Mercurial > repos > bgruening > sucos_clustering

annotate sucos_cluster.py @ 5:12725d4b90f3 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit 944ea4bb8a9cd4244152a4a4fecd0485fabc2ad0"
author bgruening
date Tue, 28 Jul 2020 08:48:16 -0400
parents f80cfac80c53
children b8725fec8c7b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
2 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
3 Cluster a set of molecules based on their 3D overlays as determined by the SuCOS score.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
4
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
5 This will generate a set of SD files, one for each cluster of molecules (presumably corresponding to a
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
6 binding pocket in the protein target).
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
7
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
8
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
9 SuCOS is the work of Susan Leung.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
10 GitHub: https://github.com/susanhleung/SuCOS
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
11 Publication: https://doi.org/10.26434/chemrxiv.8100203.v1
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
12 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
13
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
14 import sucos, utils
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
15 import argparse, gzip
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
16 from rdkit import Chem
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
17 import numpy as np
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
18 import pandas as pd
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
19 from scipy.cluster.hierarchy import linkage, fcluster
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
20
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
21 ### start main execution #########################################
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
22
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
23
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
24 def calc_distance_matrix(mols):
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
25 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
26 Calculate a full distance matrix for the given molecules. Identical molecules get a score of 0.0 with the maximum
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
27 distance possible being 1.0.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
28 :param mols: A list of molecules. It must be possible to iterate through this list multiple times
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
29 :return: A NxN 2D array of distance scores, with N being the number of molecules in the input
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
30 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
31
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
32 # TODO - do we need to calculate both sides of the matrix? Tanimoto is supposed to be a symmetric distance measure,
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
33 # but the matrix that is generated does not seem to be symmetric.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
34
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
35 mol_fm_tuples = []
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
36 for mol in mols:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
37 features = sucos.getRawFeatures(mol)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
38 mol_fm_tuples.append((mol, features))
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
39
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
40 matrix = []
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
41 for tuple1 in mol_fm_tuples:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
42 tmp = []
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
43 for tuple2 in mol_fm_tuples:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
44 if tuple1[0] == tuple2[0]:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
45 tmp.append(0.0)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
46 else:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
47 #utils.log("Calculating SuCOS between", mol1, mol2)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
48 sucos_score, fm_score, tani_score = sucos.get_SucosScore(tuple1[0], tuple2[0],
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
49 tani=True, ref_features=tuple1[1], query_features=tuple2[1])
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
50 tmp.append(1.0 - sucos_score)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
51 matrix.append(tmp)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
52
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
53
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
54 return matrix
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
55
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
56
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
57 def cluster(matrix, threshold=0.8):
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
58 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
59 Cluster the supplied distance matrix returning an array of clusters.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
60 :param matrix: the distance matrix, as calculated with the calc_distance_matrix function.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
61 :param threshold: The clustering cuttoff. The default of 0.8 is a reasonable value to use.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
62 :return: An array of clusters, each cluster being an array of the indices from the matrix.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
63 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
64
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
65 indexes = [x for x in range(0, len(matrix))]
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
66 cols = [x for x in range(0, len(matrix[0]))]
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
67 #utils.log("indexes", indexes)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
68 #utils.log("cols", cols)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
69 df = pd.DataFrame(matrix, columns=cols, index=indexes)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
70 utils.log("DataFrame:", df.shape)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
71 #utils.log(df)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
72 indices = np.triu_indices(df.shape[0], k=1)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
73 #utils.log("Indices:", indices)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
74 t = np.array(df)[indices]
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
75 Z = linkage(t, 'average')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
76 lig_clusters = []
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
77 cluster_arr = fcluster(Z, t=threshold, criterion='distance')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
78 for i in range(np.amax(cluster_arr)):
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
79 clus = df.columns[np.argwhere(cluster_arr==i+1)]
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
80 lig_clusters.append([x[0] for x in clus.tolist()])
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
81
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
82 utils.log("Clusters", lig_clusters)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
83 return lig_clusters
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
84
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
85 def write_clusters_to_sdfs(mols, clusters, basename, gzip=False):
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
86 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
87 Write the molecules to SDF files, 1 file for each cluster.
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
88 :param mols The molecules to write:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
89 :param clusters The clusters, as returned by the cluster function:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
90 :param basename The basename for the file name. e.g. if basename is 'output' then files like
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
91 output1.sdf, output2.sdf will be written:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
92 :param gzip Whether to gzip the output
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
93 :return:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
94 """
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
95
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
96 i = 0
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
97 for cluster in clusters:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
98 i += 1
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
99 filename = basename + str(i) + ".sdf"
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
100 if gzip:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
101 filename += ".gz"
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
102 utils.log("Writing ", len(cluster), "molecules in cluster", i, "to file", filename)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
103 output_file = utils.open_file_for_writing(filename)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
104 writer = Chem.SDWriter(output_file)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
105 for index in cluster:
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
106 mol = mols[index]
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
107 writer.write(mol)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
108 writer.flush()
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
109 writer.close()
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
110 output_file.close()
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
111
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
112
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
113
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
114 def main():
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
115 parser = argparse.ArgumentParser(description='Clustering with SuCOS and RDKit')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
116 parser.add_argument('-i', '--input', help='Input file in SDF format. Can be gzipped (*.gz).')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
117 parser.add_argument('-o', '--output', default="cluster", help="Base name for output files in SDF format. " +
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
118 "e.g. if value is 'output' then files like output1.sdf, output2.sdf will be created")
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
119 parser.add_argument('--gzip', action='store_true', help='Gzip the outputs generating files like output1.sdf.gz, output2.sdf.gz')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
120 parser.add_argument('-t', '--threshold', type=float, default=0.8, help='Clustering threshold')
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
121
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
122 args = parser.parse_args()
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
123 utils.log("SuCOS Cluster Args: ", args)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
124
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
125 input_file = utils.open_file_for_reading(args.input)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
126 suppl = Chem.ForwardSDMolSupplier(input_file)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
127 mols = list(suppl)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
128 matrix = calc_distance_matrix(mols)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
129 clusters = cluster(matrix, threshold=args.threshold)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
130 write_clusters_to_sdfs(mols, clusters, args.output, gzip=args.gzip)
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
131
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
132
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
133 if __name__ == "__main__":
f80cfac80c53 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/sucos commit ef86cfa5f7ab5043de420511211579d03df58645"
bgruening
parents:
diff changeset
134 main()