comparison sucos_cluster.py @ 6:b8725fec8c7b draft default tip

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit 05dc325ce687441e5d3bdbdedcc0e3529cd5e070"

5:12725d4b90f3

SuCOS is the work of Susan Leung.
GitHub: https://github.com/susanhleung/SuCOS
Publication: https://doi.org/10.26434/chemrxiv.8100203.v1
"""

import sucos, utils
import argparse, gzip
from rdkit import Chem
import numpy as np
import pandas as pd
from scipy.cluster.hierarchy import linkage, fcluster

### start main execution #########################################


def calc_distance_matrix(mols):
    """
    Calculate a full distance matrix for the given molecules. Identical molecules get a score of 0.0 with the maximum

        tmp = []
        for tuple2 in mol_fm_tuples:
            if tuple1[0] == tuple2[0]:
                tmp.append(0.0)
            else:
                #utils.log("Calculating SuCOS between", mol1, mol2)
                sucos_score, fm_score, tani_score = sucos.get_SucosScore(tuple1[0], tuple2[0],
                    tani=True, ref_features=tuple1[1], query_features=tuple2[1])
                tmp.append(1.0 - sucos_score)
        matrix.append(tmp)


    return matrix


def cluster(matrix, threshold=0.8):

    :return: An array of clusters, each cluster being an array of the indices from the matrix.
    """

    indexes = [x for x in range(0, len(matrix))]
    cols = [x for x in range(0, len(matrix[0]))]
    #utils.log("indexes", indexes)
    #utils.log("cols", cols)
    df = pd.DataFrame(matrix, columns=cols, index=indexes)
    utils.log("DataFrame:", df.shape)
    #utils.log(df)
    indices = np.triu_indices(df.shape[0], k=1)
    #utils.log("Indices:", indices)
    t = np.array(df)[indices]
    Z = linkage(t, 'average')
    lig_clusters = []
    cluster_arr = fcluster(Z, t=threshold, criterion='distance')
    for i in range(np.amax(cluster_arr)):
        clus = df.columns[np.argwhere(cluster_arr==i+1)]
        lig_clusters.append([x[0] for x in clus.tolist()])

    utils.log("Clusters", lig_clusters)
    return lig_clusters

def write_clusters_to_sdfs(mols, clusters, basename, gzip=False):
    """
    Write the molecules to SDF files, 1 file for each cluster.
    :param mols The molecules to write:

    for cluster in clusters:
        i += 1
        filename = basename + str(i) + ".sdf"
        if gzip:
            filename += ".gz"
        utils.log("Writing ", len(cluster), "molecules in cluster", i, "to file", filename)
        output_file = utils.open_file_for_writing(filename)
        writer = Chem.SDWriter(output_file)
        for index in cluster:
            mol = mols[index]
            writer.write(mol)
        writer.flush()
        writer.close()
        output_file.close()



def main():
    parser = argparse.ArgumentParser(description='Clustering with SuCOS and RDKit')
    parser.add_argument('-i', '--input', help='Input file in SDF format. Can be gzipped (*.gz).')
    parser.add_argument('-o', '--output', default="cluster", help="Base name for output files in SDF format. " +
                                               "e.g. if value is 'output' then files like output1.sdf, output2.sdf will be created")
    parser.add_argument('--gzip', action='store_true', help='Gzip the outputs generating files like output1.sdf.gz, output2.sdf.gz')
    parser.add_argument('-t', '--threshold', type=float, default=0.8, help='Clustering threshold')

    args = parser.parse_args()
    utils.log("SuCOS Cluster Args: ", args)

    input_file = utils.open_file_for_reading(args.input)

6:b8725fec8c7b

SuCOS is the work of Susan Leung.
GitHub: https://github.com/susanhleung/SuCOS
Publication: https://doi.org/10.26434/chemrxiv.8100203.v1
"""
import argparse

import numpy as np
import pandas as pd
import sucos
import utils
from rdkit import Chem
from scipy.cluster.hierarchy import fcluster, linkage

# start main execution #########################################


def calc_distance_matrix(mols):
    """
    Calculate a full distance matrix for the given molecules. Identical molecules get a score of 0.0 with the maximum

        tmp = []
        for tuple2 in mol_fm_tuples:
            if tuple1[0] == tuple2[0]:
                tmp.append(0.0)
            else:
                # utils.log("Calculating SuCOS between", mol1, mol2)
                sucos_score, fm_score, tani_score = sucos.get_SucosScore(
                    tuple1[0],
                    tuple2[0],
                    tani=True,
                    ref_features=tuple1[1],
                    query_features=tuple2[1],
                )
                tmp.append(1.0 - sucos_score)
        matrix.append(tmp)

    return matrix


def cluster(matrix, threshold=0.8):

    :return: An array of clusters, each cluster being an array of the indices from the matrix.
    """

    indexes = [x for x in range(0, len(matrix))]
    cols = [x for x in range(0, len(matrix[0]))]
    # utils.log("indexes", indexes)
    # utils.log("cols", cols)
    df = pd.DataFrame(matrix, columns=cols, index=indexes)
    utils.log("DataFrame:", df.shape)
    # utils.log(df)
    indices = np.triu_indices(df.shape[0], k=1)
    # utils.log("Indices:", indices)
    t = np.array(df)[indices]
    Z = linkage(t, "average")
    lig_clusters = []
    cluster_arr = fcluster(Z, t=threshold, criterion="distance")
    for i in range(np.amax(cluster_arr)):
        clus = df.columns[np.argwhere(cluster_arr == i + 1)]
        lig_clusters.append([x[0] for x in clus.tolist()])

    utils.log("Clusters", lig_clusters)
    return lig_clusters


def write_clusters_to_sdfs(mols, clusters, basename, gzip=False):
    """
    Write the molecules to SDF files, 1 file for each cluster.
    :param mols The molecules to write:

    for cluster in clusters:
        i += 1
        filename = basename + str(i) + ".sdf"
        if gzip:
            filename += ".gz"
        utils.log(
            "Writing ", len(cluster), "molecules in cluster", i, "to file", filename
        )
        output_file = utils.open_file_for_writing(filename)
        writer = Chem.SDWriter(output_file)
        for index in cluster:
            mol = mols[index]
            writer.write(mol)
        writer.flush()
        writer.close()
        output_file.close()


def main():
    parser = argparse.ArgumentParser(description="Clustering with SuCOS and RDKit")
    parser.add_argument(
        "-i", "--input", help="Input file in SDF format. Can be gzipped (*.gz)."
    )
    parser.add_argument(
        "-o",
        "--output",
        default="cluster",
        help="Base name for output files in SDF format. "
        + "e.g. if value is 'output' then files like output1.sdf, output2.sdf will be created",
    )
    parser.add_argument(
        "--gzip",
        action="store_true",
        help="Gzip the outputs generating files like output1.sdf.gz, output2.sdf.gz",
    )
    parser.add_argument(
        "-t", "--threshold", type=float, default=0.8, help="Clustering threshold"
    )

    args = parser.parse_args()
    utils.log("SuCOS Cluster Args: ", args)

    input_file = utils.open_file_for_reading(args.input)