Spaces:
Sleeping
Sleeping
add dataset generation script
Browse files- .gitignore +6 -0
- README.md +3 -3
- env_consts.py +1 -1
- prepare_pinder_dataset.py +622 -0
- resources/{77-182500_only_weights.ckpt → only_weights_102-240750.ckpt} +2 -2
.gitignore
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Ignore __pycache__ folders
|
| 2 |
+
__pycache__/
|
| 3 |
+
.idea/
|
| 4 |
+
|
| 5 |
+
# Ignore .DS_Store files
|
| 6 |
+
.DS_Store
|
README.md
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: indigo
|
| 5 |
colorTo: indigo
|
| 6 |
sdk: docker
|
| 7 |
pinned: false
|
| 8 |
-
license:
|
| 9 |
---
|
| 10 |
|
| 11 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
|
| 1 |
---
|
| 2 |
+
title: DockFormerPP
|
| 3 |
+
emoji: ⚡
|
| 4 |
colorFrom: indigo
|
| 5 |
colorTo: indigo
|
| 6 |
sdk: docker
|
| 7 |
pinned: false
|
| 8 |
+
license: apache-2.0
|
| 9 |
---
|
| 10 |
|
| 11 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
env_consts.py
CHANGED
|
@@ -3,7 +3,7 @@ import os
|
|
| 3 |
TEST_INPUT_DIR = None
|
| 4 |
TEST_OUTPUT_DIR = None
|
| 5 |
THIS_FILE_DIR = os.path.dirname(os.path.abspath(__file__))
|
| 6 |
-
CKPT_PATH = os.path.join(THIS_FILE_DIR, "resources", "
|
| 7 |
RUN_CONFIG_PATH = os.path.join(THIS_FILE_DIR, "resources", "run_config.json")
|
| 8 |
|
| 9 |
OUTPUT_PATH = os.path.join(THIS_FILE_DIR, "predicted_out.pdb")
|
|
|
|
| 3 |
TEST_INPUT_DIR = None
|
| 4 |
TEST_OUTPUT_DIR = None
|
| 5 |
THIS_FILE_DIR = os.path.dirname(os.path.abspath(__file__))
|
| 6 |
+
CKPT_PATH = os.path.join(THIS_FILE_DIR, "resources", "only_weights_102-240750.ckpt")
|
| 7 |
RUN_CONFIG_PATH = os.path.join(THIS_FILE_DIR, "resources", "run_config.json")
|
| 8 |
|
| 9 |
OUTPUT_PATH = os.path.join(THIS_FILE_DIR, "predicted_out.pdb")
|
prepare_pinder_dataset.py
ADDED
|
@@ -0,0 +1,622 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import json
|
| 2 |
+
import os
|
| 3 |
+
import shutil
|
| 4 |
+
import random
|
| 5 |
+
import sys
|
| 6 |
+
from typing import List, Tuple, Optional
|
| 7 |
+
|
| 8 |
+
import Bio.PDB
|
| 9 |
+
import Bio.SeqUtils
|
| 10 |
+
import pandas as pd
|
| 11 |
+
import numpy as np
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
OUTPUT_FOLDER = "/tmp/output"
|
| 16 |
+
PINDER_ANNOTATIONS = "/tmp/index.parquet"
|
| 17 |
+
GSUTIL_PATH = "/tmp/google-cloud-sdk/bin/gsutil"
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
MAX_SYSTEMS_FOR_CLUSTER = 2
|
| 21 |
+
MAX_LENGTH = 350
|
| 22 |
+
MAX_TRIES_OF_METHOD = 5
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def do_robust_chain_object_renumber(chain: Bio.PDB.Chain.Chain, new_chain_id: str) -> Optional[Bio.PDB.Chain.Chain]:
|
| 26 |
+
all_residues = [res for res in chain.get_residues()
|
| 27 |
+
if "CA" in res and Bio.SeqUtils.seq1(res.get_resname()) not in ("X", "", " ")]
|
| 28 |
+
if not all_residues:
|
| 29 |
+
return None
|
| 30 |
+
|
| 31 |
+
res_and_res_id = [(res, res.get_id()[1]) for res in all_residues]
|
| 32 |
+
|
| 33 |
+
min_res_id = min([i[1] for i in res_and_res_id])
|
| 34 |
+
if min_res_id < 1:
|
| 35 |
+
print("Negative res id", chain, min_res_id)
|
| 36 |
+
factor = -1 * min_res_id + 1
|
| 37 |
+
res_and_res_id = [(res, res_id + factor) for res, res_id in res_and_res_id]
|
| 38 |
+
|
| 39 |
+
res_and_res_id_no_collisions = []
|
| 40 |
+
for res, res_id in res_and_res_id[::-1]:
|
| 41 |
+
if res_and_res_id_no_collisions and res_and_res_id_no_collisions[-1][1] == res_id:
|
| 42 |
+
# there is a collision, usually an insertion residue
|
| 43 |
+
res_and_res_id_no_collisions = [(i, j + 1) for i, j in res_and_res_id_no_collisions]
|
| 44 |
+
res_and_res_id_no_collisions.append((res, res_id))
|
| 45 |
+
|
| 46 |
+
first_res_id = min([i[1] for i in res_and_res_id_no_collisions])
|
| 47 |
+
factor = 1 - first_res_id # start from 1
|
| 48 |
+
new_chain = Bio.PDB.Chain.Chain(new_chain_id)
|
| 49 |
+
|
| 50 |
+
res_and_res_id_no_collisions.sort(key=lambda x: x[1])
|
| 51 |
+
|
| 52 |
+
for res, res_id in res_and_res_id_no_collisions:
|
| 53 |
+
chain.detach_child(res.id)
|
| 54 |
+
res.id = (" ", res_id + factor, " ")
|
| 55 |
+
new_chain.add(res)
|
| 56 |
+
|
| 57 |
+
return new_chain
|
| 58 |
+
|
| 59 |
+
|
| 60 |
+
def robust_renumber_protein(pdb_path: str, output_path: str):
|
| 61 |
+
if pdb_path.endswith(".pdb"):
|
| 62 |
+
pdb_parser = Bio.PDB.PDBParser(QUIET=True)
|
| 63 |
+
pdb_struct = pdb_parser.get_structure("original_pdb", pdb_path)
|
| 64 |
+
elif pdb_path.endswith(".cif"):
|
| 65 |
+
pdb_struct = Bio.PDB.MMCIFParser().get_structure("original_pdb", pdb_path)
|
| 66 |
+
else:
|
| 67 |
+
raise ValueError("Unknown file type", pdb_path)
|
| 68 |
+
assert len(list(pdb_struct)) == 1, "can't extract if more than one model"
|
| 69 |
+
model = next(iter(pdb_struct))
|
| 70 |
+
chains = list(model.get_chains())
|
| 71 |
+
new_model = Bio.PDB.Model.Model(0)
|
| 72 |
+
chain_ids = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"
|
| 73 |
+
for chain, chain_id in zip(chains, chain_ids):
|
| 74 |
+
new_chain = do_robust_chain_object_renumber(chain, chain_id)
|
| 75 |
+
if new_chain is None:
|
| 76 |
+
continue
|
| 77 |
+
new_model.add(new_chain)
|
| 78 |
+
new_struct = Bio.PDB.Structure.Structure("renumbered_pdb")
|
| 79 |
+
new_struct.add(new_model)
|
| 80 |
+
io = Bio.PDB.PDBIO()
|
| 81 |
+
io.set_structure(new_struct)
|
| 82 |
+
io.save(output_path)
|
| 83 |
+
|
| 84 |
+
|
| 85 |
+
def get_chain_object_to_seq(chain: Bio.PDB.Chain.Chain) -> str:
|
| 86 |
+
res_id_to_res = {res.get_id()[1]: res for res in chain.get_residues() if "CA" in res}
|
| 87 |
+
|
| 88 |
+
if len(res_id_to_res) == 0:
|
| 89 |
+
print("skipping empty chain", chain.get_id())
|
| 90 |
+
return ""
|
| 91 |
+
seq = ""
|
| 92 |
+
for i in range(1, max(res_id_to_res) + 1):
|
| 93 |
+
if i in res_id_to_res:
|
| 94 |
+
seq += Bio.SeqUtils.seq1(res_id_to_res[i].get_resname())
|
| 95 |
+
else:
|
| 96 |
+
seq += "X"
|
| 97 |
+
return seq
|
| 98 |
+
|
| 99 |
+
|
| 100 |
+
def get_sequence_from_pdb(pdb_path: str) -> Tuple[str, List[int]]:
|
| 101 |
+
pdb_parser = Bio.PDB.PDBParser(QUIET=True)
|
| 102 |
+
pdb_struct = pdb_parser.get_structure("original_pdb", pdb_path)
|
| 103 |
+
# chain_to_seq = {chain.id: get_chain_object_to_seq(chain) for chain in pdb_struct.get_chains()}
|
| 104 |
+
all_chain_seqs = [get_chain_object_to_seq(chain) for chain in pdb_struct.get_chains()]
|
| 105 |
+
chain_lengths = [len(seq) for seq in all_chain_seqs]
|
| 106 |
+
return ("X" * 20).join(all_chain_seqs), chain_lengths
|
| 107 |
+
|
| 108 |
+
|
| 109 |
+
from Bio import PDB
|
| 110 |
+
from Bio import pairwise2
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
def extract_sequence(chain):
|
| 114 |
+
seq = ''
|
| 115 |
+
residues = []
|
| 116 |
+
for res in chain.get_residues():
|
| 117 |
+
seq_res = Bio.SeqUtils.seq1(res.get_resname())
|
| 118 |
+
if seq_res in ('X', "", " "):
|
| 119 |
+
continue
|
| 120 |
+
seq += seq_res
|
| 121 |
+
residues.append(res)
|
| 122 |
+
return seq, residues
|
| 123 |
+
|
| 124 |
+
|
| 125 |
+
def map_residues(alignment, residues_gt, residues_pred):
|
| 126 |
+
idx_gt = 0
|
| 127 |
+
idx_pred = 0
|
| 128 |
+
mapping = []
|
| 129 |
+
for i in range(len(alignment.seqA)):
|
| 130 |
+
aa_gt = alignment.seqA[i]
|
| 131 |
+
aa_pred = alignment.seqB[i]
|
| 132 |
+
res_gt = None
|
| 133 |
+
res_pred = None
|
| 134 |
+
if aa_gt != '-':
|
| 135 |
+
res_gt = residues_gt[idx_gt]
|
| 136 |
+
idx_gt += 1
|
| 137 |
+
if aa_pred != '-':
|
| 138 |
+
res_pred = residues_pred[idx_pred]
|
| 139 |
+
idx_pred += 1
|
| 140 |
+
if res_gt and res_pred:
|
| 141 |
+
mapping.append((res_gt, res_pred))
|
| 142 |
+
return mapping
|
| 143 |
+
|
| 144 |
+
|
| 145 |
+
class ResidueSelect(PDB.Select):
|
| 146 |
+
def __init__(self, residues_to_select):
|
| 147 |
+
self.residues_to_select = set(residues_to_select)
|
| 148 |
+
|
| 149 |
+
def accept_residue(self, residue):
|
| 150 |
+
return residue in self.residues_to_select
|
| 151 |
+
|
| 152 |
+
|
| 153 |
+
def count_gapped_single_aa(alignment):
|
| 154 |
+
count_non_gap = 0
|
| 155 |
+
count_fully_gapped = 0
|
| 156 |
+
for i in range(1, len(alignment.seqA) - 1):
|
| 157 |
+
if alignment.seqA[i] != '-':
|
| 158 |
+
count_non_gap += 1
|
| 159 |
+
if alignment.seqA[i - 1] == '-' and alignment.seqA[i + 1] == '-':
|
| 160 |
+
count_fully_gapped += 1
|
| 161 |
+
top_ratio = count_fully_gapped / count_non_gap
|
| 162 |
+
|
| 163 |
+
count_non_gap = 0
|
| 164 |
+
count_fully_gapped = 0
|
| 165 |
+
for i in range(1, len(alignment.seqB) - 1):
|
| 166 |
+
if alignment.seqA[i] != '-':
|
| 167 |
+
count_non_gap += 1
|
| 168 |
+
if alignment.seqA[i - 1] == '-' and alignment.seqA[i + 1] == '-':
|
| 169 |
+
count_fully_gapped += 1
|
| 170 |
+
|
| 171 |
+
if count_fully_gapped / count_non_gap > top_ratio:
|
| 172 |
+
top_ratio = count_fully_gapped / count_non_gap
|
| 173 |
+
|
| 174 |
+
return top_ratio
|
| 175 |
+
|
| 176 |
+
|
| 177 |
+
def copy_residue_numbering(gt_pdb_path, input_pdb_path):
|
| 178 |
+
parser = PDB.PDBParser(QUIET=True)
|
| 179 |
+
gt_structure = parser.get_structure('gt', gt_pdb_path)
|
| 180 |
+
input_structure = parser.get_structure('input', input_pdb_path)
|
| 181 |
+
|
| 182 |
+
for res in list(input_structure.get_residues()):
|
| 183 |
+
res.id = (' ', res.get_id()[1] + 10000, ' ')
|
| 184 |
+
|
| 185 |
+
for gt_res, input_res in zip(gt_structure.get_residues(), input_structure.get_residues()):
|
| 186 |
+
input_res.id = gt_res.id
|
| 187 |
+
|
| 188 |
+
io = PDB.PDBIO()
|
| 189 |
+
io.set_structure(input_structure)
|
| 190 |
+
io.save(input_pdb_path)
|
| 191 |
+
|
| 192 |
+
|
| 193 |
+
def align_gt_and_input(gt_pdb_path, input_pdb_path, output_gt_path, output_input_path):
|
| 194 |
+
# print("aligning", gt_pdb_path, input_pdb_path, output_gt_path, output_input_path)
|
| 195 |
+
parser = PDB.PDBParser(QUIET=True)
|
| 196 |
+
gt_structure = parser.get_structure('gt', gt_pdb_path)
|
| 197 |
+
pred_structure = parser.get_structure('pred', input_pdb_path)
|
| 198 |
+
matched_residues_gt = []
|
| 199 |
+
matched_residues_pred = []
|
| 200 |
+
|
| 201 |
+
total_gt_size = len([res for res in gt_structure.get_residues() if "CA" in res])
|
| 202 |
+
|
| 203 |
+
used_chain_pred = []
|
| 204 |
+
total_mapping_size = 0
|
| 205 |
+
for chain_gt in gt_structure.get_chains():
|
| 206 |
+
seq_gt, residues_gt = extract_sequence(chain_gt)
|
| 207 |
+
best_alignment = None
|
| 208 |
+
best_chain_pred = None
|
| 209 |
+
best_score = -1
|
| 210 |
+
best_residues_pred = None
|
| 211 |
+
# Find the best matching chain in pred
|
| 212 |
+
for chain_pred in pred_structure.get_chains():
|
| 213 |
+
# print("checking", chain_pred.get_id(), chain_gt.get_id())
|
| 214 |
+
if chain_pred in used_chain_pred:
|
| 215 |
+
continue
|
| 216 |
+
seq_pred, residues_pred = extract_sequence(chain_pred)
|
| 217 |
+
# print(seq_gt)
|
| 218 |
+
# print(seq_pred)
|
| 219 |
+
# alignments = pairwise2.align.globalxx(seq_gt, seq_pred, one_alignment_only=True)
|
| 220 |
+
alignments = pairwise2.align.globalms(seq_gt, seq_pred, 2, -10000, -1, 0, one_alignment_only=True)
|
| 221 |
+
if not alignments:
|
| 222 |
+
continue
|
| 223 |
+
# print("checking2", chain_pred.get_id(), chain_gt.get_id())
|
| 224 |
+
|
| 225 |
+
alignment = alignments[0]
|
| 226 |
+
score = alignment.score
|
| 227 |
+
if score > best_score:
|
| 228 |
+
best_score = score
|
| 229 |
+
best_alignment = alignment
|
| 230 |
+
best_chain_pred = chain_pred
|
| 231 |
+
best_residues_pred = residues_pred
|
| 232 |
+
if best_alignment and count_gapped_single_aa(best_alignment) < 0.2:
|
| 233 |
+
mapping = map_residues(best_alignment, residues_gt, best_residues_pred)
|
| 234 |
+
total_mapping_size += len(mapping)
|
| 235 |
+
used_chain_pred.append(best_chain_pred)
|
| 236 |
+
for res_gt, res_pred in mapping:
|
| 237 |
+
matched_residues_gt.append(res_gt)
|
| 238 |
+
matched_residues_pred.append(res_pred)
|
| 239 |
+
else:
|
| 240 |
+
print(f"No matching chain found for chain {chain_gt.get_id()}")
|
| 241 |
+
assert total_mapping_size / total_gt_size > 0.8, \
|
| 242 |
+
f"Mapping size too low ({total_mapping_size}/{total_gt_size}), skipping"
|
| 243 |
+
print(f"Total mapping size: {total_mapping_size}")
|
| 244 |
+
|
| 245 |
+
# Write new PDB files with only matched residues
|
| 246 |
+
io = PDB.PDBIO()
|
| 247 |
+
io.set_structure(gt_structure)
|
| 248 |
+
io.save(output_gt_path, ResidueSelect(matched_residues_gt))
|
| 249 |
+
io = PDB.PDBIO()
|
| 250 |
+
io.set_structure(pred_structure)
|
| 251 |
+
io.save(output_input_path, ResidueSelect(matched_residues_pred))
|
| 252 |
+
|
| 253 |
+
copy_residue_numbering(output_gt_path, output_input_path)
|
| 254 |
+
|
| 255 |
+
|
| 256 |
+
def validate_matching_input_gt(gt_pdb_path, input_pdb_path):
|
| 257 |
+
gt_residues = [res for res in PDB.PDBParser().get_structure('gt', gt_pdb_path).get_residues()]
|
| 258 |
+
input_residues = [res for res in PDB.PDBParser().get_structure('input', input_pdb_path).get_residues()]
|
| 259 |
+
|
| 260 |
+
if len(gt_residues) != len(input_residues):
|
| 261 |
+
print(f"Residue count mismatch: {len(gt_residues)} vs {len(input_residues)}")
|
| 262 |
+
return -1
|
| 263 |
+
|
| 264 |
+
for res_gt, res_input in zip(gt_residues, input_residues):
|
| 265 |
+
if res_gt.get_resname() != res_input.get_resname():
|
| 266 |
+
print(f"Residue name mismatch: {res_gt.get_resname()} vs {res_input.get_resname()}")
|
| 267 |
+
return -1
|
| 268 |
+
return len(input_residues)
|
| 269 |
+
|
| 270 |
+
|
| 271 |
+
def download_pdb(pdb_name, output_folder):
|
| 272 |
+
output_path = os.path.join(output_folder, pdb_name)
|
| 273 |
+
if os.path.exists(output_path):
|
| 274 |
+
return output_path
|
| 275 |
+
print("downloading", pdb_name)
|
| 276 |
+
os.system(f'{GSUTIL_PATH} -m -q cp "gs://pinder/2024-02/pdbs/{pdb_name}" {output_path}')
|
| 277 |
+
return output_path
|
| 278 |
+
|
| 279 |
+
|
| 280 |
+
INTERFACE_MIN_ATOM_DIST = 5
|
| 281 |
+
|
| 282 |
+
|
| 283 |
+
def get_filtered_res(gt_r_res, gt_l_res, max_length: int):
|
| 284 |
+
gt_r_ca = np.array([res["CA"].coord for res in gt_r_res])
|
| 285 |
+
gt_l_ca = np.array([res["CA"].coord for res in gt_l_res])
|
| 286 |
+
|
| 287 |
+
if len(gt_r_res) + len(gt_l_res) < max_length:
|
| 288 |
+
# continue without cropping
|
| 289 |
+
print("no cropping needed", len(gt_r_res), len(gt_l_res))
|
| 290 |
+
return gt_r_res, gt_l_res
|
| 291 |
+
|
| 292 |
+
# close_residues = np.argwhere(scipy.spatial.distance.cdist(gt_r_ca, gt_l_ca) < INTERFACE_MIN_ATOM_DIST)
|
| 293 |
+
# gt_r_interface, gt_l_interface = set(), set()
|
| 294 |
+
# for i, j in close_residues:
|
| 295 |
+
# gt_r_interface.add(gt_r_res[i].id[1])
|
| 296 |
+
# gt_l_interface.add(gt_l_res[j].id[1])
|
| 297 |
+
|
| 298 |
+
inter_dists = gt_r_ca[:, np.newaxis, :] - gt_l_ca[np.newaxis, :, :]
|
| 299 |
+
inter_dists = np.sqrt((inter_dists ** 2).sum(-1))
|
| 300 |
+
min_inter_dist_per_gt_l_res = inter_dists.min(axis=0)
|
| 301 |
+
min_inter_dist_per_gt_r_res = inter_dists.min(axis=1)
|
| 302 |
+
|
| 303 |
+
assert min_inter_dist_per_gt_l_res.shape[0] == len(gt_l_res)
|
| 304 |
+
assert min_inter_dist_per_gt_r_res.shape[0] == len(gt_r_res)
|
| 305 |
+
|
| 306 |
+
min_r_res, max_r_res = min(min_inter_dist_per_gt_r_res), max(min_inter_dist_per_gt_r_res)
|
| 307 |
+
min_l_res, max_l_res = min(min_inter_dist_per_gt_l_res), max(min_inter_dist_per_gt_l_res)
|
| 308 |
+
|
| 309 |
+
r_pocket = [res for res in gt_r_res if min_r_res <= res.id[1] <= max_r_res]
|
| 310 |
+
l_pocket = [res for res in gt_l_res if min_l_res <= res.id[1] <= max_l_res]
|
| 311 |
+
|
| 312 |
+
if len(r_pocket) + len(l_pocket) < max_length:
|
| 313 |
+
# add extra residues to both chains to get a total of max_length
|
| 314 |
+
res_r_before = [res for res in gt_r_res if res.id[1] < min_r_res]
|
| 315 |
+
res_r_after = [res for res in gt_r_res if res.id[1] > max_r_res]
|
| 316 |
+
res_l_before = [res for res in gt_l_res if res.id[1] < min_l_res]
|
| 317 |
+
res_l_after = [res for res in gt_l_res if res.id[1] > max_l_res]
|
| 318 |
+
|
| 319 |
+
extra_to_add = max_length - len(r_pocket) - len(l_pocket)
|
| 320 |
+
|
| 321 |
+
actions = []
|
| 322 |
+
if len(res_r_before) > 0:
|
| 323 |
+
actions.append("add_r_before")
|
| 324 |
+
if len(res_r_after) > 0:
|
| 325 |
+
actions.append("add_r_after")
|
| 326 |
+
if len(res_l_before) > 0:
|
| 327 |
+
actions.append("add_l_before")
|
| 328 |
+
if len(res_l_after) > 0:
|
| 329 |
+
actions.append("add_l_after")
|
| 330 |
+
while extra_to_add > 0 and actions:
|
| 331 |
+
action = random.choice(actions)
|
| 332 |
+
|
| 333 |
+
if action == "add_r_before":
|
| 334 |
+
r_pocket.insert(0, res_r_before.pop())
|
| 335 |
+
if not len(res_r_before):
|
| 336 |
+
actions.remove("add_r_before")
|
| 337 |
+
elif action == "add_r_after":
|
| 338 |
+
r_pocket.append(res_r_after.pop())
|
| 339 |
+
if not len(res_r_after):
|
| 340 |
+
actions.remove("add_r_after")
|
| 341 |
+
elif action == "add_l_before":
|
| 342 |
+
l_pocket.insert(0, res_l_before.pop())
|
| 343 |
+
if not len(res_l_before):
|
| 344 |
+
actions.remove("add_l_before")
|
| 345 |
+
elif action == "add_l_after":
|
| 346 |
+
l_pocket.append(res_l_after.pop())
|
| 347 |
+
if not len(res_l_after):
|
| 348 |
+
actions.remove("add_l_after")
|
| 349 |
+
extra_to_add -= 1
|
| 350 |
+
print("Extended pocket sizes", len(r_pocket), len(l_pocket), "extra_to_add", extra_to_add)
|
| 351 |
+
return r_pocket, l_pocket
|
| 352 |
+
|
| 353 |
+
print("cropping simply")
|
| 354 |
+
# remove residues that are farthest from the interface
|
| 355 |
+
res_and_dist_r = [(res, min_inter_dist_per_gt_r_res[res_idx]) for res_idx, res in enumerate(gt_r_res)]
|
| 356 |
+
res_and_dist_l = [(res, min_inter_dist_per_gt_l_res[res_idx]) for res_idx, res in enumerate(gt_l_res)]
|
| 357 |
+
|
| 358 |
+
res_and_dist_r = [(res, dist) for res, dist in res_and_dist_r if res in r_pocket]
|
| 359 |
+
res_and_dist_l = [(res, dist) for res, dist in res_and_dist_l if res in l_pocket]
|
| 360 |
+
|
| 361 |
+
res_and_dist_r = sorted(res_and_dist_r, key=lambda x: x[1], reverse=True)
|
| 362 |
+
res_and_dist_l = sorted(res_and_dist_l, key=lambda x: x[1], reverse=True)
|
| 363 |
+
|
| 364 |
+
while len(res_and_dist_r) + len(res_and_dist_l) > max_length:
|
| 365 |
+
if res_and_dist_r[0][1] > res_and_dist_l[0][1]:
|
| 366 |
+
res_and_dist_r.pop(0)
|
| 367 |
+
else:
|
| 368 |
+
res_and_dist_l.pop(0)
|
| 369 |
+
|
| 370 |
+
return [res for res, _ in res_and_dist_r], [res for res, _ in res_and_dist_l]
|
| 371 |
+
|
| 372 |
+
|
| 373 |
+
def prepare_holo(row, tmp_dir_path, max_length: int):
|
| 374 |
+
tmp_gt_r_pdb = os.path.join(tmp_dir_path, f"tmp_{row.id}_gt_r.pdb")
|
| 375 |
+
tmp_gt_l_pdb = os.path.join(tmp_dir_path, f"tmp_{row.id}_gt_l.pdb")
|
| 376 |
+
|
| 377 |
+
if os.path.exists(tmp_gt_r_pdb) and os.path.exists(tmp_gt_l_pdb):
|
| 378 |
+
return tmp_gt_r_pdb, tmp_gt_l_pdb
|
| 379 |
+
|
| 380 |
+
holo_r_pdb = download_pdb(row.holo_R_pdb, tmp_dir_path)
|
| 381 |
+
holo_l_pdb = download_pdb(row.holo_L_pdb, tmp_dir_path)
|
| 382 |
+
|
| 383 |
+
# make gt and apo that matches
|
| 384 |
+
robust_renumber_protein(holo_r_pdb, tmp_gt_r_pdb)
|
| 385 |
+
robust_renumber_protein(holo_l_pdb, tmp_gt_l_pdb)
|
| 386 |
+
|
| 387 |
+
parser = PDB.PDBParser(QUIET=True)
|
| 388 |
+
gt_r_prot = parser.get_structure('r', tmp_gt_r_pdb)
|
| 389 |
+
gt_l_prot = parser.get_structure('l', tmp_gt_l_pdb)
|
| 390 |
+
|
| 391 |
+
assert len(list(gt_r_prot.get_chains())) == 1, "can't extract if more than one chain"
|
| 392 |
+
assert len(list(gt_l_prot.get_chains())) == 1, "can't extract if more than one chain"
|
| 393 |
+
|
| 394 |
+
gt_r_res = [res for res in gt_r_prot.get_residues() if "CA" in res]
|
| 395 |
+
gt_l_res = [res for res in gt_l_prot.get_residues() if "CA" in res]
|
| 396 |
+
|
| 397 |
+
to_keep_r, to_keep_l = get_filtered_res(gt_r_res, gt_l_res, max_length)
|
| 398 |
+
|
| 399 |
+
io = PDB.PDBIO()
|
| 400 |
+
io.set_structure(gt_r_prot)
|
| 401 |
+
io.save(tmp_gt_r_pdb, ResidueSelect(to_keep_r))
|
| 402 |
+
io = PDB.PDBIO()
|
| 403 |
+
io.set_structure(gt_l_prot)
|
| 404 |
+
io.save(tmp_gt_l_pdb, ResidueSelect(to_keep_l))
|
| 405 |
+
|
| 406 |
+
return tmp_gt_r_pdb, tmp_gt_l_pdb
|
| 407 |
+
|
| 408 |
+
|
| 409 |
+
def generate_input_pdbs(tmp_input_r_pdb, tmp_input_l_pdb, tmp_gt_r_pdb, tmp_gt_l_pdb,
|
| 410 |
+
input_r_output_pdb, input_l_output_pdb, gt_r_output_pdb, gt_l_output_pdb):
|
| 411 |
+
# print("preparing input pdbs", gt_r_output_pdb)
|
| 412 |
+
if not os.path.exists(tmp_input_r_pdb) or not os.path.exists(tmp_input_l_pdb):
|
| 413 |
+
raise False
|
| 414 |
+
|
| 415 |
+
try:
|
| 416 |
+
align_gt_and_input(tmp_gt_r_pdb, tmp_input_r_pdb, gt_r_output_pdb, input_r_output_pdb)
|
| 417 |
+
protein_size_r = validate_matching_input_gt(gt_r_output_pdb, input_r_output_pdb)
|
| 418 |
+
assert protein_size_r > -1, "Failed to validate matching input and gt"
|
| 419 |
+
|
| 420 |
+
align_gt_and_input(tmp_gt_l_pdb, tmp_input_l_pdb, gt_l_output_pdb, input_l_output_pdb)
|
| 421 |
+
protein_size_l = validate_matching_input_gt(gt_l_output_pdb, input_l_output_pdb)
|
| 422 |
+
assert protein_size_l > -1, "Failed to validate matching input and gt"
|
| 423 |
+
except Exception as e:
|
| 424 |
+
print("Failed to align", e)
|
| 425 |
+
if os.path.exists(gt_r_output_pdb):
|
| 426 |
+
os.remove(gt_r_output_pdb)
|
| 427 |
+
if os.path.exists(gt_l_output_pdb):
|
| 428 |
+
os.remove(gt_l_output_pdb)
|
| 429 |
+
if os.path.exists(input_r_output_pdb):
|
| 430 |
+
os.remove(input_r_output_pdb)
|
| 431 |
+
if os.path.exists(input_l_output_pdb):
|
| 432 |
+
os.remove(input_l_output_pdb)
|
| 433 |
+
return False
|
| 434 |
+
|
| 435 |
+
return True
|
| 436 |
+
|
| 437 |
+
|
| 438 |
+
def _get_rel_path(abs_path):
|
| 439 |
+
return os.path.join(os.path.basename(os.path.dirname(abs_path)), os.path.basename(abs_path))
|
| 440 |
+
|
| 441 |
+
|
| 442 |
+
def main(start_ind: Optional[int] = None, end_ind: Optional[int] = None):
|
| 443 |
+
print("running with", start_ind, end_ind)
|
| 444 |
+
|
| 445 |
+
os.makedirs(OUTPUT_FOLDER, exist_ok=True)
|
| 446 |
+
output_models_folder = os.path.join(OUTPUT_FOLDER, "pinder_models")
|
| 447 |
+
output_train_jsons_folder = os.path.join(OUTPUT_FOLDER, "pinder_jsons_train")
|
| 448 |
+
output_val_jsons_folder = os.path.join(OUTPUT_FOLDER, "pinder_jsons_val")
|
| 449 |
+
output_test_jsons_folder = os.path.join(OUTPUT_FOLDER, "pinder_jsons_test")
|
| 450 |
+
output_info = os.path.join(OUTPUT_FOLDER, "pinder_generation_info.csv")
|
| 451 |
+
|
| 452 |
+
os.makedirs(output_models_folder, exist_ok=True)
|
| 453 |
+
os.makedirs(output_train_jsons_folder, exist_ok=True)
|
| 454 |
+
os.makedirs(output_val_jsons_folder, exist_ok=True)
|
| 455 |
+
os.makedirs(output_test_jsons_folder, exist_ok=True)
|
| 456 |
+
|
| 457 |
+
split_to_folder = {
|
| 458 |
+
"train": output_train_jsons_folder,
|
| 459 |
+
"val": output_val_jsons_folder,
|
| 460 |
+
"test": output_test_jsons_folder
|
| 461 |
+
}
|
| 462 |
+
|
| 463 |
+
# output_info_file = open(output_info, "a+")
|
| 464 |
+
systems = pd.read_parquet(PINDER_ANNOTATIONS)
|
| 465 |
+
systems = systems[systems.split.isin(['train', 'val', 'test'])]
|
| 466 |
+
|
| 467 |
+
cluster_ids = systems["cluster_id"].value_counts()
|
| 468 |
+
cluster_ids = cluster_ids[cluster_ids >= 1]
|
| 469 |
+
print("There are", len(cluster_ids), "clusters")
|
| 470 |
+
|
| 471 |
+
# clusters_with_data = 0
|
| 472 |
+
# for cluster_id in cluster_ids.index:
|
| 473 |
+
# cluster_systems = systems[systems["cluster_id"] == cluster_id]
|
| 474 |
+
# with_apo = cluster_systems[cluster_systems.apo_R & cluster_systems.apo_L]
|
| 475 |
+
# if len(with_apo) > 0:
|
| 476 |
+
# print("Cluster", cluster_id, "has", len(with_apo), "systems with apo")
|
| 477 |
+
# clusters_with_data += 1
|
| 478 |
+
# continue
|
| 479 |
+
# with_pred = cluster_systems[cluster_systems.predicted_R & cluster_systems.predicted_L]
|
| 480 |
+
# if len(with_pred) > 0:
|
| 481 |
+
# print("Cluster", cluster_id, "has", len(with_pred), "systems with pred")
|
| 482 |
+
# clusters_with_data += 1
|
| 483 |
+
# continue
|
| 484 |
+
# print("There are", clusters_with_data, "clusters with data out of", len(cluster_ids))
|
| 485 |
+
|
| 486 |
+
for cluster_ind, cluster_id in enumerate(sorted(cluster_ids.index)):
|
| 487 |
+
if (start_ind is not None and cluster_ind < start_ind) or (end_ind is not None and cluster_ind >= end_ind):
|
| 488 |
+
continue
|
| 489 |
+
# if cluster_id != "cluster_10004_p":
|
| 490 |
+
# continue
|
| 491 |
+
tmp_dir_path = os.path.join(OUTPUT_FOLDER, "tmp_" + cluster_id)
|
| 492 |
+
os.makedirs(tmp_dir_path, exist_ok=True)
|
| 493 |
+
system_id_to_method = {}
|
| 494 |
+
|
| 495 |
+
cluster_systems = systems[systems["cluster_id"] == cluster_id]
|
| 496 |
+
print("--- Starting cluster", cluster_ind, cluster_id, "size", cluster_systems.shape)
|
| 497 |
+
|
| 498 |
+
with_apo = cluster_systems[cluster_systems.apo_R & cluster_systems.apo_L]
|
| 499 |
+
print("*** APO *** Cluster", cluster_id, "has", len(with_apo), "systems with apo")
|
| 500 |
+
for try_counter, row in enumerate(with_apo.itertuples()):
|
| 501 |
+
if row.split not in ("test", "val") \
|
| 502 |
+
and (try_counter >= MAX_TRIES_OF_METHOD or len(system_id_to_method) >= MAX_SYSTEMS_FOR_CLUSTER):
|
| 503 |
+
continue
|
| 504 |
+
print("-- Trying to prepare apo", row.id, row.split)
|
| 505 |
+
try:
|
| 506 |
+
tmp_gt_r_pdb, tmp_gt_l_pdb = prepare_holo(row, tmp_dir_path, MAX_LENGTH)
|
| 507 |
+
|
| 508 |
+
gt_r_output_path = os.path.join(output_models_folder, f"{row.id}_gt_r.pdb")
|
| 509 |
+
gt_l_output_path = os.path.join(output_models_folder, f"{row.id}_gt_l.pdb")
|
| 510 |
+
|
| 511 |
+
input_r_output_path = os.path.join(output_models_folder, f"{row.id}_input_r.pdb")
|
| 512 |
+
input_l_output_path = os.path.join(output_models_folder, f"{row.id}_input_l.pdb")
|
| 513 |
+
|
| 514 |
+
input_r_pdb_path = download_pdb(row.apo_R_pdb, tmp_dir_path)
|
| 515 |
+
input_l_pdb_path = download_pdb(row.apo_L_pdb, tmp_dir_path)
|
| 516 |
+
|
| 517 |
+
if generate_input_pdbs(input_r_pdb_path, input_l_pdb_path, tmp_gt_r_pdb, tmp_gt_l_pdb,
|
| 518 |
+
input_r_output_path, input_l_output_path, gt_r_output_path, gt_l_output_path):
|
| 519 |
+
system_id_to_method[row.id] = "apo"
|
| 520 |
+
|
| 521 |
+
except Exception as e:
|
| 522 |
+
print("Failed to prepare apo", row.id, e)
|
| 523 |
+
continue
|
| 524 |
+
|
| 525 |
+
with_pred = cluster_systems[cluster_systems.predicted_R & cluster_systems.predicted_L]
|
| 526 |
+
print("*** Pred *** Cluster", cluster_id, "has", len(with_pred), "systems with pred")
|
| 527 |
+
for try_counter, row in enumerate(with_pred.itertuples()):
|
| 528 |
+
if row.id in system_id_to_method:
|
| 529 |
+
continue
|
| 530 |
+
if row.split not in ("test", "val") \
|
| 531 |
+
and (try_counter >= MAX_TRIES_OF_METHOD or len(system_id_to_method) >= MAX_SYSTEMS_FOR_CLUSTER):
|
| 532 |
+
continue
|
| 533 |
+
print("-- Trying to prepare pred", row.id, row.split)
|
| 534 |
+
try:
|
| 535 |
+
tmp_gt_r_pdb, tmp_gt_l_pdb = prepare_holo(row, tmp_dir_path, MAX_LENGTH)
|
| 536 |
+
|
| 537 |
+
gt_r_output_path = os.path.join(output_models_folder, f"{row.id}_gt_r.pdb")
|
| 538 |
+
gt_l_output_path = os.path.join(output_models_folder, f"{row.id}_gt_l.pdb")
|
| 539 |
+
|
| 540 |
+
input_r_output_path = os.path.join(output_models_folder, f"{row.id}_input_r.pdb")
|
| 541 |
+
input_l_output_path = os.path.join(output_models_folder, f"{row.id}_input_l.pdb")
|
| 542 |
+
|
| 543 |
+
input_r_pdb_path = download_pdb(row.predicted_R_pdb, tmp_dir_path)
|
| 544 |
+
input_l_pdb_path = download_pdb(row.predicted_L_pdb, tmp_dir_path)
|
| 545 |
+
|
| 546 |
+
if generate_input_pdbs(input_r_pdb_path, input_l_pdb_path, tmp_gt_r_pdb, tmp_gt_l_pdb,
|
| 547 |
+
input_r_output_path, input_l_output_path, gt_r_output_path, gt_l_output_path):
|
| 548 |
+
system_id_to_method[row.id] = "pred"
|
| 549 |
+
|
| 550 |
+
except Exception as e:
|
| 551 |
+
print("Failed to prepare pred", row.id, e)
|
| 552 |
+
|
| 553 |
+
# default - use gt
|
| 554 |
+
print("*** GT *** ")
|
| 555 |
+
for row in cluster_systems.itertuples():
|
| 556 |
+
if row.id in system_id_to_method:
|
| 557 |
+
continue
|
| 558 |
+
if row.split not in ("test", "val") and len(system_id_to_method) >= MAX_SYSTEMS_FOR_CLUSTER:
|
| 559 |
+
continue
|
| 560 |
+
try:
|
| 561 |
+
tmp_gt_r_pdb, tmp_gt_l_pdb = prepare_holo(row, tmp_dir_path, MAX_LENGTH)
|
| 562 |
+
|
| 563 |
+
gt_r_output_path = os.path.join(output_models_folder, f"{row.id}_gt_r.pdb")
|
| 564 |
+
gt_l_output_path = os.path.join(output_models_folder, f"{row.id}_gt_l.pdb")
|
| 565 |
+
|
| 566 |
+
input_r_output_path = os.path.join(output_models_folder, f"{row.id}_input_r.pdb")
|
| 567 |
+
input_l_output_path = os.path.join(output_models_folder, f"{row.id}_input_l.pdb")
|
| 568 |
+
|
| 569 |
+
shutil.copyfile(tmp_gt_r_pdb, gt_r_output_path)
|
| 570 |
+
shutil.copyfile(tmp_gt_r_pdb, input_r_output_path)
|
| 571 |
+
|
| 572 |
+
shutil.copyfile(tmp_gt_l_pdb, gt_l_output_path)
|
| 573 |
+
shutil.copyfile(tmp_gt_l_pdb, input_l_output_path)
|
| 574 |
+
|
| 575 |
+
system_id_to_method[row.id] = "gt"
|
| 576 |
+
|
| 577 |
+
except Exception as e:
|
| 578 |
+
print("Failed to prepare gt", row.id, e)
|
| 579 |
+
|
| 580 |
+
# save jsons
|
| 581 |
+
for row in cluster_systems.itertuples():
|
| 582 |
+
if row.id not in system_id_to_method:
|
| 583 |
+
continue
|
| 584 |
+
|
| 585 |
+
output_json_path = os.path.join(split_to_folder[row.split], f"{row.id}.json")
|
| 586 |
+
|
| 587 |
+
gt_r_output_path = os.path.join(output_models_folder, f"{row.id}_gt_r.pdb")
|
| 588 |
+
gt_l_output_path = os.path.join(output_models_folder, f"{row.id}_gt_l.pdb")
|
| 589 |
+
|
| 590 |
+
input_r_output_path = os.path.join(output_models_folder, f"{row.id}_input_r.pdb")
|
| 591 |
+
input_l_output_path = os.path.join(output_models_folder, f"{row.id}_input_l.pdb")
|
| 592 |
+
|
| 593 |
+
protein_r_seq_len = validate_matching_input_gt(gt_r_output_path, input_r_output_path)
|
| 594 |
+
protein_l_seq_len = validate_matching_input_gt(gt_l_output_path, input_l_output_path)
|
| 595 |
+
|
| 596 |
+
json_data = {
|
| 597 |
+
"input_r_structure": _get_rel_path(input_r_output_path),
|
| 598 |
+
"input_l_structure": _get_rel_path(input_l_output_path),
|
| 599 |
+
"gt_r_structure": _get_rel_path(gt_r_output_path),
|
| 600 |
+
"gt_l_structure": _get_rel_path(gt_l_output_path),
|
| 601 |
+
"resolution": 1.0,
|
| 602 |
+
"protein_r_seq_len": protein_r_seq_len,
|
| 603 |
+
"protein_l_seq_len": protein_l_seq_len,
|
| 604 |
+
"uniprot_r": row.uniprot_R,
|
| 605 |
+
"uniprot_l": row.uniprot_L,
|
| 606 |
+
"cluster": row.cluster_id,
|
| 607 |
+
"input_protein_source": system_id_to_method[row.id],
|
| 608 |
+
"pdb_id": row.id,
|
| 609 |
+
}
|
| 610 |
+
open(output_json_path, "w").write(json.dumps(json_data, indent=4))
|
| 611 |
+
|
| 612 |
+
print("******* saved", row.id, system_id_to_method[row.id], flush=True)
|
| 613 |
+
shutil.rmtree(tmp_dir_path)
|
| 614 |
+
|
| 615 |
+
print("total systems", len(systems))
|
| 616 |
+
|
| 617 |
+
|
| 618 |
+
if __name__ == '__main__':
|
| 619 |
+
if len(sys.argv) == 3:
|
| 620 |
+
main(int(sys.argv[1]), int(sys.argv[2]))
|
| 621 |
+
else:
|
| 622 |
+
main()
|
resources/{77-182500_only_weights.ckpt → only_weights_102-240750.ckpt}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e32cd27b63f684ed813e65db4369f5af28b1fceb3c81df66bdd4952f6b78853
|
| 3 |
+
size 52579856
|