Spaces:
Sleeping
Sleeping
File size: 94,929 Bytes
c2a02c6 699a8e2 c2a02c6 bcf16e9 c2a02c6 9215a83 c2a02c6 f4dc3e4 9215a83 c2a02c6 f4dc3e4 c2a02c6 869b53c f4dc3e4 869b53c f4dc3e4 869b53c 9215a83 869b53c 9215a83 af56dfe 869b53c 9215a83 869b53c f4dc3e4 869b53c d1d665f 869b53c f4dc3e4 869b53c c2a02c6 460d291 869b53c f4dc3e4 869b53c d1d665f f4dc3e4 5f06ca7 83997e9 5f06ca7 3c0921a ffb4736 869b53c f4dc3e4 9e930bc c2a02c6 f4dc3e4 add08b1 f4dc3e4 6d7420f f4dc3e4 4c168c7 29503f2 e27a8d4 29503f2 aba4294 4c168c7 0b6d591 4c168c7 f4dc3e4 11ae51f 0b6d591 f4dc3e4 4c168c7 f4dc3e4 4c168c7 af56dfe 4c168c7 11ae51f 0b6d591 4c168c7 bacea2c 869b53c c2a02c6 869b53c af56dfe 869b53c af56dfe 869b53c c2a02c6 f4dc3e4 869b53c c2a02c6 869b53c 9215a83 869b53c c2a02c6 869b53c 9215a83 f4dc3e4 869b53c f4dc3e4 9215a83 869b53c 9215a83 869b53c f4dc3e4 869b53c f4dc3e4 d2fda53 869b53c 92d57b2 af56dfe 869b53c f1b36e4 af56dfe bc51ec4 7b95696 869b53c af56dfe bacea2c 869b53c f4dc3e4 460d291 869b53c f4dc3e4 869b53c f4dc3e4 869b53c 9215a83 869b53c 9215a83 c2a02c6 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c f4dc3e4 869b53c f4dc3e4 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c f4dc3e4 869b53c f4dc3e4 869b53c f4dc3e4 869b53c f4dc3e4 869b53c f4dc3e4 869b53c f4dc3e4 869b53c f4dc3e4 869b53c c2a02c6 9215a83 869b53c f4dc3e4 9215a83 869b53c f4dc3e4 869b53c f4dc3e4 869b53c 6329d4b 869b53c 3ab87a2 869b53c d725c32 3ab87a2 869b53c 8337c15 3ab87a2 6c5efb0 80ef864 869b53c f4dc3e4 869b53c 9215a83 869b53c f4dc3e4 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 c2a02c6 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 9215a83 869b53c 460d291 869b53c f4dc3e4 9215a83 869b53c f4dc3e4 869b53c 152bc0f f4dc3e4 869b53c e9190e4 869b53c f4dc3e4 869b53c c2a02c6 9215a83 f4dc3e4 |
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 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 |
# IMPORT NECESSARY MODULES AND LIBRARIES
from timeit import default_timer as timer
import xml.etree.ElementTree as ET
from collections import Counter
from bs4 import BeautifulSoup
from io import StringIO
from decimal import *
import pandas as pd
import requests
import os.path as op
import subprocess
import shutil
import ssbio.utils
import warnings
import sys
import pathlib
from pathlib import Path
import os, glob
import math
import ssbio
import ssl
from Bio.Align import substitution_matrices
from Bio.PDB.Polypeptide import *
from Bio.PDB import PDBList
from Bio import Align
from Bio import SeqIO
from Bio.PDB import *
from Bio.PDB import PDBParser, PPBuilder
warnings.filterwarnings("ignore")
start = timer()
import streamlit as st
# FUNCTIONS
# FUNCTIONS
from calc_pc_property import *
from add_domains import *
from add_annotations import *
from add_sequence import *
from add_structure import *
from add_alignment import *
from manage_files import *
from add_3Dalignment import *
from add_sasa import *
from standard import *
from add_interface_pos import *
from standard import *
from uniprotSequenceMatch import uniprotSequenceMatch
from process_input import clean_data
def pdb(input_set, mode, impute):
aligner = Align.PairwiseAligner()
"""
STEP 1
Get input data as a console input.
Add datapoint identifier and remove non-standard input.
"""
data = clean_data(input_set)
path_to_input_files, path_to_output_files, path_to_domains, fisher_path, path_to_interfaces, buffer = manage_files(
mode)
out_path = path_to_output_files / 'log.txt'
print('Creating directories...')
annotation_list = ['disulfide', 'intMet', 'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink', 'mutagenesis', 'strand',
'helix', 'turn', 'metalBinding', 'repeat', 'topologicalDomain', 'caBinding', 'bindingSite',
'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil', 'peptide',
'transitPeptide', 'glycosylation', 'propeptide']
print('Feature vector generation started...\n')
if len(data) == 0:
print('Feature vectore generation terminated.')
else:
"""
STEP 2
Add physicochemical properties.
"""
print('Adding physicochemical properties...\n')
data = add_physicochemical(data)
"""
STEP 3
Add domain-related information.
"""
print('Adding domains\n')
data = add_domains(data, path_to_domains)
data = data.astype(str)
data = data.replace({'NaN': 'nan'})
data.domain = data.domain.replace({'nan': '-1'})
data.domStart = data.domStart.replace({'nan': '-1'})
data.domEnd = data.domEnd.replace({'nan': '-1'})
data.distance = data.distance.replace({'nan': '-1'})
"""
STEP 4
Retrieve canonical and isoform UniProt sequences.
Add to the data frame.
"""
print('Retrieving UniProt sequences...\n')
canonical_fasta = pd.DataFrame(columns=['uniprotID', 'uniprotSequence'])
up_list = list(set(data['uniprotID'].to_list()))
for i in range(len(up_list)):
canonical_fasta.at[i, 'uniprotSequence'] = get_uniprot_seq(up_list[i])
canonical_fasta.at[i, 'uniprotID'] = up_list[i]
canonical_fasta = canonical_fasta.drop_duplicates()
isoform_fasta = pd.DataFrame(columns=['uniprotID', 'isoformSequence'])
iso_dict = []
for i in range(len(up_list)):
iso_dict.append(get_isoforms(up_list[i]))
index = 0
for i in iso_dict:
for key, val in i.items():
isoform_fasta.at[index, 'uniprotID'] = key
isoform_fasta.at[index, 'isoformSequence'] = val
index += 1
isoform_fasta = isoform_fasta.drop_duplicates()
for i in isoform_fasta.index:
isoform_fasta.at[i, 'whichIsoform'] = isoform_fasta.at[i, 'uniprotID'][7:10].strip()
isoform_fasta.at[i, 'uniprotID'] = isoform_fasta.at[i, 'uniprotID'][0:6]
print('Sequence files created...\n')
data = data.merge(canonical_fasta, on='uniprotID', how='left')
data = data.astype(str)
data['whichIsoform'] = 'nan'
data.replace({'': 'nan'}, inplace=True)
data['wt_sequence_match'] = ''
for i in data.index:
if len(data.at[i, 'uniprotSequence']) >= int(data.at[i, 'pos']):
wt = data.at[i, 'wt']
can = str(data.at[i, 'uniprotSequence'])[int(data.at[i, 'pos']) - 1]
if wt == can:
data.at[i, 'wt_sequence_match'] = 'm'
elif wt != can:
isoList = isoform_fasta[
isoform_fasta['uniprotID'] == data.at[i, 'uniprotID']].isoformSequence.to_list()
for k in isoList:
if len(k) >= int(data.at[i, 'pos']):
resInIso = k[int(int(data.at[i, 'pos']) - 1)]
if wt == resInIso:
whichIsoform = isoform_fasta[isoform_fasta.isoformSequence == k].whichIsoform.to_list()[
0]
data.at[i, 'wt_sequence_match'] = 'i'
data.at[i, 'whichIsoform'] = whichIsoform
break
elif len(data.at[i, 'uniprotSequence']) < int(data.at[i, 'pos']):
isoList = isoform_fasta[isoform_fasta['uniprotID'] == data.at[i, 'uniprotID']].isoformSequence.to_list()
for k in isoList:
if len(k) >= int(data.at[i, 'pos']):
resInIso = k[int(int(data.at[i, 'pos']) - 1)]
wt = data.at[i, 'wt']
if wt == resInIso:
whichIsoform = isoform_fasta[isoform_fasta.isoformSequence == k].whichIsoform.to_list()[0]
data.at[i, 'wt_sequence_match'] = 'i'
data.at[i, 'whichIsoform'] = whichIsoform
break
data.wt_sequence_match = data.wt_sequence_match.astype('str')
data.replace({'': 'nan'}, inplace=True)
data_size = len(data.drop_duplicates(['datapoint']))
not_match_in_uniprot = data[(data.uniprotSequence == 'nan') | (data.wt_sequence_match == 'nan')]
uniprot_matched = data[(data.uniprotSequence != 'nan') & (data.wt_sequence_match != 'nan')]
data = None
print('You have %d data points that failed to match a UniProt Sequence\nProceeding with %d remaining...\n'
% (len(not_match_in_uniprot.drop_duplicates(['datapoint'])),
len(uniprot_matched.drop_duplicates(['datapoint']))))
"""
STEP 5
Retrieve related PDB sequences, extract their sequences.
Add to the data frame.
"""
from urllib.error import HTTPError
pdb_fasta = pd.DataFrame(columns=['pdbID', 'chain', 'pdbSequence'])
pdb_info = pd.DataFrame(columns=['uniprotID', 'pdbID', 'chain', 'resolution'])
print('Retrieving PDB structures...\n')
pdbs = []
protein = uniprot_matched.uniprotID.to_list()
protein = list(set(protein))
for prot in protein:
pdbs.append(get_pdb_ids(prot))
if len(pdbs) >= 1:
pdbs = [item for sublist in pdbs for item in sublist]
else:
pdbs = []
print('Processing PDB structures...\n')
if pdbs == []:
print('No PDB structure found for the query. ')
print('Starting PDB structures download...\n')
pdbs = list(filter(None, pdbs))
pdbs = (set(pdbs))
pdbs = [i.lower() for i in pdbs]
pdbl = PDBList()
parser = PDBParser()
index = 0
try:
shutil.rmtree('obsolete')
except OSError as e:
pass
cnt = 0
pdbs = [i.upper() for i in pdbs]
def fetch_uniprot_ids(pdb_code):
response = requests.get(f"https://www.ebi.ac.uk/pdbe/api/mappings/uniprot/{pdb_code}")
response.raise_for_status() # Check for a successful response
data = response.json()
return list(list(list(data.values())[0].values())[0].keys())
for search in pdbs:
# Step 1: Fetch the PDB file
pdb_url = f"https://files.rcsb.org/download/{search}.pdb"
try:
response = requests.get(pdb_url)
response.raise_for_status() # Check for a successful response
except :
continue # Skip to the next PDB code if fetching fails
# Step 2: Parse the PDB file from memory
pdb_data = response.text
pdb_parser = PDBParser(QUIET=True) # QUIET=True suppresses warnings
pdb_file_content = StringIO(pdb_data)
structure = pdb_parser.get_structure(search, pdb_file_content)
ppb = PPBuilder()
pdb_data_list = pdb_data.split('\n')
pdb_data_list_sequence = [i for i in pdb_data_list if i.startswith('SEQRES')]
pdb_data_list_sequence = [ list(filter(None,i.split(' '))) for i in pdb_data_list_sequence]
seqs = {}
for i in pdb_data_list_sequence:
if i[2] in seqs.keys():
seqs[i[2]] += i[4:]
else:
seqs[i[2]] = i[4:]
for key, val in seqs.items():
seqs[key] = ''.join([threeToOne(i) for i in val])
pdb_data_list = [i for i in pdb_data_list if i.startswith('DBREF')]
pdb_data_list = [[list(filter(None,i.split(' '))) for j in i.split(' ') if j == 'UNP'] for i in pdb_data_list]
pdb_data_list = [i for i in pdb_data_list if i != []]
pdb_data_list_uniprot = [[j[6] for j in i] for i in pdb_data_list]
#pdb_data_list = [[list(filter(None,j)) for j in i] for i in pdb_data_list]
pdb_data_list = [[j[2] for j in i] for i in pdb_data_list]
pdb_data_list = [i[0] for i in pdb_data_list]
for model in structure:
for pp in ppb.build_peptides(model):
sequence = pp.get_sequence()
for chain, up in zip(model,pdb_data_list_uniprot ):
chain_id = chain.get_id()
# Extract UniProt ID if available in the chain's annotations
uniprot_ids = fetch_uniprot_ids(search)
# Get the resolution from the PDB header
header = structure.header
resolution = header.get('resolution', 'N/A')
if chain_id in pdb_data_list:
# Print UniProt IDs, chain ID, and resolution for the current model
chain_id = chain.get_id()
pdb_fasta.at[index, 'pdbID'] = search
pdb_fasta.at[index, 'chain'] = chain_id
pdb_fasta.at[index, 'pdbSequence'] = str(seqs[chain_id])
pdb_info.at[index, 'uniprotID'] = ', '.join(up)
pdb_info.at[index, 'pdbID'] = search
pdb_info.at[index, 'chain'] = chain_id
pdb_info.at[index, 'resolution'] = resolution
index += 1
print('PDB file processing finished..')
for filename in list(Path(path_to_output_files / 'pdb_structures').glob("*")):
try:
filename_replace_ext = filename.with_suffix(".pdb")
filename.rename(filename_replace_ext)
except:
FileNotFoundError
for filename in list(Path(path_to_output_files / 'pdb_structures').glob("*")):
try:
if filename.stem.startswith("pdb"):
filename_replace_ext = filename.with_name(filename.stem[3:])
filename.rename(filename_replace_ext.with_suffix('.pdb'))
except:
FileNotFoundError
uniprot_matched = pd.merge(uniprot_matched, pdb_info, on='uniprotID', how='left')
uniprot_matched = uniprot_matched.astype(str)
uniprot_matched = uniprot_matched.drop_duplicates()
uniprot_matched = uniprot_matched.merge(pdb_fasta, on=['pdbID', 'chain'], how='left')
uniprot_matched = uniprot_matched.astype(str)
with_pdb = uniprot_matched[(uniprot_matched.pdbID != 'nan') & (
(uniprot_matched.resolution != 'nan') & (uniprot_matched.resolution != 'OT') & (
uniprot_matched.resolution != 'None'))].drop_duplicates()
no_pdb = uniprot_matched[(uniprot_matched.pdbID == 'nan') | (
(uniprot_matched.resolution == 'nan') | (uniprot_matched.resolution == 'OT') | (
uniprot_matched.resolution == 'None'))]
no_pdb = no_pdb[~no_pdb.datapoint.isin(with_pdb.datapoint.to_list())]
no_pdb.drop(columns=['chain', 'pdbID', 'pdbSequence', 'resolution'], inplace=True)
print(
'PDB Information successfully added...\nPDB structures are found for %d of %d.\n%d of %d failed to match with PDB structure.\n'
% (len(with_pdb.drop_duplicates(['datapoint'])), len(uniprot_matched.drop_duplicates(['datapoint'])),
len(no_pdb.drop_duplicates(['datapoint'])), len(uniprot_matched.drop_duplicates(['datapoint']))))
with_pdb = with_pdb.sort_values(['uniprotID', 'resolution'], axis=0, ascending=True)
with_pdb = with_pdb.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos', 'pdbSequence'], keep='first')
with_pdb.replace({'': 'nan'}, inplace=True)
if len(with_pdb) == 0:
with_pdb['pdbInfo'] = ''
else:
for i in with_pdb.index:
try:
res = str(with_pdb.at[i, 'resolution'])
chain = with_pdb.at[i, 'chain']
new = with_pdb.at[i, 'pdbID'] + ':' + chain + ':' + res
with_pdb.at[i, 'pdbInfo'] = new
except:
TypeError
with_pdb.at[i, 'pdbInfo'] = 'nan'
with_pdb = with_pdb[['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence', 'pdbSequence',
'wt_sequence_match',
'whichIsoform', 'pdbID', 'resolution', 'chain', 'pdbInfo', 'datapoint']]
# If the query data points are found in no_match_in_uniprot data frame, it will not give any results.
# If the query data points are found in no_pdb data frame, it will be searched in the modbase and swiss_model steps.
# If the query data points are found in with_pdb data frame, it will be searched in the following steps.
"""
STEP 6
Retrieve sequence annotations.
Add to the data frame.
"""
if len(with_pdb) > 0:
with_pdb = add_annotations(with_pdb)
else:
new_cols = with_pdb.columns.to_list() + ['disulfide', 'intMet', 'intramembrane', 'naturalVariant',
'dnaBinding',
'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane',
'crosslink', 'mutagenesis', 'strand',
'helix', 'turn', 'metalBinding', 'repeat', 'topologicalDomain',
'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif',
'coiledCoil', 'peptide',
'transitPeptide', 'glycosylation', 'propeptide', 'disulfideBinary',
'intMetBinary', 'intramembraneBinary',
'naturalVariantBinary', 'dnaBindingBinary', 'activeSiteBinary',
'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary']
with_pdb = pd.DataFrame(columns=new_cols)
try:
with_pdb.whichIsoform = with_pdb.whichIsoform.astype('str')
except:
AttributeError
with_pdb['whichIsoform'] = ''
with_pdb = with_pdb.astype(str)
with_pdb = with_pdb.replace({'NaN': 'nan'})
with_pdb.replace({'[]': 'nan'}, inplace=True)
with_pdb.replace({'nan-nan': 'nan'}, inplace=True)
with_pdb.replace({'': 'nan'}, inplace=True)
"""
STEP 7
Do alignment for PDB
"""
# Canonical matches, i.e. labelled as m, canonical sequences will be aligned with PDB sequences.
# Isoform matches, i.e. labelled as i, isoform sequences will be aligned with PDB sequences.
with_pdb['uniprotSequence'] = with_pdb['uniprotSequence'].str.replace('U', 'C')
with_pdb['pdbSequence'] = with_pdb['pdbSequence'].str.replace('U', 'C')
dfM = with_pdb[with_pdb.wt_sequence_match == 'm']
dfM = dfM.sort_values(['uniprotID', 'resolution'], axis=0, ascending=True)
dfM = dfM.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos', 'pdbSequence'], keep='first')
dfNM = with_pdb[with_pdb.wt_sequence_match == 'i']
dfNM = dfNM.sort_values(['uniprotID', 'resolution'], axis=0, ascending=True)
dfNM = dfNM.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos', 'pdbSequence'], keep='first')
dfNM.rename(columns={'isoformSequence': 'uniprotSequence'}, inplace=True)
dfM = dfM.astype(str)
dfNM = dfNM.astype(str)
dfM.reset_index(inplace=True)
dfM.drop(['index'], axis=1, inplace=True)
dfNM.reset_index(inplace=True)
dfNM.drop(['index'], axis=1, inplace=True)
uniprot_matched_size = len(uniprot_matched.drop_duplicates(['datapoint']))
uniprot_matched = None
pdb_fasta = None
pdb_info = None
pdbs = None
existing_pdb = None
with_pdb_size = len(with_pdb.drop_duplicates(['datapoint']))
with_pdb = None
print('Aligning sequences...\n')
aligned_m = final_stage(dfM, annotation_list, Path(path_to_output_files / 'alignment_files'))
aligned_nm = final_stage(dfNM, annotation_list, Path(path_to_output_files / 'alignment_files'))
# When PDB sequence is nan, it is wrongly aligned to the UniProt sequence. Fix them.
for i in aligned_m.index:
if aligned_m.at[i, 'pdbSequence'] == 'nan':
aligned_m.at[i, 'mutationPositionOnPDB'] = 'nan'
aligned_m.at[i, 'domainStartonPDB'] = 'nan'
aligned_m.at[i, 'domainEndonPDB'] = 'nan'
aligned_m.at[i, 'pdb_alignStatus'] = 'nan'
for i in aligned_nm.index:
if aligned_nm.at[i, 'pdbSequence'] == 'nan':
aligned_nm.at[i, 'mutationPositionOnPDB'] = 'nan'
aligned_nm.at[i, 'domainStartonPDB'] = 'nan'
aligned_nm.at[i, 'domainEndonPDB'] = 'nan'
aligned_nm.at[i, 'pdb_alignStatus'] = 'nan'
# Check if they the same column name before merging.
aligned_m = aligned_m.astype(str)
aligned_nm = aligned_nm.astype(str)
frames = [aligned_m, aligned_nm]
after_up_pdb_alignment = pd.concat(frames, sort=False)
if len(after_up_pdb_alignment) == 0:
after_up_pdb_alignment['pdb_alignStatus'] = ''
after_up_pdb_alignment['mutationPositionOnPDB'] = ''
after_up_pdb_alignment['domainStartonPDB'] = ''
after_up_pdb_alignment['domainEndonPDB'] = ''
after_up_pdb_alignment = after_up_pdb_alignment.sort_values(
by=['uniprotID', 'wt', 'mut', 'pos', 'pdb_alignStatus', 'resolution', 'chain'],
ascending=[True, True, True, True, True, True, True])
after_up_pdb_alignment = after_up_pdb_alignment.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos'], keep='first')
after_up_pdb_alignment = after_up_pdb_alignment.astype('str')
pdb_aligned = after_up_pdb_alignment[
(after_up_pdb_alignment.pdbID != 'nan') & (after_up_pdb_alignment.mutationPositionOnPDB != 'nan')]
yes_pdb_no_match = after_up_pdb_alignment[
(after_up_pdb_alignment.pdbID != 'nan') & (after_up_pdb_alignment.mutationPositionOnPDB == 'nan')]
no_pdb = no_pdb.copy()
print('PDB matching is completed...\n')
print('SUMMARY')
print('-------')
print('%d data points that failed to match a UniProt Sequence are discarded.' % len(
not_match_in_uniprot.drop_duplicates(['datapoint'])))
print('Of the remaining %d:' % uniprot_matched_size)
print('--%d of %d successfully aligned with PDB structures.' % (
len(pdb_aligned.drop_duplicates(['datapoint'])), with_pdb_size))
print('--%d of %d not found on the covered area by the structure.' % (
len(yes_pdb_no_match.drop_duplicates(['datapoint'])), with_pdb_size))
print('--PDB structures not found for %d datapoints.' % len(no_pdb.drop_duplicates(['datapoint'])))
print('--%d will be searched in Swiss-Model database.\n' % (
len(yes_pdb_no_match.drop_duplicates(['datapoint'])) + len(no_pdb.drop_duplicates(['datapoint']))))
dfM = None
dfNM = None
aligned_nm = None
aligned_m = None
after_up_pdb_alignment = None
print('Proceeding to SwissModel search...')
print('------------------------------------\n')
# At this point we have 4 dataframes
# 1. after_up_pdb_alignment --- This is after PDB sequence alignment. There may be mutations that wasnt found matching to after the alignment. Will be searched in other databases as well.
# 1a. aligned --- we are done with this.
# 1b. yes_pdb_no_match --- They have PDB structures but not matched, so will be searched in the other databases.
# 2. not_match_in_uniprot --- This wont be aligned with anything because these proteins dont have a uniprot ID. Only basic info is present.
# 3. no_pdb --- No PDB structures were found for them. Will be searched in other databases.
"""
Step 8
Neutralize data points that are to be searched in Swiss-Model
# One point is that yes_pdb_no_match's annotations are the adjusted according to the PDBs they are matched before.
# They need to be converted to their old original UniProt annotation positions.
"""
yes_pdb_no_match.drop(['disulfide', 'intMet',
'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'caBinding', 'topologicalDomain', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide', 'disulfideBinary',
'intMetBinary', 'intramembraneBinary',
'naturalVariantBinary', 'dnaBindingBinary', 'activeSiteBinary',
'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary', 'pdbSequence', 'pdbInfo', 'pdbID',
'chain', 'resolution', 'pdb_alignStatus', 'mutationPositionOnPDB',
'domainStartonPDB', 'domainEndonPDB'], axis=1, inplace=True)
to_swiss = pd.concat([yes_pdb_no_match.drop_duplicates(['datapoint']), no_pdb.drop_duplicates(['datapoint'])])
no_pdb = None
to_swiss.reset_index(inplace=True)
to_swiss.drop(['index'], axis=1, inplace=True)
to_swiss = to_swiss.astype('str')
to_swiss = to_swiss.replace({'NaN': 'nan'})
# Create model summary dataframe.
if len(to_swiss) != 0:
print('Generating SwissModel file...\n')
swiss_model = pd.read_csv(Path(path_to_input_files / 'swissmodel_structures.txt'), sep='\t',
dtype=str, header=None, skiprows=1,
names=['UniProtKB_ac', 'iso_id', 'uniprot_seq_length', 'uniprot_seq_md5',
'coordinate_id', 'provider', 'from', 'to', 'template', 'qmean',
'qmean_norm', 'seqid', 'url'])
else:
swiss_model = pd.DataFrame(
columns=['UniProtKB_ac', 'iso_id', 'uniprot_seq_length', 'uniprot_seq_md5', 'coordinate_id',
'provider', 'from', 'to', 'template', 'qmean', 'qmean_norm', 'seqid', 'url', 'whichIsoform'])
swiss_model = swiss_model.astype('str')
try:
swiss_model.iso_id = swiss_model.iso_id.astype('str')
except:
AttributeError
swiss_model['iso_id'] = 'nan'
swiss_model = swiss_model[swiss_model.UniProtKB_ac != 'nan']
for ind in swiss_model.index:
swiss_model.at[ind, 'UniProtKB_ac'] = swiss_model.at[ind, 'UniProtKB_ac'].split('-')[0]
if swiss_model.at[ind, 'iso_id'] != 'nan':
swiss_model.at[ind, 'whichIsoform'] = swiss_model.at[ind, 'iso_id'].split('-')[1]
else:
swiss_model.at[ind, 'whichIsoform'] = 'nan'
# swiss_model.drop(['input'], axis=1, inplace=True)
swiss_model = swiss_model[swiss_model.provider == 'SWISSMODEL']
print('Index File Processed...\n')
# Get relevant columns
swiss_model = swiss_model[
['UniProtKB_ac', 'from', 'to', 'template', 'qmean_norm', 'seqid', 'url', 'whichIsoform']]
# Sort models on qmean score and identity. Some proteins have more than one models, we will pick one.
swiss_model = swiss_model.sort_values(by=['UniProtKB_ac', 'qmean_norm', 'seqid'], ascending=False)
swiss_model.reset_index(inplace=True)
swiss_model.drop(['index'], axis=1, inplace=True)
# Get protein IDs for which there exist models.
swiss_model_ids = set(swiss_model.UniProtKB_ac.to_list())
to_swiss = to_swiss.astype(str)
no_swiss_models = pd.DataFrame()
for i in to_swiss.index:
if to_swiss.at[i, 'uniprotID'] not in swiss_model_ids:
k = pd.Series(to_swiss.iloc[i])
no_swiss_models = no_swiss_models.append(k, ignore_index=True)
no_swiss_models = no_swiss_models.astype(str)
if len(no_swiss_models) == 0:
no_swiss_models = pd.DataFrame(columns=to_swiss.columns)
else:
no_swiss_models = no_swiss_models[to_swiss.columns]
no_swiss_models.reset_index(inplace=True)
no_swiss_models.drop('index', axis=1, inplace=True)
with_swiss_models = pd.concat([to_swiss, no_swiss_models]).drop_duplicates(['datapoint'], keep=False)
with_swiss_models = with_swiss_models[to_swiss.columns]
# Add model info.
with_swiss_models = with_swiss_models.astype(str)
swiss_model = swiss_model.astype(str)
swiss_models_with_data = pd.merge(with_swiss_models, swiss_model, left_on=['uniprotID', 'whichIsoform'],
right_on=['UniProtKB_ac', 'whichIsoform'],
how='left')
swiss_models_with_data = swiss_models_with_data.astype(str)
swiss_models_with_data = swiss_models_with_data.sort_values(by=['uniprotID', 'wt', 'mut', 'pos', 'qmean_norm'],
ascending=False)
swiss_models_with_data = swiss_models_with_data.drop_duplicates()
swiss_models_with_data = swiss_models_with_data.drop(['UniProtKB_ac', 'seqid'], axis=1)
swiss_models_with_data.pos = swiss_models_with_data.pos.astype('int')
swiss_models_with_data = swiss_models_with_data.astype(str)
# Get the ones in the list but without model url and add to the list to go to modbase.
url_nan = swiss_models_with_data[swiss_models_with_data.url == 'nan']
# Add this nan's to no_model. These will be searched in MODBASE because here they dont have urls.
url_nan = url_nan.drop(['from', 'qmean_norm', 'template', 'to', 'url'], axis=1)
no_swiss_models_2 = pd.concat([no_swiss_models, url_nan])
swiss_models_with_data = swiss_models_with_data[swiss_models_with_data.url != 'nan']
for i in swiss_models_with_data.index:
try:
swiss_models_with_data.at[i, 'chain'] = swiss_models_with_data.at[i, 'template'].split('.')[2]
swiss_models_with_data.at[i, 'template'] = swiss_models_with_data.at[i, 'template'].split('.')[0]
except:
IndexError
if len(swiss_models_with_data) == 0:
swiss_models_with_data['chain'] = ''
swiss_models_with_data['template'] = ''
swiss_models_with_data.qmean_norm = swiss_models_with_data.qmean_norm.astype('str')
swiss_models_with_data.chain = swiss_models_with_data.chain.astype('str')
swiss_models_with_data['qmean_norm'] = swiss_models_with_data.qmean_norm.apply(lambda x: round(float(x), 2))
swiss_models_with_data = swiss_models_with_data.astype(str)
# swiss_models_with_data: These data points will be aligned with their corresponding model sequences.
# Add sequences
no_swiss_models_2.reset_index(inplace=True)
no_swiss_models_2.drop('index', axis=1, inplace=True)
swiss_models_with_data.reset_index(inplace=True)
swiss_models_with_data.drop('index', axis=1, inplace=True)
swiss_model_ids = None
with_swiss_models = None
swiss_model = None
no_swiss_models = None
url_nan = None
# At this point we have:
# pdb_aligned --- Align in the PDB phase
# not_match_in_uniprot --- This wont be aligned with anything because these proteins dont have a uniprot ID. Only basic info is present.
# to_swiss (no_pdb + yes_pdb_no_match) --- to be searched in SwissModel database
# to_swiss (with_swiss_models & no_swiss_models)
# swiss_models_with_data --- We found swiss models for them.
# no_swiss_models_2 (no_swiss_models + url_nan)--- to be searched in modbase (the ones having swissmodels but not matching with the boundaries & broken_swiss will be added here)
"""
STEP 9
Associated model IDs are added.
Download model files.
"""
print('Beginning SwissModel files download...')
existing_swiss = list(Path(path_to_output_files / 'swissmodel_structures').glob("*"))
existing_swiss = [str(i) for i in existing_swiss]
existing_swiss = ['.'.join(i.split('/')[-1].split('.')[:-1]) for i in existing_swiss]
swissmodels_fasta = pd.DataFrame()
for i in swiss_models_with_data.index:
protein = swiss_models_with_data.at[i, 'uniprotID']
template = swiss_models_with_data.at[i, 'template'].split('.')[0]
qmean_norm = str(round(float(swiss_models_with_data.at[i, 'qmean_norm']), 2))
if protein + '_' + template + '_' + qmean_norm not in existing_swiss:
url = swiss_models_with_data.at[i, 'url'].strip('\"').strip('}').replace('\\', '').strip('\"').replace(
'https',
'https:')
req = requests.get(url)
name = Path(path_to_output_files / 'swissmodel_structures' / f'{protein}_{template}_{qmean_norm}.txt')
print('Downloading for Protein:', protein + ' Model: ' + template)
with open(name, 'wb') as f:
f.write(req.content)
else:
print('Model exists.')
name = Path(path_to_output_files / 'swissmodel_structures' / f'{protein}_{template}_{qmean_norm}.txt')
with open(name, encoding="utf8") as f:
fasta = ''
lines = f.readlines()
chain = ''
for row in lines:
if row[0:4] == 'ATOM' and row[13:15] == 'CA':
chain = row[20:22].strip()
fasta += threeToOne(row[17:20])
if row[0:3] == 'TER':
k = pd.Series([protein, template, qmean_norm, chain.upper(), fasta])
swissmodels_fasta = swissmodels_fasta.append(k, ignore_index=True)
fasta = ''
if len(swissmodels_fasta) == 0:
swissmodels_fasta = pd.DataFrame(columns=['uniprotID', 'template', 'qmean_norm', 'chain', 'fasta'])
else:
swissmodels_fasta.columns = ['uniprotID', 'template', 'qmean_norm', 'chain', 'fasta']
swissmodels_fasta = swissmodels_fasta.astype(str)
swiss_models_with_data.qmean_norm = swiss_models_with_data.qmean_norm.astype(float)
swissmodels_fasta.qmean_norm = swissmodels_fasta.qmean_norm.astype(float)
swissmodels_fasta = swissmodels_fasta.sort_values(['uniprotID', 'template', 'qmean_norm', 'chain'],
axis=0) # example = 3gdh
swissmodels_fasta.reset_index(inplace=True)
swissmodels_fasta.drop(['index'], axis=1, inplace=True)
swissmodels_fasta = swissmodels_fasta.drop_duplicates(['uniprotID', 'template', 'qmean_norm', 'chain'])
swissmodels_fasta = swissmodels_fasta.drop_duplicates(['uniprotID', 'template', 'chain', 'fasta'])
swissmodels_fasta = swissmodels_fasta.drop_duplicates(['uniprotID', 'template', 'fasta'])
# Some files were broken, thus their PDBs couldnt be recorded.
swissmodels_fasta = swissmodels_fasta.drop_duplicates()
swissmodels_fasta = swissmodels_fasta.astype(str)
swiss_models_with_data = swiss_models_with_data.astype(str)
swissmodels_fasta = swissmodels_fasta.astype(str)
swiss_models_with_data1 = swiss_models_with_data.merge(swissmodels_fasta,
on=['uniprotID', 'template', 'qmean_norm', 'chain'])
swiss_models_with_data1 = swiss_models_with_data1.sort_values(['datapoint', 'fasta'], axis=0,
ascending=[True, False])
swiss_models_with_data1 = swiss_models_with_data1.drop_duplicates(['datapoint', 'template'])
swiss_models_with_data1_dp = list(set(swiss_models_with_data1.datapoint.to_list()))
swiss_models_with_data.reset_index(inplace=True)
swiss_models_with_data.drop(['index'], axis=1, inplace=True)
broken_swiss = pd.DataFrame()
c = 0
for i in swiss_models_with_data.index: # en baştaki dfde var ama model gelende yok.
if swiss_models_with_data.at[i, 'datapoint'] not in swiss_models_with_data1_dp:
k = pd.Series(swiss_models_with_data.iloc[i])
broken_swiss = broken_swiss.append(k, ignore_index=True)
c += 1
if len(broken_swiss) == 0:
broken_swiss = pd.DataFrame(columns=swiss_models_with_data.columns.to_list())
swiss_models_with_data = swiss_models_with_data1.copy()
swiss_models_with_data.qmean_norm = swiss_models_with_data.qmean_norm.astype('float')
swiss_models_with_data = swiss_models_with_data.sort_values(['uniprotID', 'wt', 'mut', 'qmean_norm'],
axis=0, ascending=[True, True, True, False])
# Delete the same model sequence with lower quality
swiss_models_with_data = swiss_models_with_data.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos', 'fasta'],
keep='first')
swiss_models_with_data.uniprotSequence = swiss_models_with_data.uniprotSequence.astype('str')
swiss_models_with_data.pos = swiss_models_with_data.pos.astype('int')
len(swiss_models_with_data.drop_duplicates(['datapoint'])) + len(
broken_swiss.drop_duplicates(['datapoint'])) + len(
no_swiss_models_2.drop_duplicates(['datapoint'])) == len(to_swiss.drop_duplicates(['datapoint']))
# This printed data here includes all possible models with different qualities,
# because we may get a hit in either of them.
swiss_models_with_data.rename({'fasta': 'pdbSequence'}, axis=1, inplace=True) # for convenience.
# NOW DO ALIGNMENT HERE
swiss_models_with_data = swiss_models_with_data.replace({'[\'?\']': 'nan'})
swiss_models_with_data = swiss_models_with_data.replace({'[]': 'nan'})
swiss_models_with_data.rename({'template': 'pdbID'}, axis=1,
inplace=True) # Only to be able use the alignment code above.
swiss_models_with_data = swiss_models_with_data.astype(str)
swiss_models_with_data.pdbSequence = swiss_models_with_data.pdbSequence.astype('str')
swiss_models_with_data = add_annotations(swiss_models_with_data)
swiss_models_with_data = swiss_models_with_data.astype(str)
swiss_models_with_data.replace({'NaN': 'nan'}, inplace=True)
swiss_models_with_data_copy = swiss_models_with_data.copy()
swiss_models_with_data1_dp = None
swiss_models_with_data1 = None
existing_swiss = None
swissmodels_fasta = None
print('Aligning sequences...\n')
swiss_models_with_data['uniprotSequence'] = swiss_models_with_data['uniprotSequence'].str.replace('U', 'C')
swiss_models_with_data['pdbSequence'] = swiss_models_with_data['pdbSequence'].str.replace('U', 'C')
swiss_model_aligned = alignment(swiss_models_with_data, annotation_list,
path_to_output_files / 'alignment_files')
swiss_models_with_data = None
if len(swiss_model_aligned) == 0:
swiss_model_aligned = pd.DataFrame(columns=pdb_aligned.columns)
swiss_model_aligned['qmean_norm'] = 'nan'
else:
swiss_model_aligned = swiss_model_aligned.astype(str)
swiss_model_aligned.replace({'NaN': 'nan'}, inplace=True)
# Some datapoints appear in both nan and not_nan. If not_nan we take it only once.
nan = swiss_model_aligned[swiss_model_aligned.mutationPositionOnPDB == 'nan']
not_nan = swiss_model_aligned[swiss_model_aligned.mutationPositionOnPDB != 'nan']
not_nan.qmean_norm = not_nan.qmean_norm.astype('float')
not_nan.sort_values(['datapoint', 'pdb_alignStatus', 'qmean_norm'], ascending=[True, True, False], inplace=True)
which_ones_are_match = pd.concat([not_nan, nan]).drop_duplicates(['datapoint'], keep='first')
swiss_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB != 'nan']
swiss_not_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB == 'nan']
swiss_match.qmean_norm = swiss_match.qmean_norm.astype('float')
swiss_match.sort_values(['uniprotID', 'wt', 'pos', 'mut', 'pdb_alignStatus', 'qmean_norm'],
ascending=[True, True, True, True, True, False], inplace=True)
swiss_match.drop_duplicates(['uniprotID', 'wt', 'pos', 'mut'], keep='first', inplace=True)
swiss_not_match = swiss_not_match[no_swiss_models_2.columns]
broken_swiss = broken_swiss[no_swiss_models_2.columns]
swiss_not_match = swiss_not_match.drop_duplicates(['datapoint'])
broken_swiss = broken_swiss.drop_duplicates(['datapoint'])
to_modbase = pd.concat([no_swiss_models_2, broken_swiss]).drop_duplicates()
to_modbase = pd.concat([to_modbase, swiss_not_match]).drop_duplicates()
to_modbase = to_modbase.astype(str)
to_swiss_columns = to_swiss.columns
to_swiss_size = len(to_swiss.drop_duplicates(['datapoint']))
to_swiss = None
# CONTROL
"""
# This should be the whole data.
len(swiss_match.drop_duplicates(['datapoint'])) + len(aligned.drop_duplicates(['datapoint'])) + len(to_modbase.drop_duplicates(['datapoint'])) + len(not_match_in_uniprot.drop_duplicates(['datapoint'])) ,len(data)
len(aligned.drop_duplicates(['datapoint'])) + len(not_match_in_uniprot.drop_duplicates(['datapoint'])) +len(to_swiss.drop_duplicates(['datapoint']))== len(data)
"""
print('SwissModel matching is completed...\n')
print('SUMMARY')
print('-------')
print('%d data points that failed to match a UniProt Sequence are discarded.' % len(
not_match_in_uniprot.drop_duplicates(['datapoint'])))
print('Of the remaining %d:' % uniprot_matched_size)
print('--%d of %d successfully aligned with PDB structures.' % (
len(pdb_aligned.drop_duplicates(['datapoint'])), with_pdb_size))
print('--%d of %d successfully aligned with SwissModels structures.' % (
len(swiss_match.drop_duplicates(['datapoint'])), to_swiss_size))
print('--%d will be searched in ModBase database.\n' % len(to_modbase.drop_duplicates(['datapoint'])))
print('Proceeding to ModBase search...')
print('------------------------------------\n')
no_swiss_models_2 = None
broken_swiss = None
swiss_model_aligned = None
nan = None
not_nan = None
which_ones_are_match = None
swiss_not_match = None
# STEP : GO TO MODBASE
# Should not include anything related to prev models.
if len(to_modbase) != 0:
to_modbase = to_modbase.astype(str)
# GET MODBASE MODELS
# Get IDs from data to retrieve only their models from MODBASE
to_modbase.reset_index(inplace=True)
to_modbase.drop(['index'], axis=1, inplace=True)
existing_modbase_models = list(Path(path_to_output_files / 'modbase_structures').glob("*"))
existing_modbase_models = [str(i) for i in existing_modbase_models]
existing_modbase_models = [i.split('/')[-1].split('.')[0] for i in existing_modbase_models]
existing_modbase_models_ind = list(Path(path_to_output_files / 'modbase_structures_individual').glob("*"))
existing_modbase_models_ind = [str(i) for i in existing_modbase_models_ind]
existing_modbase_models_ind = [i.split('/')[-1].split('.')[0] for i in existing_modbase_models_ind]
modbase_reduced = pd.DataFrame()
modbase_fasta = pd.DataFrame()
print('Retrieving ModBase models...\n')
# Get model files associated with each UniProtID
for protein in list(set(to_modbase.uniprotID.to_list())):
if protein not in existing_modbase_models:
print('Downloading Modbase models for ', protein)
url = 'https://salilab.org/modbase/retrieve/modbase/?databaseID=' + protein
req = requests.get(url)
name = path_to_output_files / 'modbase_structures' / f'{protein}.txt'
with open(name, 'wb') as f:
f.write(req.content)
else:
print('Model exists for', protein)
name = Path(path_to_output_files / 'modbase_structures' / f'{protein}.txt')
with open(name, encoding="utf8") as f:
a = open(name, 'r').read()
soup = BeautifulSoup(a, 'lxml')
for pdb in soup.findAll('pdbfile'):
model_id = str(pdb.contents[1])[10:-11]
if model_id not in existing_modbase_models_ind:
with open(path_to_output_files / 'modbase_structures_individual' / f'{model_id}.txt', 'w',
encoding="utf8") as individual:
individual.write(str('UniProt ID: ' + protein))
individual.write('\n')
individual.write(str(pdb.contents[3])[10:-11].strip())
with open(path_to_output_files / 'modbase_structures_individual' / f'{model_id}.txt',
encoding="utf8") as f:
fasta = ''
chain = ''
template_chain = ''
score = -999
for ind_line in f.readlines():
if ind_line[0:10] == 'UniProt ID':
uniprot_id = ind_line.split(':')[1].strip()
if ind_line[0:23] == 'REMARK 220 TARGET BEGIN':
target_begin = ind_line[40:43].strip()
if ind_line[0:21] == 'REMARK 220 TARGET END':
target_end = ind_line[40:43].strip()
if ind_line[0:25] == 'REMARK 220 TEMPLATE BEGIN':
pdb_begin = ind_line[40:43].strip()
if ind_line[0:23] == 'REMARK 220 TEMPLATE END':
pdb_end = ind_line[40:43].strip()
if ind_line[0:23] == 'REMARK 220 TEMPLATE PDB':
pdb_code = ind_line[40:43].strip()
if ind_line[0:25] == 'REMARK 220 TEMPLATE CHAIN':
pdb_chain = ind_line[40:43].strip()
if ind_line[0:32] == 'REMARK 220 ModPipe Quality Score':
quality_score = ind_line[40:].strip()
if ind_line[0:27] == 'REMARK 220 MODPIPE MODEL ID':
model_id = ind_line[40:].strip()
if ind_line[0:25] == 'REMARK 220 TEMPLATE CHAIN':
template_chain = ind_line[40:42].strip()
if ind_line[0:4] == 'ATOM' and ind_line[13:15] == 'CA':
fasta += threeToOne(ind_line[17:20])
if ind_line[0:32] == 'REMARK 220 ModPipe Quality Score':
try:
score = ind_line[40:].strip()
except (ValueError):
score = -999
if ind_line[0:3] == 'TER' or ind_line[0:3] == 'END':
k = pd.Series([uniprot_id, model_id, str(score), template_chain, fasta])
modbase_fasta = modbase_fasta.append(k, ignore_index=True)
fasta = ''
try:
k = pd.Series(
[uniprot_id, target_begin, target_end, pdb_code, pdb_chain, pdb_begin, pdb_end,
quality_score,
model_id])
modbase_reduced = modbase_reduced.append(k, ignore_index=True)
except:
NameError
print('This file doesnt have Quality Score. Replacer: -999', model_id)
quality_score = -999
print()
if len(modbase_fasta) != 0:
modbase_fasta.columns = ['uniprotID', 'template', 'score', 'chain', 'fasta']
else:
modbase_fasta = pd.DataFrame(columns=['uniprotID', 'template', 'score', 'chain', 'fasta'])
modbase_fasta = modbase_fasta.astype(str)
modbase_fasta = modbase_fasta.replace({'': 'nan'})
modbase_fasta = modbase_fasta.replace({'NaN': 'nan'})
modbase_fasta = modbase_fasta[modbase_fasta.fasta != 'nan']
print('Modbase model frame constructed.\n')
if len(modbase_reduced) != 0:
modbase_reduced.columns = ['UniprotID', 'TargetBeg', 'TargetEnd', 'PDBCode', 'PDBChain', 'PDBBegin',
'PDBEnd',
'ModPipeQualityScore', 'ModelID']
else:
modbase_reduced = pd.DataFrame(
columns=['UniprotID', 'TargetBeg', 'TargetEnd', 'PDBCode', 'PDBChain', 'PDBBegin', 'PDBEnd',
'ModPipeQualityScore', 'ModelID'])
to_modbase = add_annotations(to_modbase)
to_modbase = to_modbase.astype(str)
to_modbase.fillna('nan', inplace=True)
to_modbase = to_modbase.replace({'NaN': 'nan'})
to_modbase.replace({'[]': 'nan'}, inplace=True)
to_modbase.replace({'nan-nan': 'nan'}, inplace=True)
to_modbase.replace({'': 'nan'}, inplace=True)
model_info_added = to_modbase.merge(modbase_reduced, right_on='UniprotID', left_on='uniprotID',
how='left')
modbase_reduced = None
existing_modbase_models = None
existing_modbase_models_ind = None
model_info_added = model_info_added.drop(['UniprotID'], axis=1)
model_info_added = model_info_added.rename(columns={'TargetBeg': 'from', 'TargetEnd': 'to',
'PDBCode': 'template', 'PDBChain': 'chain',
'ModPipeQualityScore': 'score',
'ModelID': 'pdbID'})
model_info_added.drop(['PDBEnd', 'PDBBegin'], axis=1, inplace=True)
model_info_added.score = model_info_added.score.astype(float)
model_info_added = model_info_added.sort_values(by=['datapoint', 'score'],
ascending=False)
model_info_added.reset_index(inplace=True)
model_info_added.drop(['index'], axis=1, inplace=True)
model_info_added = model_info_added.drop_duplicates()
model_info_added = model_info_added.astype(str)
model_info_added = model_info_added.replace({'NaN': 'nan'})
no_info = model_info_added[model_info_added.pdbID == 'nan']
with_modbase_info = model_info_added[model_info_added.pdbID != 'nan']
model_info_added = None
len(no_info.drop_duplicates(['datapoint'])), len(with_modbase_info.drop_duplicates(['datapoint']))
len(no_info.drop_duplicates(['datapoint'])) + len(with_modbase_info.drop_duplicates(['datapoint'])) == len(
to_modbase.drop_duplicates(['datapoint']))
# Add no_info to the rest down below!
no_info = no_info[to_swiss_columns]
with_modbase_info.score = with_modbase_info.score.astype(float)
modbase_fasta.score = modbase_fasta.score.astype(float)
modbase_fasta = modbase_fasta.sort_values(['uniprotID', 'score', 'template', 'chain'],
ascending=[True, False, True, True], axis=0) # example = 3gdh
# I added this newly downloaded ones to the main model file.
modbase_fasta = modbase_fasta.rename(columns={'template': 'pdbID'})
with_modbase_info.pos = with_modbase_info.pos.astype('int')
with_modbase_info.score = with_modbase_info.score.astype(float)
with_modbase_info.score = with_modbase_info.score.apply(lambda x: round(x, 2))
modbase_fasta.score = modbase_fasta.score.astype(float)
modbase_fasta.score = modbase_fasta.score.apply(lambda x: round(x, 2))
with_modbase_info = with_modbase_info.merge(modbase_fasta, on='pdbID', how='left')
with_modbase_info.drop(['score_y'], axis=1, inplace=True)
with_modbase_info.rename(columns={'score_x': 'score'}, inplace=True)
with_modbase_info.drop(['uniprotID_y', 'chain_y'], axis=1, inplace=True)
with_modbase_info.rename(columns={'uniprotID_x': 'uniprotID', 'chain_x': 'chain'}, inplace=True)
with_modbase_info.score = with_modbase_info.score.astype('float')
with_modbase_info = with_modbase_info.sort_values(['uniprotID', 'wt', 'mut', 'pos', 'score', 'from', 'to'],
axis=0,
ascending=[True, True, True, True, False, True, False])
with_modbase_info = with_modbase_info.drop_duplicates(['uniprotID', 'wt', 'mut', 'pos', 'fasta'],
keep='first')
with_modbase_info = with_modbase_info.replace({'[\'?\']': 'nan'})
with_modbase_info = with_modbase_info.replace({'[]': 'nan'})
with_modbase_info = with_modbase_info.replace({'\'?\', ': ''})
with_modbase_info = with_modbase_info.replace({', \'?\'': ''})
with_modbase_info = with_modbase_info.replace({'(': ''})
with_modbase_info = with_modbase_info.replace(
{')': ''})
with_modbase_info = with_modbase_info.astype(str)
with_modbase_info.fasta = with_modbase_info.fasta.astype('str')
with_modbase_info.reset_index(inplace=True)
with_modbase_info.drop('index', axis=1, inplace=True)
align = with_modbase_info[
with_modbase_info.fasta != 'nan']
yes_pdb_no_match = with_modbase_info[
with_modbase_info.fasta == 'nan']
yes_pdb_no_match = yes_pdb_no_match[~yes_pdb_no_match.datapoint.isin(align.datapoint.to_list())]
align.rename(columns={'fasta': 'pdbSequence'}, inplace=True)
align['uniprotSequence'] = align['uniprotSequence'].str.replace('U', 'C')
align['pdbSequence'] = align['pdbSequence'].str.replace('U', 'C')
to_modbase_size = len(to_modbase.drop_duplicates(['datapoint']))
modbase_fasta = None
to_modbase = None
print('Aligning sequences...\n')
modbase_aligned = alignment(align, annotation_list, path_to_output_files / 'alignment_files')
modbase_aligned = modbase_aligned.astype(str)
modbase_aligned = modbase_aligned.replace({'NaN': 'nan'})
# Get the ones whose models couldn't be found. Add to no_modbase (yani hiçbir şey de eşleşmemiş artık.)
if len(with_modbase_info) != 0:
not_in_aligned = pd.concat([modbase_aligned.drop_duplicates(['datapoint']),
with_modbase_info.drop_duplicates(['datapoint'])]).drop_duplicates(
['datapoint'],
keep=False)
else:
not_in_aligned = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint', 'disulfide',
'intMet',
'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane',
'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif',
'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide',
'disulfide',
'intMet', 'intramembrane', 'naturalVariant', 'dnaBinding',
'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane',
'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif',
'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide', 'from',
'to', 'template', 'chain', 'score', 'pdbID', 'pdbSequence', 'fasta'])
with_modbase_info = None
if len(not_in_aligned) != 0:
not_models = pd.concat([yes_pdb_no_match.drop_duplicates(['datapoint']),
not_in_aligned.drop_duplicates(['datapoint'])]).drop_duplicates(['datapoint'],
keep='first')
# Retain the best model among the aligned ones.
else:
not_models = pd.DataFrame(columns=not_in_aligned.columns)
yes_pdb_no_match = None
# # Some datapoints appear in both nan and not_nan. If not_nan we take it only once.
modbase_aligned = modbase_aligned.astype(str)
if len(modbase_aligned) != 0:
nan = modbase_aligned[modbase_aligned.mutationPositionOnPDB == 'nan']
not_nan = modbase_aligned[modbase_aligned.mutationPositionOnPDB != 'nan']
not_nan.score = not_nan.score.astype(float)
not_nan.sort_values(['datapoint', 'pdb_alignStatus', 'score'], ascending=[True, True, False],
inplace=True)
not_nan = not_nan.sort_values(['datapoint', 'mutationPositionOnPDB', 'score'],
ascending=[True, True, False])
not_nan = not_nan.drop_duplicates(['datapoint'], keep='first')
else:
nan = pd.DataFrame(columns=modbase_aligned.columns)
not_nan = pd.DataFrame(columns=modbase_aligned.columns)
modbase_aligned = None
which_ones_are_match = pd.concat([not_nan, nan]).drop_duplicates(['datapoint'], keep='first')
if len(which_ones_are_match) == 0:
which_ones_are_match = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint', 'disulfide', 'intMet',
'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide',
'disulfideBinary', 'intMetBinary', 'intramembraneBinary',
'naturalVariantBinary', 'dnaBindingBinary', 'activeSiteBinary',
'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary', 'from', 'to', 'template',
'chain', 'score', 'pdbID', 'pdbSequence', 'pdb_alignStatus',
'mutationPositionOnPDB', 'domainStartonPDB', 'domainEndonPDB'])
modbase_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB != 'nan']
modbase_not_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB == 'nan']
else:
modbase_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB != 'nan']
modbase_not_match = which_ones_are_match[which_ones_are_match.mutationPositionOnPDB == 'nan']
which_ones_are_match = None
modbase_match.score = modbase_match.score.astype('float')
modbase_match = modbase_match.sort_values(['datapoint', 'mutationPositionOnPDB', 'score'],
ascending=[True, True, False])
modbase_match.drop_duplicates(['datapoint'], keep='first', inplace=True)
not_nan = None
nan = None
# merge not_in_align and modbase_not_match as they were both excluded from modbase match.
# No model
no_info = no_info[to_swiss_columns]
no_info = no_info.drop_duplicates()
# Model present, no sequence
not_models = not_models[to_swiss_columns]
not_models = not_models.drop_duplicates()
# Modbase model and sequence present, no match in PDB
modbase_not_match = modbase_not_match[to_swiss_columns]
modbase_not_match = modbase_not_match.drop_duplicates()
if len(not_in_aligned) != 0 and len(modbase_not_match) != 0 and len(no_info) != 0:
rest = pd.concat([not_in_aligned, modbase_not_match, no_info])
elif len(not_in_aligned) != 0 and len(modbase_not_match) != 0 and len(no_info) == 0:
rest = pd.concat([not_in_aligned, modbase_not_match])
elif len(not_in_aligned) == 0 and len(modbase_not_match) != 0 and len(no_info) != 0:
rest = pd.concat([modbase_not_match, no_info])
elif len(not_in_aligned) != 0 and len(modbase_not_match) == 0 and len(no_info) != 0:
rest = pd.concat([not_in_aligned, no_info])
elif len(not_in_aligned) != 0 and len(modbase_not_match) == 0 and len(no_info) == 0:
rest = not_in_aligned
elif len(not_in_aligned) == 0 and len(modbase_not_match) != 0 and len(no_info) == 0:
rest = modbase_not_match
elif len(not_in_aligned) == 0 and len(modbase_not_match) == 0 and len(no_info) != 0:
rest = no_info
else:
rest = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint'])
rest = rest[to_swiss_columns]
rest = rest.drop_duplicates()
rest.reset_index(inplace=True)
rest.drop(['index'], axis=1, inplace=True)
rest = rest.astype('str')
else:
modbase_match = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint', 'disulfide', 'intMet',
'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide',
'disulfideBinary', 'intMetBinary', 'intramembraneBinary',
'naturalVariantBinary', 'dnaBindingBinary', 'activeSiteBinary',
'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary', 'from', 'to', 'template',
'chain', 'score', 'pdbID', 'pdbSequence', 'pdb_alignStatus',
'mutationPositionOnPDB', 'domainStartonPDB', 'domainEndonPDB'])
not_in_aligned = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint', 'disulfide', 'intMet',
'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide', 'disulfide',
'intMet', 'intramembrane', 'naturalVariant', 'dnaBinding', 'activeSite',
'nucleotideBinding', 'lipidation', 'site', 'transmembrane', 'crosslink',
'mutagenesis', 'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide', 'from',
'to', 'template', 'chain', 'score', 'pdbID', 'pdbSequence', 'fasta'])
no_info = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint'])
rest = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'wt_sequence_match', 'whichIsoform', 'datapoint'])
rest = rest[to_swiss_columns]
rest = rest.drop_duplicates()
rest.reset_index(inplace=True)
rest.drop(['index'], axis=1, inplace=True)
rest = rest.astype('str')
to_modbase_size = 0
print('Modbase matching is completed...\n')
print('SUMMARY')
print('-------')
print('%d data points that failed to match a UniProt Sequence are discarded.' % len(
not_match_in_uniprot.drop_duplicates(['datapoint'])))
print('Of the remaining %d:' % uniprot_matched_size)
print('--%d of %d successfully aligned with PDB structures.' % (
len(pdb_aligned.drop_duplicates(['datapoint'])), with_pdb_size))
print('--%d of %d successfully aligned with SwissModels structures.' % (
len(swiss_match.drop_duplicates(['datapoint'])), to_swiss_size))
print('--%d of %d successfully aligned with Modbase structures.\n' % (
len(modbase_match.drop_duplicates(['datapoint'])), to_modbase_size))
print('--Remaining %d not found to match any models.' % len(rest.drop_duplicates(['datapoint'])))
print('--A total of %d datapoints will not be evaluated.\n' % (
len(rest.drop_duplicates(['datapoint'])) + len(not_match_in_uniprot.drop_duplicates(['datapoint']))))
print('FOR CHECKING : ',
len(rest.drop_duplicates(['datapoint'])) + len(not_match_in_uniprot.drop_duplicates(['datapoint'])) + len(
pdb_aligned.drop_duplicates(['datapoint'])) + len(swiss_match.drop_duplicates(['datapoint'])) + len(
modbase_match.drop_duplicates(['datapoint'])) == data_size)
no_info = None
align = None
not_in_aligned = None
not_models = None
modbase_not_match = None
# Final corrections
# Now 3D alignment.
pdb = pdb_aligned.copy()
swiss = swiss_match.copy()
modbase = modbase_match.copy()
pdb_aligned = None
swiss_match = None
modbase_match = None
"""
WHAT DO WE HAVE NOW?
- uniprot sequence not found
- pdb aligned
- swiss aligned
- modbase aligned
- not aligned with anything (rest)
"""
# Fix the axes and merge all data.
pdb.drop(['pdbInfo'], axis=1, inplace=True)
pdb.rename(columns={'resolution': 'score'}, inplace=True)
swiss.rename(columns={'qmean_norm': 'score'}, inplace=True)
modbase.rename(columns={'qmean_norm': 'score'}, inplace=True)
swiss = swiss[pdb.columns]
modbase = modbase[pdb.columns]
pdb['source'] = 'PDB'
swiss['source'] = 'SWISSMODEL'
modbase['source'] = 'MODBASE'
data = pd.concat([swiss, modbase, pdb])
data.reset_index(inplace=True)
data.drop(['index'], axis=1, inplace=True)
data = data.astype('str')
data_spare = pd.concat([not_match_in_uniprot, rest])
not_match_in_uniprot = None
pdb = None
swiss = None
modbase = None
rest = None
print('Generating FreeSASA files...')
print('------------------------------------\n')
# Folder to calculated RSA values.
existing_free_sasa = list(Path(path_to_output_files / 'freesasa_files').glob("*"))
existing_free_sasa = [str(i) for i in existing_free_sasa]
existing_free_sasa = [i.split('/')[-1].split('.')[0] for i in existing_free_sasa]
print('Calculation RSA for PDB Structure Files...\n')
pdb_only = data[data.source == 'PDB']
for pdbID in pdb_only.pdbID.to_list():
if pdbID not in existing_free_sasa:
(run_freesasa(Path(path_to_output_files / 'pdb_structures' / f'{pdbID.lower()}.pdb'),
Path(path_to_output_files / 'freesasa_files' / f'{pdbID.lower()}.txt'),
include_hetatms=True,
outdir=None, force_rerun=False, file_type='pdb'))
print('Calculation RSA for SwissModel Files...\n')
swiss_only = data[data.source == 'SWISSMODEL']
swiss_dp = []
for i in swiss_only.index:
swiss_dp.append(swiss_only.at[i, 'uniprotID'] + '_' + swiss_only.at[i, 'pdbID'].lower() + '_' + str(
round(float(swiss_only.at[i, 'score']), 2)))
for pdbID in swiss_dp:
if pdbID not in existing_free_sasa:
(run_freesasa(Path(path_to_output_files / 'swissmodel_structures' / f'{pdbID}.txt'),
Path(path_to_output_files / 'freesasa_files' / f'{pdbID}.txt'), include_hetatms=True,
outdir=None, force_rerun=False, file_type='pdb'))
print('Calculation RSA for Modbase Model Files...\n')
modbase_only = data[data.source == 'MODBASE']
for pdbID in modbase_only.pdbID.to_list():
if pdbID not in existing_free_sasa:
(run_freesasa(Path(path_to_output_files / 'modbase_structures_individual' / f'{pdbID.lower()}.txt'),
Path(path_to_output_files / 'freesasa_files' / f'{pdbID.lower()}.txt'),
include_hetatms=True,
outdir=None, force_rerun=False, file_type='pdb'))
# This annotation list is different than the prev one, keep it.
annotation_list += ['domainStartonPDB', 'domainEndonPDB']
folder_path = path_to_output_files / 'freesasa_files'
aligner = Align.PairwiseAligner()
print('Proceeding to 3D distance calculation...\n')
data.domainEndonPDB = data.domainEndonPDB.astype(str)
data.domainStartonPDB = data.domainStartonPDB.astype(str)
existing_free_sasa = None
swiss_dp = None
pdb_only = None
swiss_only = None
modbase_only = None
data['uniprotSequence'] = data['uniprotSequence'].str.replace('U', 'C')
data['pdbSequence'] = data['pdbSequence'].str.replace('U', 'C')
for i in data.index:
id_ = data.at[i, 'pdbID'].lower()
up_id_ = data.at[i, 'uniprotID']
score_ = str(data.at[i, 'score'])
if data.at[i, 'source'] == 'PDB':
pdb_path = Path(path_to_output_files / 'pdb_structures' / f'{id_}.pdb')
elif data.at[i, 'source'] == 'MODBASE':
pdb_path = Path(path_to_output_files / 'modbase_structures_individual' / f'{id_}.txt')
elif data.at[i, 'source'] == 'SWISSMODEL':
pdb_path = Path(path_to_output_files / 'swissmodel_structures' / f'{up_id_}_{id_}_{score_}.txt')
pdbSequence = data.at[i, 'pdbSequence']
source = data.at[i, 'source']
chain = data.at[i, 'chain']
uniprotID = data.at[i, 'uniprotID']
pdbID = data.at[i, 'pdbID']
alignments = get_alignments_3D(uniprotID, 'nan', pdb_path, pdbSequence, source, chain, pdbID, mode, Path(path_to_output_files / '3D_alignment'), file_format = 'gzip')
mutPos = data.at[i, 'mutationPositionOnPDB']
try:
coordMut = get_coords(mutPos, alignments, 'nan', 'nan', mode)[0]
except:
ValueError
coordMut = 'nan'
try:
sasa_pos = get_coords(mutPos, alignments, 'nan', 'nan', mode)[2]
data.at[i, 'sasa'] = sasa(data.at[i, 'source'], data.at[i, 'pdbID'], data.at[i, 'uniprotID'], sasa_pos,
data.at[i, 'wt'], mode, path_to_output_files, file_type='pdb')
except:
ValueError
data.at[i, 'sasa'] = 'nan' # mutation position is nan
for annot in annotation_list:
annotx = []
try:
positions_of_annotations = data.at[i, annot].split(',')
for pos in positions_of_annotations:
pos = pos.strip().strip('\'').strip('[\'').strip('\']')
try:
if '-' not in pos:
pos = int(float(pos))
coordAnnot = get_coords(pos, alignments, 'nan', 'nan', mode)[0]
try:
annotx.append(find_distance(coordMut, coordAnnot))
except:
ValueError
else:
for r in range(int(pos.split('-')[0]), int(pos.split('-')[1]) + 1):
coordAnnot = get_coords(r, alignments, 'nan', 'nan', mode)[0]
annotx.append(find_distance(coordMut, coordAnnot))
except:
ValueError
try:
data.at[i, annot] = min([float(i) for i in annotx])
except:
ValueError
data.at[i, annot] = 'nan'
except:
ValueError
if (str(data.at[i, 'domainStartonPDB']) == 'NaN' or str(data.at[i, 'domainStartonPDB']) == 'nan') and (
str(data.at[i, 'domainEndonPDB']) != 'NaN' and str(data.at[i, 'domainEndonPDB']) != 'nan'):
data.at[i, 'domainStartonPDB'] = 100000
elif (str(data.at[i, 'domainEndonPDB']) == 'NaN' or str(data.at[i, 'domainEndonPDB']) == 'nan') and (
str(data.at[i, 'domainStartonPDB']) != 'NaN' and str(data.at[i, 'domainStartonPDB']) != 'nan'):
data.at[i, 'domainEndonPDB'] = 100000
elif (str(data.at[i, 'domainStartonPDB']) == 'NaN' and str(data.at[i, 'domainEndonPDB']) == 'nan'):
data.at[i, 'domaindistance3D'] = 'nan'
data.at[i, 'domaindistance3D'] = min(float(data.at[i, 'domainStartonPDB']),
float(data.at[i, 'domainEndonPDB']))
data.at[i, 'domaindistance3D'] = min(float(data.at[i, 'domainStartonPDB']),
float(data.at[i, 'domainEndonPDB']))
data = data.astype(str)
data.replace({'NaN': 'nan'}, inplace=True)
# Now unify all 3 separate data. We have with_pdb. The ones that have pdb structyres, swiss, modbase, the ones didnt match with ant and the ones didnt have wt seq match.
# Get interface positions from ECLAIR. Download HQ human
print()
print('Assigning surface regions...')
print('------------------------------------\n')
print('Extracting interface residues...\n')
data_interface = pd.read_csv(path_to_interfaces, sep='\t')
positions = get_interface_positions(data_interface, 'P1', 'P2')
interface_dataframe = pd.DataFrame()
for key, val in positions.items():
k = pd.Series((key, str(list(set(val)))))
interface_dataframe = interface_dataframe.append(k, ignore_index=True)
interface_dataframe.columns = ['uniprotID', 'positions']
if len(data) == 0:
data = pd.DataFrame(
columns=['uniprotID', 'wt', 'mut', 'pos', 'composition', 'polarity', 'volume', 'granthamScore',
'domain', 'domStart', 'domEnd', 'distance', 'uniprotSequence',
'pdbSequence', 'wt_sequence_match', 'whichIsoform', 'pdbID', 'score',
'chain', 'datapoint', 'disulfide', 'intMet', 'intramembrane',
'naturalVariant', 'dnaBinding', 'activeSite', 'nucleotideBinding',
'lipidation', 'site', 'transmembrane', 'crosslink', 'mutagenesis',
'strand', 'helix', 'turn', 'metalBinding', 'repeat',
'topologicalDomain', 'caBinding', 'bindingSite', 'region',
'signalPeptide', 'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil',
'peptide', 'transitPeptide', 'glycosylation', 'propeptide',
'disulfideBinary', 'intMetBinary', 'intramembraneBinary',
'naturalVariantBinary', 'dnaBindingBinary', 'activeSiteBinary',
'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'topologicalDomainBinary', 'caBindingBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary', 'pdb_alignStatus',
'mutationPositionOnPDB', 'domainStartonPDB', 'domainEndonPDB',
'source', 'sasa', 'domaindistance3D', 'threeState_trsh4_HQ', 'domain_fisher'])
else:
data.sasa = data.sasa.astype('str')
for i in data.index:
if '*' in data.at[i, 'sasa']:
data.at[i, 'sasa'] = data.at[i, 'sasa'].split('*')[0]
data.sasa = data.sasa.replace({'N/A': 'nan'})
data.sasa = data.sasa.replace({'None': 'nan'})
data.replace({' N/A': 'nan'}, inplace=True)
data.replace({'None': 'nan'}, inplace=True)
data.sasa = data.sasa.astype(float)
data = data.astype(str)
for i in data.index:
if float(data.at[i, 'sasa']) < 5:
data.at[i, 'trsh4'] = 'core'
elif float(data.at[i, 'sasa']) >= 5:
data.at[i, 'trsh4'] = 'surface'
elif data.at[i, 'sasa'] == 'nan':
data.at[i, 'trsh4'] = 'nan'
data = data.merge(interface_dataframe, on='uniprotID', how='left')
data.positions = data.positions.astype('str')
for i in data.index:
if (str(data.at[i, 'pos']) in data.at[i, 'positions']) and data.at[i, 'trsh4'] == 'surface':
data.at[i, 'threeState_trsh4_HQ'] = 'interface'
elif (str(data.at[i, 'pos']) not in data.at[i, 'positions']) and data.at[i, 'trsh4'] == 'surface':
data.at[i, 'threeState_trsh4_HQ'] = 'surface'
elif (str(data.at[i, 'pos']) not in data.at[i, 'positions']) and data.at[i, 'trsh4'] == 'core':
data.at[i, 'threeState_trsh4_HQ'] = 'core'
elif (str(data.at[i, 'pos']) in data.at[i, 'positions']) and data.at[i, 'trsh4'] == 'core':
data.at[i, 'threeState_trsh4_HQ'] = 'conflict'
elif data.at[i, 'trsh4'] == 'nan':
data.at[i, 'threeState_trsh4_HQ'] = 'nan'
data.drop(['positions'], axis=1, inplace=True)
# OPTIONAL
# DOMAIN SELECTION
# Next step: Delete all other domains with 'NULL.' R is capable of handling 53 categories. We will keep 52 most
# significant domains and 53th category will be NULL.
fisherResult = pd.read_csv(fisher_path, sep='\t')
significant_domains = fisherResult.domain.to_list()
for i in data.index:
if data.at[i, 'domain'] in significant_domains:
data.at[i, 'domain_fisher'] = data.at[i, 'domain']
else:
data.at[i, 'domain_fisher'] = 'NULL'
# Change the numbering for binary annotations and create 3 classes:
# nan--> 0, 0 -->1 and 1 -->2
print('Final adjustments are being done...\n')
binaryCols = ['disulfideBinary', 'intMetBinary', 'intramembraneBinary', 'naturalVariantBinary',
'dnaBindingBinary',
'activeSiteBinary', 'nucleotideBindingBinary', 'lipidationBinary', 'siteBinary',
'transmembraneBinary', 'crosslinkBinary', 'mutagenesisBinary',
'strandBinary', 'helixBinary', 'turnBinary', 'metalBindingBinary',
'repeatBinary', 'caBindingBinary', 'topologicalDomainBinary',
'bindingSiteBinary', 'regionBinary', 'signalPeptideBinary',
'modifiedResidueBinary', 'zincFingerBinary', 'motifBinary',
'coiledCoilBinary', 'peptideBinary', 'transitPeptideBinary',
'glycosylationBinary', 'propeptideBinary']
data = data.astype(str)
data.replace({'NaN': 'nan'}, inplace=True)
for i in data.index:
for j in binaryCols:
data[j] = data[j].astype('str')
if (data.at[i, j] == '0') or (data.at[i, j] == '0.0'):
data.at[i, j] = '1'
elif data.at[i, j] == 'nan':
data.at[i, j] = '0'
elif (data.at[i, j] == '1') or (data.at[i, j] == '1.0'):
data.at[i, j] = '2'
annotCols = ['disulfide', 'intMet', 'intramembrane',
'naturalVariant', 'dnaBinding', 'activeSite', 'nucleotideBinding',
'lipidation', 'site', 'transmembrane', 'crosslink', 'mutagenesis',
'strand', 'helix', 'turn', 'metalBinding', 'repeat', 'caBinding',
'topologicalDomain', 'bindingSite', 'region', 'signalPeptide',
'modifiedResidue', 'zincFinger', 'motif', 'coiledCoil', 'peptide',
'transitPeptide', 'glycosylation', 'propeptide']
for i in data.index:
for annot in annotCols:
binaryName = str(annot) + 'Binary'
if data.at[i, binaryName] == '2':
data.at[i, annot] = '0.0'
data.replace({'100000': 'nan'}, inplace=True)
data = add_physicochemical(data)
data.rename(
columns={'uniprotID': 'prot_uniprotAcc', 'wt': 'wt_residue', 'pos': 'position', 'mut': 'mut_residue',
'datapoint': 'meta_merged', 'datapoint_disease': 'meta-lab_merged', 'label': 'source_db',
'family': 'prot_family', 'domain': 'domains_all', 'domain_fisher': 'domains_sig',
'domaindistance3D': 'domains_3Ddist', 'threeState_trsh4_HQ': 'location_3state',
'disulfideBinary': 'disulfide_bin', 'intMetBinary': 'intMet_bin',
'intramembraneBinary': 'intramembrane_bin',
'naturalVariantBinary': 'naturalVariant_bin', 'dnaBindingBinary': 'dnaBinding_bin',
'activeSiteBinary': 'activeSite_bin',
'nucleotideBindingBinary': 'nucleotideBinding_bin', 'lipidationBinary': 'lipidation_bin',
'siteBinary': 'site_bin',
'transmembraneBinary': 'transmembrane_bin', 'crosslinkBinary': 'crosslink_bin',
'mutagenesisBinary': 'mutagenesis_bin',
'strandBinary': 'strand_bin', 'helixBinary': 'helix_bin', 'turnBinary': 'turn_bin',
'metalBindingBinary': 'metalBinding_bin',
'repeatBinary': 'repeat_bin', 'topologicalDomainBinary': 'topologicalDomain_bin',
'caBindingBinary': 'caBinding_bin',
'bindingSiteBinary': 'bindingSite_bin', 'regionBinary': 'region_bin',
'signalPeptideBinary': 'signalPeptide_bin',
'modifiedResidueBinary': 'modifiedResidue_bin', 'zincFingerBinary': 'zincFinger_bin',
'motifBinary': 'motif_bin',
'coiledCoilBinary': 'coiledCoil_bin', 'peptideBinary': 'peptide_bin',
'transitPeptideBinary': 'transitPeptide_bin',
'glycosylationBinary': 'glycosylation_bin', 'propeptideBinary': 'propeptide_bin',
'disulfide': 'disulfide_dist', 'intMet': 'intMet_dist',
'intramembrane': 'intramembrane_dist', 'naturalVariant': 'naturalVariant_dist',
'dnaBinding': 'dnaBinding_dist', 'activeSite': 'activeSite_dist',
'nucleotideBinding': 'nucleotideBinding_dist', 'lipidation': 'lipidation_dist',
'site': 'site_dist',
'transmembrane': 'transmembrane_dist', 'crosslink': 'crosslink_dist',
'mutagenesis': 'mutagenesis_dist', 'strand': 'strand_dist', 'helix': 'helix_dist',
'turn': 'turn_dist',
'metalBinding': 'metalBinding_dist', 'repeat': 'repeat_dist',
'topologicalDomain': 'topologicalDomain_dist', 'caBinding': 'caBinding_dist',
'bindingSite': 'bindingSite_dist', 'region': 'region_dist',
'signalPeptide': 'signalPeptide_dist', 'modifiedResidue': 'modifiedResidue_dist',
'zincFinger': 'zincFinger_dist', 'motif': 'motif_dist', 'coiledCoil': 'coiledCoil_dist',
'peptide': 'peptide_dist', 'transitPeptide': 'transitPeptide_dist',
'glycosylation': 'glycosylation_dist', 'propeptide': 'propeptide_dist'}, inplace=True)
data = data[
['prot_uniprotAcc', 'wt_residue', 'mut_residue', 'position', 'meta_merged', 'composition', 'polarity',
'volume',
'granthamScore', 'domains_all',
'domains_sig', 'domains_3Ddist', 'sasa', 'location_3state', 'disulfide_bin', 'intMet_bin',
'intramembrane_bin', 'naturalVariant_bin', 'dnaBinding_bin',
'activeSite_bin', 'nucleotideBinding_bin', 'lipidation_bin', 'site_bin',
'transmembrane_bin', 'crosslink_bin', 'mutagenesis_bin', 'strand_bin',
'helix_bin', 'turn_bin', 'metalBinding_bin', 'repeat_bin',
'caBinding_bin', 'topologicalDomain_bin', 'bindingSite_bin',
'region_bin', 'signalPeptide_bin', 'modifiedResidue_bin',
'zincFinger_bin', 'motif_bin', 'coiledCoil_bin', 'peptide_bin',
'transitPeptide_bin', 'glycosylation_bin', 'propeptide_bin', 'disulfide_dist', 'intMet_dist',
'intramembrane_dist',
'naturalVariant_dist', 'dnaBinding_dist', 'activeSite_dist',
'nucleotideBinding_dist', 'lipidation_dist', 'site_dist',
'transmembrane_dist', 'crosslink_dist', 'mutagenesis_dist',
'strand_dist', 'helix_dist', 'turn_dist', 'metalBinding_dist',
'repeat_dist', 'caBinding_dist', 'topologicalDomain_dist',
'bindingSite_dist', 'region_dist', 'signalPeptide_dist',
'modifiedResidue_dist', 'zincFinger_dist', 'motif_dist',
'coiledCoil_dist', 'peptide_dist', 'transitPeptide_dist',
'glycosylation_dist', 'propeptide_dist']]
ready = data.copy()
# Imputation
if (impute == 'True') or (impute == 'true') or (impute == True):
filler = [17.84, 30.8, 24.96, 13.12, 23.62, 18.97, 20.87, 29.59, 20.7, 12.7, 22.85, 17.21, 9.8, 9, 15.99,
16.82,
20.46, 24.58, 9.99, 17.43, 20.08, 30.91, 20.86, 22.14, 21.91, 28.45, 17.81, 25.12, 20.33, 22.36]
col_index = 0
for col_ in ready.columns[-30:]:
ready[col_] = ready[col_].fillna(filler[col_index])
ready[col_] = ready[col_].replace({'nan': filler[col_index]})
col_index += 1
ready['domains_3Ddist'] = ready['domains_3Ddist'].fillna(24.5)
ready['sasa'] = ready['sasa'].fillna(29.5)
ready['location_3state'] = ready['location_3state'].fillna('unknown')
elif (impute == 'False') or (impute == 'false') or (impute == False):
pass
ready = ready.replace({'nan': np.NaN})
ready.to_csv(path_to_output_files / 'featurevector_pdb.txt', sep='\t', index=False)
if len(ready) == 0:
print(
'No feature vector could be produced for input data. Please check the presence of a structure for the input proteins.')
print(ready)
print('Feature vector successfully created...')
return ready
end = timer()
hours, rem = divmod(end - start, 3600)
minutes, seconds = divmod(rem, 60)
print("Time passed: {:0>2}:{:0>2}:{:05.2f}".format(int(hours), int(minutes), seconds))
return ready |