File size: 2,195 Bytes
638ea38 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 | import json
import urllib.request
import numpy as np
from Bio.Align import PairwiseAligner, substitution_matrices
def alphafold_model(acc, directory):
directory.mkdir(parents=True, exist_ok=True)
f = directory / f"{acc}.pdb"
if not f.exists():
meta = json.load(urllib.request.urlopen(f"https://alphafold.ebi.ac.uk/api/prediction/{acc}", timeout=60))[0]
urllib.request.urlretrieve(meta["pdbUrl"], f)
return f
ALIGNER = PairwiseAligner(mode="global", open_gap_score=-11, extend_gap_score=-1)
ALIGNER.substitution_matrix = substitution_matrices.load("BLOSUM62")
def correspondence(s1, s2):
aln = ALIGNER.align(s1, s2)[0]
i1, i2 = [], []
for (a0, a1), (b0, b1) in zip(*aln.aligned):
i1.extend(range(a0, a1))
i2.extend(range(b0, b1))
return np.array(i1), np.array(i2)
def kabsch(P, Q):
pc, qc = P.mean(0), Q.mean(0)
U, _, Vt = np.linalg.svd((P - pc).T @ (Q - qc))
d = np.sign(np.linalg.det(U @ Vt))
R = U @ np.diag([1.0, 1.0, d]) @ Vt
return R, pc, qc
def tm_score(P, Q, L):
n = len(P)
d0 = max(1.24 * (max(L, 19) - 15) ** (1 / 3) - 1.8, 0.5)
d0s = min(max(d0, 4.5), 8.0)
best, best_rmsd = 0.0, np.nan
sizes, f = [], n
while f >= 4 and len(sizes) < 6:
sizes.append(f)
f //= 2
for size in sizes:
step = max(1, size // 2)
for start in range(0, n - size + 1, step):
sel = np.arange(start, start + size)
for _ in range(20):
R, pc, qc = kabsch(P[sel], Q[sel])
d = np.linalg.norm((P - pc) @ R + qc - Q, axis=1)
tm = np.sum(1 / (1 + (d / d0) ** 2)) / L
if tm > best:
best, best_rmsd = tm, float(np.sqrt(np.mean(d[d < 5] ** 2))) if np.any(d < 5) else np.nan
cut = d0s
new = np.where(d < cut)[0]
while len(new) < 3:
cut += 0.5
new = np.where(d < cut)[0]
if len(new) == len(sel) and np.all(new == sel):
break
sel = new
return best, best_rmsd
|