dockonsurf / modules / screening.py @ a5cc42ff
Historique | Voir | Annoter | Télécharger (5,08 ko)
1 |
import logging |
---|---|
2 |
import numpy as np |
3 |
import ase |
4 |
|
5 |
logger = logging.getLogger('DockOnSurf')
|
6 |
|
7 |
|
8 |
def vect_avg(vects): |
9 |
"""Computes the element-wise mean of a set of vectors.
|
10 |
|
11 |
@param vects: np.ndarray with shape (num_vectors, length_vector).
|
12 |
@return: vector average computed doing the element-wise mean.
|
13 |
"""
|
14 |
from utilities import try_command |
15 |
err = "vect_avg parameter vects must be a list-like, able to be converted" \
|
16 |
" np.array"
|
17 |
vects = try_command(np.array, [(ValueError, err)], vects)
|
18 |
if len(vects.shape) == 1: |
19 |
return vects
|
20 |
else:
|
21 |
num_vects = vects.shape[1]
|
22 |
return np.array([np.average(vects[:, i]) for i in range(num_vects)]) |
23 |
|
24 |
|
25 |
def get_atom_coords(atoms: ase.Atoms, ctrs_list): |
26 |
"""Gets the coordinates of the specified atoms from a ase.Atoms object.
|
27 |
|
28 |
Given an ase.Atoms object and a list of atom indices specified in ctrs_list
|
29 |
it gets the coordinates of the specified atoms. If the element in the
|
30 |
ctrs_list is not an index but yet a list of indices, it computes the
|
31 |
element-wise mean of the coordinates of the atoms specified in the inner
|
32 |
list.
|
33 |
@param atoms: ase.Atoms object for which to obtain the coordinates of.
|
34 |
@param ctrs_list: list of (indices/list of indices) of the atoms for which
|
35 |
the coordinates should be extracted.
|
36 |
@return: np.ndarray of atomic coordinates.
|
37 |
"""
|
38 |
coords = [] |
39 |
for i, elem in enumerate(ctrs_list): |
40 |
if isinstance(elem, list): |
41 |
coords.append(vect_avg(np.array([atoms[c].position for c in elem]))) |
42 |
elif isinstance(elem, int): |
43 |
coords.append(atoms[elem].position) |
44 |
else:
|
45 |
err = f"'ctrs_list must be a list of ints or lists, {type(elem)} " \
|
46 |
"found."
|
47 |
logger.error(err) |
48 |
raise ValueError |
49 |
return np.array(coords)
|
50 |
|
51 |
|
52 |
def ads_euler(site, ctr, pts_angle, neigh_ctr): |
53 |
pass
|
54 |
|
55 |
|
56 |
def ads_chemcat(site, ctr, pts_angle): |
57 |
return "TO IMPLEMENT" |
58 |
|
59 |
|
60 |
def adsorb_confs(conf_list, surf, ads_ctrs, sites, algo, num_pts, neigh_ctrs): |
61 |
"""Generates a number of adsorbate-surface structure coordinates.
|
62 |
|
63 |
Given a list of conformers, a surface, a list of atom indices (or list of
|
64 |
list of indices) of both the surface and the adsorbate, it generates a
|
65 |
number of adsorbate-surface structures for every possible combination of
|
66 |
them at different orientations.
|
67 |
@param conf_list: list of ase.Atoms of the different conformers
|
68 |
@param surf: the ase.Atoms object of the surface
|
69 |
@param ads_ctrs: the list atom indices of the adsorbate.
|
70 |
@param sites: the list of atom indices of the surface.
|
71 |
@param algo: the algorithm to use for the generation of adsorbates.
|
72 |
@param num_pts: the number of points per angle orientation to sample
|
73 |
@param neigh_ctrs: the indices of the neighboring atoms to the adsorption
|
74 |
atoms.
|
75 |
@return: list of ase.Atoms for the adsorbate-surface structures
|
76 |
"""
|
77 |
surf_ads_list = [] |
78 |
sites_coords = get_atom_coords(surf, sites) |
79 |
for conf in conf_list: |
80 |
molec_ctr_coords = get_atom_coords(conf, ads_ctrs) |
81 |
molec_neigh_coords = get_atom_coords(conf, neigh_ctrs) |
82 |
for site in sites_coords: |
83 |
for i, molec_ctr in enumerate(molec_ctr_coords): |
84 |
if algo == 'euler': |
85 |
surf_ads_list.append(ads_euler(site, molec_ctr, num_pts, |
86 |
molec_neigh_coords[i])) |
87 |
elif algo == 'chemcat': |
88 |
surf_ads_list.append(ads_chemcat(site, molec_ctr, num_pts)) |
89 |
return surf_ads_list
|
90 |
|
91 |
|
92 |
def run_screening(inp_vars): |
93 |
"""Carry out the screening of adsorbate coordinates on a surface
|
94 |
|
95 |
@param inp_vars: Calculation parameters from input file.
|
96 |
"""
|
97 |
import os |
98 |
from modules.formats import read_coords, read_energies, \ |
99 |
rdkit_mol_to_ase_atoms, adapt_format |
100 |
from modules.clustering import get_rmsd, clustering |
101 |
from modules.isolated import get_moments_of_inertia |
102 |
from modules.calculation import run_calc |
103 |
|
104 |
if not os.path.isdir("isolated"): |
105 |
err = "'isolated' directory not found. It is needed in order to carry "
|
106 |
"out the screening of structures to be adsorbed"
|
107 |
logger.error(err) |
108 |
raise ValueError(err) |
109 |
|
110 |
conf_list = read_coords(inp_vars['code'], 'isolated', 'rdkit') |
111 |
# TODO Implement neighbors algorithm
|
112 |
# neigh_list = get_neighbors(conf_list[0], inp_vars['molec_ads_ctrs'])
|
113 |
conf_enrgs = read_energies(inp_vars['code'], 'isolated') |
114 |
mois = np.array([get_moments_of_inertia(conf)[0] for conf in conf_list]) |
115 |
rmsd_mtx = get_rmsd(conf_list) |
116 |
exemplars = clustering(rmsd_mtx) |
117 |
conf_list = [conf_list[idx] for idx in exemplars] |
118 |
conf_list = [rdkit_mol_to_ase_atoms(conf) for conf in conf_list] |
119 |
surf = adapt_format('ase', inp_vars['surf_file']) |
120 |
surf_ads_list = adsorb_confs(conf_list, surf, inp_vars['molec_ads_ctrs'],
|
121 |
inp_vars['sites'], inp_vars['ads_algo'], |
122 |
inp_vars['sample_points_per_angle'],
|
123 |
inp_vars['molec_neigh_ctrs'])
|
124 |
run_calc('screening', inp_vars, surf_ads_list)
|