf_sched_alpha=args.inf_sched_alpha, inf_sched_beta=args.inf_sched_beta,
                                 t_max=t_max)
    t_schedule = None
    rot_schedule = tr_schedule
    tor_schedule = tr_schedule
    print('common t schedule', tr_schedule)

    rmsds_list, obrmsds, centroid_distances_list, failures, skipped, min_cross_distances_list, base_min_cross_distances_list, confidences_list, names_list = [], [], [], 0, 0, [], [], [], []
    run_times, min_self_distances_list, without_rec_overlap_list = [], [], []
    gnina_rmsds_list, gnina_score_list = [], []
    N = args.samples_per_complex
    #names_no_rec_overlap = read_strings_from_txt(f'data/splits/timesplit_test_no_rec_overlap')
    #names_no_rec_overlap = np.load("data/BindingMOAD_2020_processed/test_names_bootstrapping.npy")
    names_no_rec_overlap = []
    print('Size of test dataset: ', len(test_dataset))

    if args.save_complexes:
        sampled_complexes = {}

    if args.save_gnina_metrics:
        # key is complex_name, value is the gnina metrics for all samples
        gnina_metrics = {}

    for idx, orig_complex_graph in tqdm(enumerate(test_loader)):
        torch.cuda.empty_cache()

        if confidence_model is not None and not (confidence_args.use_original_model_cache or confidence_args.transfer_weights) \
                and orig_complex_graph.name[0] not in confidence_complex_dict.keys():
            skipped += 1
            print(f"HAPPENING | The confidence dataset did not contain {orig_complex_graph.name[0]}. We are skipping this complex.")
            continue
        success = 0
        bs = args.batch_size
        while 0 >= success > -args.limit_failures:
            try:
                data_list = [copy.deepcopy(orig_complex_graph) for _ in range(N)]
                if args.resample_rdkit:
                    for i, g in enumerate(data_list):
                        g['ligand'].pos = g['ligand'].pos[i]

                randomize_position(data_list, score_model_args.no_torsion, args.no_random or args.no_random_pocket,
                                   score_model_args.tr_sigma_max if not args.pocket_knowledge else args.pocket_tr_max,
                                   args.pocket_knowledge, args.pocket_cutoff,
                                   initial_noise_std_proportion=args.initial_noise_std_proportion,
                                   choose_residue=args.choose_residue)


                pdb = None
                if args.save_visualisation:
                    visualization_list = []
                    for idx, graph in enumerate(data_list):
                        lig = orig_complex_graph.mol[0]
                        pdb = PDBFile(lig)
                        pdb.add(lig, 0, 0)
                        pdb.add(((orig_complex_graph['ligand'].pos if not args.resample_rdkit else orig_complex_graph['ligand'].pos[idx]) + orig_complex_graph.original_center).detach().cpu(), 1, 0)
                        pdb.add((graph['ligand'].pos + graph.original_center).detach().cpu(), part=1, order=1)
                        visualization_list.append(pdb)
                else:
                    visualization_list = None

                start_time = time.time()
                if not args.no_model:
                    if confidence_model is not None and not (
                            confidence_args.use_original_model_cache or confidence_args.transfer_weights):
                        confidence_data_list = [copy.deepcopy(confidence_complex_dict[orig_complex_graph.name[0]]) for _ in
                                               range(N)]
                    else:
                        confidence_data_list = None

                    data_list, confidence = sampling(data_list=data_list, model=model,
                                                     inference_steps=args.actual_steps if args.actual_steps is not None else args.inference_steps,
                                                     tr_schedule=tr_schedule, rot_schedule=rot_schedule,
                                                     tor_schedule=tor_schedule,
                                                     device=device, t_to_sigma=t_to_sigma, model_args=score_model_args,
                                                     no_random=args.no_random,
                                                     ode=args.ode, visualization_list=visualization_list,
                                                     confidence_model=confidence_model,
                                                     confidence_data_list=confidence_data_list,
                                                     confidence_model_args=confidence_model_args,
                                                     t_schedule=t_schedule,
                                                     batch_size=bs,
                                                     no_final_step_noise=args.no_final_step_noise, pivot=None,
                                                     temp_sampling=[args.temp_sampling_tr, args.temp_sampling_rot, args.temp_sampling_tor],
                                                     temp_psi=[args.temp_psi_tr, args.temp_psi_rot, args.temp_psi_tor],
                                                     temp_sigma_data=[args.temp_sigma_data_tr, args.temp_sigma_data_rot, args.temp_sigma_data_tor])

                run_times.append(time.time() - start_time)
                if score_model_args.no_torsion:
                    orig_complex_graph['ligand'].orig_pos = (orig_complex_graph['ligand'].pos.cpu().numpy() + orig_complex_graph.original_center.cpu().numpy())

                filterHs = torch.not_equal(data_list[0]['ligand'].x[:, 0], 0).cpu().numpy()

                if isinstance(orig_complex_graph['ligand'].orig_pos, list):
                    # Same pair with multiple binding positions
                    # print(f'Number of ground truth poses: {len(orig_complex_graph['ligand'].orig_pos)}')
                    if args.dataset == 'moad' or args.dataset == 'posebusters':
                        orig_ligand_pos = np.array([pos[filterHs] - orig_complex_graph.original_center.cpu().numpy() for pos in orig_complex_graph['ligand'].orig_pos[0]])
                    else:
                        orig_ligand_pos = np.array([pos[filterHs] - orig_complex_graph.original_center.cpu().numpy() for pos in [orig_complex_graph['ligand'].orig_pos[0]]])
                    print('Found ', len(orig_ligand_pos), ' ground truth poses')
                else:
                    print('default path')
                    orig_ligand_pos = np.expand_dims(
                        orig_complex_graph['ligand'].orig_pos[filterHs] - orig_complex_graph.original_center.cpu().numpy(),
                        axis=0)

                ligand_pos = np.asarray(
                        [complex_graph['ligand'].pos.cpu().numpy()[filterHs] for complex_graph in data_list])

                # Use gnina to minimize energy for predicted ligands.
                if args.gnina_minimize:
                    print('Running gnina on all predicted ligand positions for energy minimization.')
                    gnina_rmsds, gnina_scores = [], []
                    lig = copy.deepcopy(orig_complex_graph.mol[0])
                    positions = np.asarray([complex_graph['ligand'].pos.cpu().numpy() for complex_graph in data_list])

                    conf = confidence
                    if conf is not None and isinstance(confidence_args.rmsd_classification_cutoff, list):
                        conf = conf[:, 0]
                    if conf is not None:
                        conf = conf.cpu().numpy()
                        conf = np.nan_to_num(conf, nan=-1e-6)
                        re_order = np.argsort(conf)[::-1]
                        positions = positions[re_order]

                    for pos in positions[:args.gnina_poses_to_optimize]:
                        center = orig_complex_graph.original_center.cpu().numpy()
                        gnina_ligand_pos, gnina_mol, gnina_score = get_gnina_poses(args, lig, pos, center, name=orig_complex_graph.name[0],
                                                                                   folder=args.folder, gnina_path=args.gnina_path) # TODO set the right folder

                        mol = RemoveAllHs(orig_complex_graph.mol[0])
                        rmsds = []
                        for i in range(len(orig_ligand_pos)):
                            try:
                                rmsd = get_symmetry_rmsd(mol, orig_ligand_pos[i], gnina_ligand_pos, gnina_mol)
                            except Exception as e:
                                print("Using non corrected RMSD because of the error:", e)
                                rmsd = np.sqrt(((gnina_ligand_pos - orig_ligand_pos[i]) ** 2).sum(axis=1).mean(axis=0))
                            rmsds.append(rmsd)
                        rmsds = np.asarray(rmsds)
                        rmsd = np.min(rmsds, axis=0)
                        gnina_rmsds.append(rmsd)
                        gnina_scores.append(gnina_score)

                    gnina_rmsds = np.asarray(gnina_rmsds)
                    assert gnina_rmsds.shape == (args.gnina_poses_to_optimize,), str(gnina_rmsds.shape) + " " + str(args.gnina_poses_to_optimize)
                    gnina_rmsds_list.append(gnina_rmsds)
                    gnina_scores = np.asarray(gnina_scores)
                    gnina_score_list.append(gnina_scores)

                mol = RemoveAllHs(orig_complex_graph.mol[0])
                rmsds = []
                for i in range(len(orig_ligand_pos)):
                    try:
                        rmsd = get_symmetry_rmsd(mol, orig_ligand_pos[i], [l for l in ligand_pos])
                    except Exception as e:
                        print("Using non corrected RMSD because of the error:", e)
                        rmsd = np.sqrt(((ligand_pos - orig_ligand_pos[i]) ** 2).sum(axis=2).mean(axis=1))
                    rmsds.append(rmsd)
                rmsds = np.asarray(rmsds)
                rmsd = np.min(rmsds, axis=0)
                
                centroid_distance = np.min(np.linalg.norm(ligand_pos.mean(axis=1)[None, :] - orig_ligand_pos.mean(axis=1)[:, None], axis=2), axis=0)

                if confidence is not None and isinstance(confidence_args.rmsd_classification_cutoff, list):
                    confidence = confidence[:, 0]
                if confidence is not None:
                    confidence = confidence.cpu().numpy()
                    confidence = np.nan_to_num(confidence, nan=-1e-6)
                    re_order = np.argsort(confidence)[::-1]
                    print(orig_complex_graph['name'], ' rmsd', np.around(rmsd, 1)[re_order], ' centroid distance',
                          np.around(centroid_distance, 1)[re_order], ' confidences ', np.around(confidence, 4)[re_order],
                          (' gnina rmsd ' + str(np.around(gnina_rmsds, 1))) if args.gnina_minimize else '')
                    confidences_list.append(confidence)
                else:
                    print(orig_complex_graph['name'], ' rmsd', np.around(rmsd, 1), ' centroid distance',
                          np.around(centroid_distance, 1))
                centroid_distances_list.append(centroid_distance)

                self_distances = np.linalg.norm(ligand_pos[:, :, None, :] - ligand_pos[:, None, :, :], axis=-1)
                self_distances = np.where(np.eye(self_distances.shape[2]), np.inf, self_distances)
                min_self_distances_list.append(np.min(self_distances, axis=(1, 2)))

                if args.save_complexes:
                    sampled_complexes[orig_complex_graph.name[0]] = data_list

                if args.save_visualisation:
                    if confidence is not None:
                        for rank, batch_idx in enumerate(re_order):
                            visualization_list[batch_idx].write(
                                f'{args.out_dir}/{data_list[batch_idx]["name"][0]}_{rank + 1}_{rmsd[batch_idx]:.1f}_{(confidence)[batch_idx]:.1f}.pdb')
                    else:
                        for rank, batch_idx in enumerate(np.argsort(rmsd)):
                            visualization_list[batch_idx].write(
                                f'{args.out_dir}/{data_list[batch_idx]["name"][0]}_{rank + 1}_{rmsd[batch_idx]:.1f}.pdb')
                without_rec_overlap_list.append(1 if orig_complex_graph.name[0] in names_no_rec_overlap else 0)
                names_list.append(orig_complex_graph.name[0])
                rmsds_list.append(rmsd)
                success = 1
            except Exception as e:
                print("Failed on", orig_complex_graph["name"], e)
                success -= 1
                if bs > 1:
                    bs = bs // 2

        if success != 1:
            rmsds_list.append(np.zeros(args.samples_per_complex) + 10000)
            if confidence_model_args is not None:
                confidences_list.append(np.zeros(args.samples_per_complex) - 10000)
            centroid_distances_list.append(np.zeros(args.samples_per_complex) + 10000)
            min_self_distances_list.append(np.zeros(args.samples_per_complex) + 10000)
            without_rec_overlap_list.append(1 if orig_complex_graph.name[0] in names_no_rec_overlap else 0)
            names_list.append(orig_complex_graph.name[0])
            failures += 1

    print('Performance without hydrogens included in the loss')
    print(failures, "failures due to exceptions")
    print(skipped, ' skipped because complex was not in confidence dataset')

    if args.save_complexes:
        print("Saving complexes.")
        if args.complexes_save_path is not None:
            with open(os.path.join(args.complexes_save_path, "ligands.pkl"), 'wb') as f:
                pickle.dump(sampled_complexes, f)
    
    if args.save_gnina_metrics:
        with open(f'{args.out_dir}/gnina_metrics.pkl', 'wb') as f:
            pickle.dump(gnina_metrics, f)
        print("Saved gnina metrics")

    performance_metrics = {}
    for overlap in ['', 'no_overlap_']:
        if 'no_overlap_' == overlap:
            without_rec_overlap = np.array(without_rec_overlap_list, dtype=bool)
            if without_rec_overlap.sum() == 0: continue
            rmsds = np.array(rmsds_list)[without_rec_overlap]
            min_self_distances = np.array(min_self_distances_list)[without_rec_overlap]
            centroid_distances = np.array(centroid_distances_list)[without_rec_overlap]
            if args.confidence_model_dir is not None:
                confidences = np.array(confidences_list)[without_rec_overlap]
            else:
                confidences = np.array(confidences_list)
            names = np.array(names_list)[without_rec_overlap]
            gnina_rmsds = np.array(gnina_rmsds_list)[without_rec_overlap] if args.gnina_minimize else None
            gnina_score = np.array(gnina_score_list)[without_rec_overlap] if args.gnina_minimize else None

        else:
            rmsds = np.array(rmsds_list)
            gnina_rmsds = np.array(gnina_rmsds_list) if args.gnina_minimize else None
            gnina_score = np.array(gnina_score_list) if args.gnina_minimize else None
            min_self_distances = np.array(min_self_distances_list)
            centroid_distances = np.array(centroid_distances_list)
            confidences = np.array(confidences_list)
            names = np.array(names_list)

        run_times = np.array(run_times)
        np.save(f'{args.out_dir}/{overlap}min_self_distances.npy', min_self_distances)
        np.save(f'{args.out_dir}/{overlap}rmsds.npy', rmsds)
        np.save(f'{args.out_dir}/{overlap}centroid_distances.npy', centroid_distances)
        np.save(f'{args.out_dir}/{overlap}confidences.npy', confidences)
        np.save(f'{args.out_dir}/{overlap}run_times.npy', run_times)
        np.save(f'{args.out_dir}/{overlap}complex_names.npy', np.array(names))
        np.save(f'{args.out_dir}/{overlap}gnina_rmsds.npy', gnina_rmsds)
        np.save(f'{args.out_dir}/{overlap}gnina_score.npy', gnina_score)

        performance_metrics.update({
            f'{overlap}run_times_std': run_times.std().__round__(2),
            f'{overlap}run_times_mean': run_times.mean().__round__(2),
            f'{overlap}mean_rmsd': rmsds.mean(),
            f'{overlap}rmsds_below_2': (100 * (rmsds < 2).sum() / len(rmsds) / N),
            f'{overlap}rmsds_below_5': (100 * (rmsds < 5).sum() / len(rmsds) / N),
            f'{overlap}rmsds_percentile_25': np.percentile(rmsds, 25).round(2),
            f'{overlap}rmsds_percentile_50': np.percentile(rmsds, 50).round(2),
            f'{overlap}rmsds_percentile_75': np.percentile(rmsds, 75).round(2),
            f'{overlap}min_rmsds_below_2': (100 * (np.min(rmsds, axis=1) < 2).sum() / len(rmsds)),
            f'{overlap}min_rmsds_below_5': (100 * (np.min(rmsds, axis=1) < 5).sum() / len(rmsds)),

            f'{overlap}mean_centroid': centroid_distances.mean().__round__(2),
            f'{overlap}centroid_below_2': (100 * (centroid_distances < 2).sum() / len(centroid_distances) / N).__round__(2),
            f'{overlap}centroid_below_5': (100 * (centroid_distances < 5).sum() / len(centroid_distances) / N).__round__(2),
            f'{overlap}centroid_percentile_25': np.percentile(centroid_distances, 25).round(2),
            f'{overlap}centroid_percentile_50': np.percentile(centroid_distances, 50).round(2),
            f'{overlap}centroid_percentile_75': np.percentile(centroid_distances, 75).round(2),
        })

        if args.gnina_minimize:
            score_ordering = np.argsort(gnina_score, axis=1)[:, ::-1]
            filtered_rmsds_gnina = gnina_rmsds[np.arange(gnina_rmsds.shape[0])[:, None], score_ordering][:, 0]

            performance_metrics.update({
                f'{overlap}gnina_rmsds_below_2': (100 * (gnina_rmsds < 2).sum() / len(gnina_rmsds) / args.gnina_poses_to_optimize) if args.gnina_minimize else None,
                f'{overlap}gnina_rmsds_below_5': (100 * (gnina_rmsds < 5).sum() / len(gnina_rmsds) / args.gnina_poses_to_optimize) if args.gnina_minimize else None,
                f'{overlap}gnina_min_rmsds_below_2': (100 * (np.min(gnina_rmsds, axis=1) < 2).sum() / len(gnina_rmsds)) if args.gnina_minimize else None,
                f'{overlap}gnina_min_rmsds_below_5': (100 * (np.min(gnina_rmsds, axis=1) < 5).sum() / len(gnina_rmsds)) if args.gnina_minimize else None,
                f'{overlap}gnina_filtered_rmsds_below_2': (100 * (filtered_rmsds_gnina < 2).sum() / len(filtered_rmsds_gnina)).__round__(2),
                f'{overlap}gnina_filtered_rmsds_below_5': (100 * (filtered_rmsds_gnina < 5).sum() / len(filtered_rmsds_gnina)).__round__(2),
                f'{overlap}gnina_rmsds_percentile_25': np.percentile(gnina_rmsds, 25).round(2),
                f'{overlap}gnina_rmsds_percentile_50': np.percentile(gnina_rmsds, 50).round(2),
                f'{overlap}gnina_rmsds_percentile_75': np.percentile(gnina_rmsds, 75).round(2),

            })

        if N >= 5:
            top5_rmsds = np.min(rmsds[:, :5], axis=1)
            top5_centroid_distances = centroid_distances[
                                          np.arange(rmsds.shape[0])[:, None], np.argsort(rmsds[:, :5], axis=1)][:, 0]
            top5_min_self_distances = min_self_distances[
                                          np.arange(rmsds.shape[0])[:, None], np.argsort(rmsds[:, :5], axis=1)][:, 0]
            performance_metrics.update({
                f'{overlap}top5_self_intersect_fraction': (
                            100 * (top5_min_self_distances < 0.4).sum() / len(top5_min_self_distances)).__round__(2),
                f'{overlap}top5_rmsds_below_2': (100 * (top5_rmsds < 2).sum() / len(top5_rmsds)).__round__(2),
                f'{overlap}top5_rmsds_below_5': (100 * (top5_rmsds < 5).sum() / len(top5_rmsds)).__round__(2),
                f'{overlap}top5_rmsds_percentile_25': np.percentile(top5_rmsds, 25).round(2),
                f'{overlap}top5_rmsds_percentile_50': np.percentile(top5_rmsds, 50).round(2),
                f'{overlap}top5_rmsds_percentile_75': np.percentile(top5_rmsds, 75).round(2),

                f'{overlap}top5_centroid_below_2': (
                            100 * (top5_centroid_distances < 2).sum() / len(top5_centroid_distances)).__round__(2),
                f'{overlap}top5_centroid_below_5': (
                            100 * (top5_centroid_distances < 5).sum() / len(top5_centroid_distances)).__round__(2),
                f'{overlap}top5_centroid_percentile_25': np.percentile(top5_centroid_distances, 25).round(2),
                f'{overlap}top5_centroid_percentile_50': np.percentile(top5_centroid_distances, 50).round(2),
                f'{overlap}top5_centroid_percentile_75': np.percentile(top5_centroid_distances, 75).round(2),
            })

        if N >= 10:
            top10_rmsds = np.min(rmsds[:, :10], axis=1)
            top10_centroid_distances = centroid_distances[
                                           np.arange(rmsds.shape[0])[:, None], np.argsort(rmsds[:, :10], axis=1)][:, 0]
            top10_min_self_distances = min_self_distances[
                                           np.arange(rmsds.shape[0])[:, None], np.argsort(rmsds[:, :10], axis=1)][:, 0]
            performance_metrics.update({
                f'{overlap}top10_self_intersect_fraction': (
                            100 * (top10_min_self_distances < 0.4).sum() / len(top10_min_self_distances)).__round__(2),
                f'{overlap}top10_rmsds_below_2': (100 * (top10_rmsds < 2).sum() / len(top10_rmsds)).__round__(2),
                f'{overlap}top10_rmsds_below_5': (100 * (top10_rmsds < 5).sum() / len(top10_rmsds)).__round__(2),
                f'{overlap}top10_rmsds_percentile_25': np.percentile(top10_rmsds, 25).round(2),
                f'{overlap}top10_rmsds_percentile_50': np.percentile(top10_rmsds, 50).round(2),
                f'{overlap}top10_rmsds_percentile_75': np.percentile(top10_rmsds, 75).round(2),

                f'{overlap}top10_centroid_below_2': (
                            100 * (top10_centroid_distances < 2).sum() / len(top10_centroid_distances)).__round__(2),
                f'{overlap}top10_centroid_below_5': (
                            100 * (top10_centroid_distances < 5).sum() / len(top10_centroid_distances)).__round__(2),
                f'{overlap}top10_centroid_percentile_25': np.percentile(top10_centroid_distances, 25).round(2),
                f'{overlap}top10_centroid_percentile_50': np.percentile(top10_centroid_distances, 50).round(2),
                f'{overlap}top10_centroid_percentile_75': np.percentile(top10_centroid_distances, 75).round(2),
            })

        if confidence_model is not None:
            confidence_ordering = np.argsort(confidences, axis=1)[:, ::-1]

            filtered_rmsds = rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, 0]
            filtered_centroid_distances = centroid_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, 0]
            filtered_min_self_distances = min_self_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, 0]
            performance_metrics.update({
                f'{overlap}filtered_self_intersect_fraction': (
                            100 * (filtered_min_self_distances < 0.4).sum() / len(filtered_min_self_distances)).__round__(
                    2),
                f'{overlap}filtered_rmsds_below_2': (100 * (filtered_rmsds < 2).sum() / len(filtered_rmsds)).__round__(2),
                f'{overlap}filtered_rmsds_below_5': (100 * (filtered_rmsds < 5).sum() / len(filtered_rmsds)).__round__(2),
                f'{overlap}filtered_rmsds_percentile_25': np.percentile(filtered_rmsds, 25).round(2),
                f'{overlap}filtered_rmsds_percentile_50': np.percentile(filtered_rmsds, 50).round(2),
                f'{overlap}filtered_rmsds_percentile_75': np.percentile(filtered_rmsds, 75).round(2),

                f'{overlap}filtered_centroid_below_2': (
                            100 * (filtered_centroid_distances < 2).sum() / len(filtered_centroid_distances)).__round__(2),
                f'{overlap}filtered_centroid_below_5': (
                            100 * (filtered_centroid_distances < 5).sum() / len(filtered_centroid_distances)).__round__(2),
                f'{overlap}filtered_centroid_percentile_25': np.percentile(filtered_centroid_distances, 25).round(2),
                f'{overlap}filtered_centroid_percentile_50': np.percentile(filtered_centroid_distances, 50).round(2),
                f'{overlap}filtered_centroid_percentile_75': np.percentile(filtered_centroid_distances, 75).round(2),
            })

            if N >= 5:
                top5_filtered_rmsds = np.min(rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :5], axis=1)
                top5_filtered_centroid_distances = \
                centroid_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :5][
                    np.arange(rmsds.shape[0])[:, None], np.argsort(
                        rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :5], axis=1)][:, 0]
                top5_filtered_min_self_distances = \
                min_self_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :5][
                    np.arange(rmsds.shape[0])[:, None], np.argsort(
                        rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :5], axis=1)][:, 0]
                performance_metrics.update({
                    f'{overlap}top5_filtered_rmsds_below_2': (
                                100 * (top5_filtered_rmsds < 2).sum() / len(top5_filtered_rmsds)).__round__(2),
                    f'{overlap}top5_filtered_rmsds_below_5': (
                                100 * (top5_filtered_rmsds < 5).sum() / len(top5_filtered_rmsds)).__round__(2),
                    f'{overlap}top5_filtered_rmsds_percentile_25': np.percentile(top5_filtered_rmsds, 25).round(2),
                    f'{overlap}top5_filtered_rmsds_percentile_50': np.percentile(top5_filtered_rmsds, 50).round(2),
                    f'{overlap}top5_filtered_rmsds_percentile_75': np.percentile(top5_filtered_rmsds, 75).round(2),

                    f'{overlap}top5_filtered_centroid_below_2': (100 * (top5_filtered_centroid_distances < 2).sum() / len(
                        top5_filtered_centroid_distances)).__round__(2),
                    f'{overlap}top5_filtered_centroid_below_5': (100 * (top5_filtered_centroid_distances < 5).sum() / len(
                        top5_filtered_centroid_distances)).__round__(2),
                    f'{overlap}top5_filtered_centroid_percentile_25': np.percentile(top5_filtered_centroid_distances,
                                                                                    25).round(2),
                    f'{overlap}top5_filtered_centroid_percentile_50': np.percentile(top5_filtered_centroid_distances,
                                                                                    50).round(2),
                    f'{overlap}top5_filtered_centroid_percentile_75': np.percentile(top5_filtered_centroid_distances,
                                                                                    75).round(2),
                })
            if N >= 10:
                top10_filtered_rmsds = np.min(rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :10],
                                              axis=1)
                top10_filtered_centroid_distances = \
                centroid_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :10][
                    np.arange(rmsds.shape[0])[:, None], np.argsort(
                        rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :10], axis=1)][:, 0]
                top10_filtered_min_self_distances = \
                min_self_distances[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :10][
                    np.arange(rmsds.shape[0])[:, None], np.argsort(
                        rmsds[np.arange(rmsds.shape[0])[:, None], confidence_ordering][:, :10], axis=1)][:, 0]
                performance_metrics.update({
                    f'{overlap}top10_filtered_rmsds_below_2': (
                                100 * (top10_filtered_rmsds < 2).sum() / len(top10_filtered_rmsds)).__round__(2),
                    f'{overlap}top10_filtered_rmsds_below_5': (
                                100 * (top10_filtered_rmsds < 5).sum() / len(top10_filtered_rmsds)).__round__(2),
                    f'{overlap}top10_filtered_rmsds_percentile_25': np.percentile(top10_filtered_rmsds, 25).round(2),
                    f'{overlap}top10_filtered_rmsds_percentile_50': np.percentile(top10_filtered_rmsds, 50).round(2),
                    f'{overlap}top10_filtered_rmsds_percentile_75': np.percentile(top10_filtered_rmsds, 75).round(2),

                    f'{overlap}top10_filtered_centroid_below_2': (100 * (top10_filtered_centroid_distances < 2).sum() / len(
                        top10_filtered_centroid_distances)).__round__(2),
                    f'{overlap}top10_filtered_centroid_below_5': (100 * (top10_filtered_centroid_distances < 5).sum() / len(
                        top10_filtered_centroid_distances)).__round__(2),
                    f'{overlap}top10_filtered_centroid_percentile_25': np.percentile(top10_filtered_centroid_distances,
                                                                                     25).round(2),
                    f'{overlap}top10_filtered_centroid_percentile_50': np.percentile(top10_filtered_centroid_distances,
                                                                                     50).round(2),
                    f'{overlap}top10_filtered_centroid_percentile_75': np.percentile(top10_filtered_centroid_distances,
                                                                                     75).round(2),
                })

    for k in performance_metrics:
        print(k, performance_metrics[k])

    if args.wandb:
        wandb.log(performance_metrics)
        histogram_metrics_list = [('rmsd', rmsds[:, 0]),
                                  ('centroid_distance', centroid_distances[:, 0]),
                                  ('mean_rmsd', rmsds.mean(axis=1)),
                                  ('mean_centroid_distance', centroid_distances.mean(axis=1))]
        if N >= 5:
            histogram_metrics_list.append(('top5_rmsds', top5_rmsds))
            histogram_metrics_list.append(('top5_centroid_distances', top5_centroid_distances))
        if N >= 10:
            histogram_metrics_list.append(('top10_rmsds', top10_rmsds))
            histogram_metrics_list.append(('top10_centroid_distances', top10_centroid_distances))
        if confidence_model is not None:
            histogram_metrics_list.append(('reverse_filtered_rmsds', reverse_filtered_rmsds))
            histogram_metrics_list.append(('reverse_filtered_centroid_distances', reverse_filtered_centroid_distances))
            histogram_metrics_list.append(('filtered_rmsd', filtered_rmsds))
            histogram_metrics_list.append(('filtered_centroid_distance', filtered_centroid_distances))
            if N >= 5:
                histogram_metrics_list.append(('top5_filtered_rmsds', top5_filtered_rmsds))
                histogram_metrics_list.append(('top5_filtered_centroid_distances', top5_filtered_centroid_distances))
                histogram_metrics_list.append(('top5_reverse_filtered_rmsds', top5_reverse_filtered_rmsds))
                histogram_metrics_list.append(
                    ('top5_reverse_filtered_centroid_distances', top5_reverse_filtered_centroid_distances))
            if N >= 10:
                histogram_metrics_list.append(('top10_filtered_rmsds', top10_filtered_rmsds))
                histogram_metrics_list.append(('top10_filtered_centroid_distances', top10_filtered_centroid_distances))
                histogram_metrics_list.append(('top10_reverse_filtered_rmsds', top10_reverse_filtered_rmsds))
                histogram_metrics_list.append(
                    ('top10_reverse_filtered_centroid_distances', top10_reverse_filtered_centroid_distances))