| |
| import numpy as np |
| cimport numpy as np |
| import cython |
|
|
| @cython.boundscheck(False) |
| @cython.wraparound(False) |
| def residue_distances(float [:,:] atom_coordinates1, float [:,:] atom_coordinates2, long [:] atoms_per_res1, long [:] atoms_per_res2): |
| cdef: |
| int i, j, x, y, i_atoms, j_atoms, cum_i_atoms, cum_j_atoms, cum_i_atoms_end, cum_j_atoms_end |
| int n_res_i = atoms_per_res1.shape[0] |
| int n_res_j = atoms_per_res2.shape[0] |
| float this_d, min_d |
| float [:,:] res_distances = np.zeros((n_res_i, n_res_j), dtype=np.float32) |
|
|
| cum_i_atoms = 0 |
| for i in range(n_res_i): |
| i_atoms = atoms_per_res1[i] |
| cum_i_atoms_end = cum_i_atoms + i_atoms |
| cum_j_atoms = 0 |
| for j in range(n_res_j): |
| j_atoms = atoms_per_res2[j] |
| min_d = 100000.0 |
| cum_j_atoms_end = cum_j_atoms + j_atoms |
| for x in range(cum_i_atoms, cum_i_atoms_end): |
| for y in range(cum_j_atoms, cum_j_atoms_end): |
| this_d = (atom_coordinates1[x][0] - atom_coordinates2[y][0])**2 + (atom_coordinates1[x][1] - atom_coordinates2[y][1])**2 + (atom_coordinates1[x][2] - atom_coordinates2[y][2])**2 |
| if this_d < min_d: |
| min_d = this_d |
| if min_d > 1000.0: |
| break |
| if min_d > 1000.0: |
| break |
| res_distances[i, j] = min_d |
| cum_j_atoms = cum_j_atoms + j_atoms |
| cum_i_atoms = cum_i_atoms + i_atoms |
|
|
| return res_distances |
|
|
| @cython.boundscheck(False) |
| @cython.wraparound(False) |
| def get_fnat_stats(float [:,:] model_res_distances, float [:,:] native_res_distances, float threshold=5.0): |
| cdef: |
| int native_shape_0 = native_res_distances.shape[0] |
| int native_shape_1 = native_res_distances.shape[1] |
| int i, j |
| int n_native_contacts = 0 |
| int n_model_contacts = 0 |
| int n_shared_contacts = 0 |
| int n_non_native_contacts = 0 |
| float threshold_squared |
|
|
| threshold_squared = threshold * threshold |
| for i in range(native_shape_0): |
| for j in range(native_shape_1): |
| if native_res_distances[i, j] < threshold_squared: |
| n_native_contacts += 1 |
| if model_res_distances[i, j] < threshold_squared: |
| n_shared_contacts += 1 |
| if model_res_distances[i, j] < threshold_squared: |
| n_model_contacts += 1 |
| if native_res_distances[i, j] >= threshold_squared: |
| n_non_native_contacts += 1 |
|
|
| return ( |
| n_shared_contacts, |
| n_non_native_contacts, |
| n_native_contacts, |
| n_model_contacts, |
| ) |
|
|
