|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
""" |
|
4
|
|
|
Author: Guillaume Bouvier -- [email protected] |
|
5
|
|
|
https://research.pasteur.fr/en/member/guillaume-bouvier/ |
|
6
|
|
|
2018-01-24 15:23:33 (UTC+0100) |
|
7
|
|
|
https://bougui505.github.io/2018/01/24/compute_the_global_distance_test_(gdt)_with_pymol.html |
|
8
|
|
|
""" |
|
9
|
|
|
|
|
10
|
|
|
from pymol import cmd |
|
11
|
|
|
import numpy as np |
|
12
|
|
|
|
|
13
|
|
|
print("Loaded calc_gdt.py: gdt(sel1, sel2), gdt_hs(sel1, sel2)") |
|
14
|
|
View Code Duplication |
@cmd.extend |
|
|
|
|
|
|
15
|
|
|
def gdt(sel1, sel2): |
|
16
|
|
|
""" |
|
17
|
|
|
Calculate the GDT (Global Distance Test) between two RNA (using P atoms) selections. |
|
18
|
|
|
Args: |
|
19
|
|
|
sel1 (str): selection string for the protein to compare. |
|
20
|
|
|
sel2 (str): selection string for the reference protein. |
|
21
|
|
|
""" |
|
22
|
|
|
# Select only C-alpha atoms |
|
23
|
|
|
sel1 += ' and name P' |
|
24
|
|
|
sel2 += ' and name P' |
|
25
|
|
|
cutoffs = [1.0, 2.0, 4.0, 8.0] |
|
26
|
|
|
# Align without transforming for raw alignment data |
|
27
|
|
|
cmd.align(sel1, sel2, cycles=0, transform=0, object='aln') |
|
28
|
|
|
# Get raw alignment data |
|
29
|
|
|
mapping = cmd.get_raw_alignment('aln') |
|
30
|
|
|
distances = [] |
|
31
|
|
|
for pair in mapping: |
|
32
|
|
|
atom1 = '%s and id %d' % (pair[0][0], pair[0][1]) |
|
33
|
|
|
atom2 = '%s and id %d' % (pair[1][0], pair[1][1]) |
|
34
|
|
|
dist = cmd.get_distance(atom1, atom2) |
|
35
|
|
|
distances.append(dist) |
|
36
|
|
|
distances = np.array(distances) |
|
37
|
|
|
gdts = [(distances <= cutoff).mean() for cutoff in cutoffs] |
|
38
|
|
|
out = np.array(list(zip(cutoffs, gdts))).flatten() |
|
39
|
|
|
print("GDT_%d: %.4f; GDT_%d: %.4f; GDT_%d: %.4f; GDT_%d: %.4f;" % tuple(out)) |
|
40
|
|
|
GDT = np.mean(gdts) |
|
41
|
|
|
print("GDT: %.4f" % GDT) |
|
42
|
|
|
# Color by distance |
|
43
|
|
|
cmd.spectrum('b', 'green_yellow_red', sel1) |
|
44
|
|
|
return GDT |
|
45
|
|
|
|
|
46
|
|
View Code Duplication |
@cmd.extend |
|
|
|
|
|
|
47
|
|
|
def gdt_hs(sel1, sel2): |
|
48
|
|
|
""" |
|
49
|
|
|
Calculate the GDT (Global Distance Test) High Accuracy between two RNA (using P atoms) selections. |
|
50
|
|
|
Args: |
|
51
|
|
|
sel1 (str): selection string for the protein to compare. |
|
52
|
|
|
sel2 (str): selection string for the reference protein. |
|
53
|
|
|
""" |
|
54
|
|
|
# Select only C-alpha atoms |
|
55
|
|
|
sel1 += ' and name P' |
|
56
|
|
|
sel2 += ' and name P' |
|
57
|
|
|
cutoffs = [0.5, 1, 2, 4] |
|
58
|
|
|
# Align without transforming for raw alignment data |
|
59
|
|
|
cmd.align(sel1, sel2, cycles=0, transform=0, object='aln') |
|
60
|
|
|
# Get raw alignment data |
|
61
|
|
|
mapping = cmd.get_raw_alignment('aln') |
|
62
|
|
|
distances = [] |
|
63
|
|
|
for pair in mapping: |
|
64
|
|
|
atom1 = '%s and id %d' % (pair[0][0], pair[0][1]) |
|
65
|
|
|
atom2 = '%s and id %d' % (pair[1][0], pair[1][1]) |
|
66
|
|
|
dist = cmd.get_distance(atom1, atom2) |
|
67
|
|
|
distances.append(dist) |
|
68
|
|
|
distances = np.array(distances) |
|
69
|
|
|
gdts = [(distances <= cutoff).mean() for cutoff in cutoffs] |
|
70
|
|
|
out = np.array(list(zip(cutoffs, gdts))).flatten() |
|
71
|
|
|
print("GDT_HS%d: %.4f; GDT_%d: %.4f; GDT_%d: %.4f; GDT_%d: %.4f;" % tuple(out)) |
|
72
|
|
|
GDT = np.mean(gdts) |
|
73
|
|
|
print("GDT_HS: %.4f" % GDT) |
|
74
|
|
|
# Color by distance |
|
75
|
|
|
cmd.spectrum('b', 'green_yellow_red', sel1) |
|
76
|
|
|
return GDT |
|
77
|
|
|
|
|
78
|
|
|
if __name__ == '__main__': |
|
79
|
|
|
# Assume the first two objects are the models to compare |
|
80
|
|
|
model1, reference = cmd.get_object_list('all')[:2] |
|
81
|
|
|
gdt_gs(model1, reference) |
|
|
|
|
|
|
82
|
|
|
|
mmagnus /
rna-tools
Loading history...