OneScience-Group/DiffDock
053
1o
2 �aOj�: � @ s� d dl Z d dlZd dlZd dlZd dlmZ d dlmZ d dlZ d dl3Z4d dl Z d dl m
Z
d dlmZ d dlmZ d dlmZ d dlmZmZ d d lmZmZ d d5lmZm Z d d l m Z z d d
l m Z m!Z!m"Z" W n e#y} d d
l m Z m!Z!m"Z" Y nw dd� Z$dd� Z%dd� Z&dd� Z'dd� Z(dd� Z)dd� Z*d d � Z+d d � Z,d d!� Z-d"d#� Z.d$d%� Z/d&d'� Z0d(d)� Z1d*d+� Z2e3d,kr�e2� dS dS )-� N)�Path��SimpleNamespace)� RemoveAllHs)�6DataLoader)�write_mol_with_coords)�get_t_schedule)�InferenceDataset� set_nones)�configure_logger�7get_logger)�randomize_position�sampling)� validate_sampling_entrypoint� )�load_model_args�load_score_model�model_uses_lm_embeddingsc C s t �� } | jdddd� | �� S )Nz--configTz!Path to the sampling YAML config.)�required�help)�argparse�ArgumentParser� add_argument�8parse_args)�parser� r �.modelscope/diffdock/scripts/sample_diffdock.pyr s r c C sN t | t�r tj�| �S t | t�rdd� | �� D �S t | t�r%dd� | D �S | S )Nc S s i | ] \}}|t |��qS r ��_resolve_env_vars)�.0�k�vr r r �9<dictcomp>$ s z%_resolve_env_vars.<locals>.<dictcomp>c S s g | ]}t |��qS r r )r r! r r r �10<listcomp>& s z%_resolve_env_vars.<locals>.<listcomp>)�11isinstance�str�os�path�12expandvars�dict�items�list)�objr r r r s 13 1415r c C sD t | ddd��}tt�|�pi �W d � S 1 sw Y d S )N�r�utf-8)�encoding)�openr �yamlZ safe_load)Z config_path�handler r r � load_config, s $�r3 c C s8 i }| � � D ]\}}t|t�r|�|� q|||< q|S �N)r* r$ r) �update)�configZflat�key�valuer r r �flatten_config1 s 16 17r9 c C s t di | ��S )Nr r )r6 r r r � to_namespace; s r: c C s, | dv rt �t j�� rd�S d�S t �| �S )N> �autoN�cuda�cpu)�torch�devicer< Z is_available)Z device_namer r r �resolve_device? s 18r@ c C s� | j d urJt| j dddd��}tt�|��}W d � n1 s w Y tdd� |D ��}tdd� |D ��}tdd� |D ��}td d� |D ��}n| jpNd19g}| jg}| jg}| j g}d d� t20|�D �}||||fS ) Nr- r. � )r/ �newlinec S � g | ]}|� d ��qS )� complex_name��get�r �rowr r r r# I � z load_inputs.<locals>.<listcomp>c S rC )� protein_pathrE rG r r r r# J rI c S rC )�protein_sequencerE rG r r r r# K rI c S rC )�ligand_descriptionrE rG r r r r# L rI Z complex_0c S s& g | ]\}}|d ur |nd|� ��qS )NZcomplex_r )r �idx�namer r r r# S s & ) Zprotein_ligand_csvr0 r+ �csvZ21DictReaderr22 rD rJ rK rL � enumerate)�argsr2 Zrows�
complex_names�
protein_paths�protein_sequences�ligand_descriptionsr r r � load_inputsE s 23 � rV c C s4 t j| dd� |D ]}t jt j�| |�dd� q d S )NT)�exist_ok)r& �makedirsr' �join)�out_dirrR rN r r r �ensure_output_dirsW s �r[ c C s | j }t|ttf�r|d S |S )Nr )�molr$ r+ �tuple)�
complex_graphr\ r r r �get_ligand_mol] s r_ c C s t | dd �}|d u rt|�S |S )N�
lm_embeddings)�getattrr )rQ �24model_argsr` r r r �resolve_lm_embeddings_flagb s rc c C s: t | j||||||j|j|j|j|j|jt|dd� d�
S )N�not_knn_only_graphF)
rZ rR Z
protein_filesrU rT r` �receptor_radius� remove_hs�c_alpha_max_neighbors� all_atoms� atom_radius�atom_max_neighborsZknn_only_graph) r rZ re rf rg rh ri rj ra )rQ rb rR rS rT rU r` r r r �build_inference_dataseti s 25 �rk c 26 C sR t | dd �t | dd �t | dd �t | dd �t | dd �t | dd �t | dd� t|�fS ) Nre rf rg rh ri rj rd F)ra �bool)rb r` r r r �inference_graph_signature� s 272829303132 �rm c C sB | }t t|dd �t�r|d d �df }t�|�� �� �� ��d�S )NZrmsd_classification_cutoffr �����) r$ ra r+ �np�asarray�detachr= �numpyZreshape)�33confidence�confidence_model_args�confidence_scoresr r r �extract_confidence_scores� s rv c 34 C s� t | dd �}|d u r dS t|�}|| j }|�� s!td|� d���t|�}t|dddd� t|| j|ddt | dd �d35�\}}}t| d �rH| j |_ |�36d |� ||fS )
N�confidence_model_dir)NNz!Confidence checkpoint not found: �-. This example does not auto-download models.z-DiffDock sampling confidence-model checkpointT)�context�include_confidence�confidence_mode�old_confidence_modelF)� model_dir�ckptr? � no_parallelr{ �old� crop_beyondz(Loaded optional confidence model from %s) ra r Zconfidence_ckpt�exists�FileNotFoundErrorr r r �hasattrr� �info) rQ r? �loggerrw Zconfidence_checkpoint_pathZ confidence_model_args_preview�confidence_modelrt �_r r r � load_optional_confidence_model� s8 3738��39 �40 r� c - s$ t � } t| j�}tt|��}tt|dd��}t|t|dd �d up%t|dd�d� tt|dd�� t � }t41|j �}||j }|�
� sItd |� d42���t|�}t|d d � t||j |d
t|dd�d�\}} }43t|d�rm|j| _|�d|� t|||�\} } t|�\}
}}}t|j|
� t|| �}t|| |
||||d�}t|ddd�}d }| d ur�t|| �}t| |�t| |�k}|r�|�d� t|| |
||||d�}t|ddd�}n|�d� t |j |j |j |j d�}d}d}|j!}t"|�} |�d| � |d u r�dd� |D �} nt#||�} t$| �D �]s\} \�� �j%d �s"|d7 }|�&d |j'| � �q� d u�r<� j%d �s<|d7 }|�&d |j'| � �q�z �fd d �t(|�D �} d } � d u�r[� fd d �t(|�D �} t)| | j*|j+| j,|j-|j.d!� t/��}!t0dDi d"| �d#|�d$|j1d u�r�|j1n|j �d%|�d&|�d'|�d|�d(|44�d)| �d*|j+�d+|j2�d,| �d-| �d.| �d/|j3�d0|j4�d1|j5|j6|j7g�d2|j8|j9|j:g�d3|j;|j<|j=g��\} }"t>�?�fd4d �| D ��}#t>�@t"|#��}$d }%|"d u�rtA|"| �}%t>jB|%d5d6�}%t>�C|%�d d d7� }$|�d8|
| t>jD|%|$ d9d:�� tEjF�G|j|
| �}&t$|$dd;�D ]8\}'}(|#|( })tH�I|!�}*| jJ�r3tK|*�}*d<|'� d=�}+|%d u�rJd<|'� d>|%|( d?�d=�}+tL|*|)tEjF�G|&|+�� �q W �q tM�yz }, z|�&d@|
| |,� |d7 }W Y d },~,�qd },~,ww |�dA|| � |�dB|| � |�dC|j� d S )ENr? r; rw r| F)rz Zloglevel�INFOzCheckpoint not found: rx z(DiffDock sampling score-model checkpoint)ry TZold_score_model)r} r~ r? r r� r� z DiffDock sampling will run on %s)rR rS rT rU r` � )Zdataset�45batch_sizeZshufflez`Confidence rerank requires independent inference graphs; building a separate confidence dataset.z>Confidence rerank will reuse the score-model inference graphs.)�sigma_schedule�inference_steps�inf_sched_alpha�inf_sched_betar zSize of test dataset: %sc s s � | ]}|d fV qd S r4 r )r �orig_complex_graphr r r � <genexpr> s � zmain.<locals>.<genexpr>z)Skipping %s because preprocessing failed.z4Skipping %s because confidence preprocessing failed.c � g | ]}t �� ��qS r ��copy�deepcopy�r r� �r� r r r# . rI zmain.<locals>.<listcomp>c r� r r� r� )� confidence_orig_complex_graphr r r# 1 rI )� initial_noise_std_proportion�choose_residue� data_list�modelr� � tr_scheduleZ rot_scheduleZ tor_schedule�46t_to_sigmarb � no_random�oder� �confidence_data_listrt r� �no_final_step_noiseZ
temp_samplingZtemp_psiZtemp_sigma_datac s, g | ]}|d j �� �� � j�� �� �qS )�ligand)�posr= rr Zoriginal_center)r r^ r� r r r# _ s ��g�����ư�)�nanrn z8Applied confidence rerank for %s. Ranked confidences: %s� )Z precision)�start�rankz.sdfZ_confz.4fzFailed on %s with error: %sz Failed for %s / %s complexes.zSkipped %s / %s complexes.zResults saved in %sr )Nr r3 r6 r: r9 r@ ra r r r r r} r~ r� r� r r r� r� r� r� rV r[ rZ rc rk r rm r r� r� r� r� Zsamples_per_complex�len�ziprP ZsuccessZwarningrR �ranger
Z47no_torsionr� Z tr_sigma_maxr� r� r_ r Z actual_stepsr� r� r� Ztemp_sampling_trZtemp_sampling_rotZtemp_sampling_torZ temp_psi_trZ temp_psi_rotZ temp_psi_torZtemp_sigma_data_trZtemp_sigma_data_rotZtemp_sigma_data_torro rp Zarangerv Z48nan_to_numZargsortZ array2stringr& r' rY r� r� rf r r � Exception)-ZparsedZ49raw_configrQ r? r� r} Zcheckpoint_pathZscore_model_args_previewr� Zscore_model_argsr� r� rt rR rS rT rU Zscore_lm_embeddingsZ test_datasetZ test_loaderZconfidence_loaderZconfidence_lm_embeddingsZ"confidence_needs_independent_graphZconfidence_datasetr� ZfailuresZskippedZ num_samplesZ test_ds_sizeZ loader_iterrM r� r� r� rs Zligand_positionsZ rerank_orderru Z write_dirr� Z50sample_idxr� Zmol_pred�filename�excr )r� r� r �main� s� 51 52�535455��56 �57 58� 59���� 60� 61 ��62� ��������� �63� � �
�������� ��"64��6566�67 68�69��r� �__main__)4r r� rO r& Zpathlibr �typesr rr ro r> r1 Z70rdkit.Chemr Ztorch_geometric.loaderr Z*onescience.datapipes.diffdock.process_molsr Z)onescience.utils.diffdock.diffusion_utilsr Z)onescience.utils.diffdock.inference_utilsr r71 Z'onescience.utils.diffdock.logging_utilsr r Z"onescience.utils.diffdock.samplingr
r Z$onescience.utils.diffdock.validationr Zmodels.score_wrapperr r r � ImportErrorr r r3 r9 r: r@ rV r[ r_ rc rk rm rv r� r� �__name__r r r r �<module> sR � 72
# K73�