inference_app.py

import time

import gradio as gr

from gradio_molecule3d import Molecule3D

import sys
import os
import os
import numpy as np
from rdkit import Chem
from rdkit.Chem import AllChem
from rdkit.Chem import Draw
from rdkit.Chem.Draw import IPythonConsole
from rdkit.Chem import DataStructs
from rdkit.Chem import RDConfig
from rdkit.Chem import rdBase
import pickle

from Bio.PDB import *
from Bio import PDB
import requests
import subprocess

import mdtraj as md
from enspara import geometry
from sklearn.cluster import DBSCAN
import pandas as pd


def run_smina(
    ligand_path, protein_path, out_path, pocket_center, pocket_size, num_poses=1, exhaustiveness=1
):
    """
    Perform docking with Smina.

    Parameters
    ----------
    ligand_path: str or pathlib.Path
        Path to ligand PDBQT file that should be docked.
    protein_path: str or pathlib.Path
        Path to protein PDBQT file that should be docked to.
    out_path: str or pathlib.Path
        Path to which docking poses should be saved, SDF or PDB format.
    pocket_center: iterable of float or int
        Coordinates defining the center of the binding site.
    pocket_size: iterable of float or int
        Lengths of edges defining the binding site.
    num_poses: int
        Maximum number of poses to generate.
    exhaustiveness: int
        Accuracy of docking calculations.

    Returns
    -------
    output_text: str
        The output of the Smina calculation.
    """
    output_text = subprocess.check_output(
        [
            "./smina.static",
            "--ligand",
            str(ligand_path),
            "--receptor",
            str(protein_path),
            "--out",
            str(out_path),
            "--center_x",
            str(pocket_center[0]),
            "--center_y",
            str(pocket_center[1]),
            "--center_z",
            str(pocket_center[2]),
            "--size_x",
            str(pocket_size[0]),
            "--size_y",
            str(pocket_size[1]),
            "--size_z",
            str(pocket_size[2]),
            "--num_modes",
            str(num_poses),
            "--exhaustiveness",
            str(exhaustiveness),
        ],
        universal_newlines=True,  # needed to capture output text
    )
    return output_text

def predict (input_sequence, input_ligand, input_protein, exhaustiveness):
    """
    Main prediction function that calls ligsite and smina

    Parameters
    ----------
    input_sequence: str
        monomer sequence
    input_ligand: str
        ligand as SMILES string
    protein_path: gradio.File
        Gradio file object to monomer protein structure as PDB
    exhaustiveness: int
        SMINA parameter

    Returns
    -------
    output_structures: tuple
        (output_protein, output_ligand_sdf)
    run_time: float
        run time of the program
    """
    start_time = time.time()

    if input_protein==None:
        raise gr.Error("need pdb input")
    m=Chem.MolFromSmiles(input_ligand)

    m2=Chem.AddHs(m)
    AllChem.EmbedMolecule(m2)
    AllChem.MMFFOptimizeMolecule(m2)
    
    Chem.SDWriter("/usr/src/app/ligand.sdf").write(m2)

    os.system(f"obabel {input_protein.name} -xr -O /usr/src/app/receptor.pdbqt")
    os.system("obabel -isdf /usr/src/app/ligand.sdf -O /usr/src/app/ligand.pdbqt")

    #Find pocket
    pdb = md.load(input_protein.name)
    # run ligsite
    pockets_xyz = geometry.pockets.get_pocket_cells(struct=pdb)
    eps_value = 0.15
    min_samples_value = 5
    dbscan = DBSCAN(eps=eps_value, min_samples=min_samples_value)
    labels = dbscan.fit_predict(pockets_xyz)
    # Find the unique clusters and their sizes
    unique_labels, counts = np.unique(labels, return_counts=True)
    # Exclude noise points
    valid_clusters = unique_labels[unique_labels != -1]
    valid_counts = counts[unique_labels != -1]
    # Find the cluster with the most points (highest density)
    densest_cluster_label = valid_clusters[np.argmax(valid_counts)]
    densest_cluster_points = pockets_xyz[labels == densest_cluster_label]
    # write cluster to PDB   
    top_df = pd.DataFrame()
    top_df['serial'] = list(range(densest_cluster_points.shape[0]))
    top_df['name'] = 'PK'
    top_df['element'] = 'H'
    top_df['resSeq'] = list(range(densest_cluster_points.shape[0]))
    top_df['resName'] = 'PCK'
    top_df['chainID'] = 0
    pocket_top = md.Topology.from_dataframe(top_df, np.array([]))
    pocket_trj = md.Trajectory(xyz=densest_cluster_points, topology=pocket_top)
    pocket_trj.save('/usr/src/app/pockets_dense.pdb')
    parser = PDBParser()
    struc = parser.get_structure("X", "/usr/src/app/pockets_dense.pdb")
    coords = [x.coord for x in struc.get_atoms()]
    pocket_center = np.mean(coords, axis=0)
    # run smina
    output_text = run_smina(
        "/usr/src/app/ligand.pdbqt",
        "/usr/src/app/receptor.pdbqt",
        "/usr/src/app/docking_pose.sdf",
        pocket_center,
        [10,10,10],
        exhaustiveness=exhaustiveness
    )    
    end_time = time.time()
    run_time = end_time - start_time
    return [input_protein.name,"/usr/src/app/docking_pose.sdf"], run_time

with gr.Blocks() as app:

    gr.Markdown("# LigSite + Smina")

    gr.Markdown("Example model using LigSite and DBScan to find a binding pocket in the protein and then SMINA to dock the ligand in the found pocket.")   
    with gr.Row():
        input_sequence = gr.Textbox(lines=3, label="Input Protein sequence (FASTA)")
        input_ligand = gr.Textbox(lines=3, label="Input ligand SMILES")
        input_protein = gr.File(label="Input protein monomer")
        
    
    # define any options here

    # for automated inference the default options are used
    exhaustiveness = gr.Slider(1,10,value=1, label="Slider Option")
    # checkbox_option = gr.Checkbox(label="Checkbox Option")
    # dropdown_option = gr.Dropdown(["Option 1", "Option 2", "Option 3"], label="Radio Option")

    btn = gr.Button("Run Inference")

    gr.Examples(
        [
            [
                "SVKSEYAEAAAVGQEAVAVFNTMKAAFQNGDKEAVAQYLARLASLYTRHEELLNRILEKARREGNKEAVTLMNEFTATFQTGKSIFNAMVAAFKNGDDDSFESYLQALEKVTAKGETLADQIAKAL:SVKSEYAEAAAVGQEAVAVFNTMKAAFQNGDKEAVAQYLARLASLYTRHEELLNRILEKARREGNKEAVTLMNEFTATFQTGKSIFNAMVAAFKNGDDDSFESYLQALEKVTAKGETLADQIAKAL",
                "COc1ccc(cc1)n2c3c(c(n2)C(=O)N)CCN(C3=O)c4ccc(cc4)N5CCCCC5=O",
                "input_test.pdb"
            ],
        ],
        [input_sequence, input_ligand, input_protein],
    )
    reps =    [
    {
      "model": 0,
      "style": "cartoon",
      "color": "whiteCarbon",
    },
    {
      "model": 0,
      "resname": "UNK",
      "style": "stick",
      "color": "greenCarbon",
    },
        {
      "model": 0,
      "resname": "LIG",
      "style": "stick",
      "color": "greenCarbon",
    },
         {
      "model": 1,
      "style": "stick",
      "color": "greenCarbon",
    }
        
  ]
    
    out = Molecule3D(reps=reps)
    run_time = gr.Textbox(label="Runtime")

    btn.click(predict, inputs=[input_sequence, input_ligand, input_protein, exhaustiveness], outputs=[out, run_time])

app.launch()