[c66173]: / bin / DeepMod_scripts / myDetect.py

Download this file

1287 lines (1144 with data), 63.2 kB

   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
import os;
import sys;
import string;
import glob;
import time
import copy
import h5py
import numpy as np
import multiprocessing
from collections import defaultdict
from distutils.version import LooseVersion
import tempfile
import subprocess
import re;
from . import myCom
#import myCom
import tensorflow as tf
from tensorflow.contrib import rnn
from . import myMultiBiRNN
from . import EventTable
from . import MoveTable
rnn_pred_batch_size = 512
# default path for data in fast5 files
fast5_channel_id= 'UniqueGlobalKey/channel_id'
fast5_analysis = ''.join(['/', myCom.analyses_base]) #
fast5_events = myCom.basecall_events_base #
fast5_rawReads = ''.join(['/', myCom.raw_base, '/', myCom.reads_base]) #
fast5_basecall_fq = myCom.basecall_fastq_base #
fast5_signal = myCom.signal_base #
pre_base_str = 'rnn.pred.ind'
#
# get digitisation, offset, range, sampling_rate from fast5 files
#
def get_channel_info(moptions, sp_param):
if not sp_param['f5status']=="": return;
try:
channel_info = sp_param['f5reader'][fast5_channel_id].attrs
sp_param["channel_info"] = {'digitisation':channel_info['digitisation'], 'offset':channel_info['offset'], 'range':channel_info['range'], 'sampling_rate':channel_info['sampling_rate'], 'channel_number':channel_info['channel_number']}
except:
raiseError("No Channel Info", sp_param, "No Channel Info")
# print error message
def raiseError(sp_info, sp_param, errk):
sp_param['f5status'] = errk
print ('Error!!! %s in %s' % (sp_info, sp_param['mfile_path']))
sys.stdout.flush()
#
# get Albacore version used
# only support v1+ and v2+
#
def getAlbacoreVersion(moptions, sp_param):
if not sp_param['f5status']=="": return;
try:
ver_path = ''.join([fast5_analysis,'/', moptions['basecall_1d'] ])
#add .decode("utf-8") to make it compatible to py3
try:
used_version = LooseVersion(sp_param['f5reader'][ver_path].attrs['version'].decode("utf-8") if 'version' in sp_param['f5reader'][ver_path].attrs else "0.0")
except:
used_version = LooseVersion(sp_param['f5reader'][ver_path].attrs['version'] if 'version' in sp_param['f5reader'][ver_path].attrs else "0.0")
sp_param['get_albacore_version'] = used_version
if used_version < LooseVersion("1.0"): #
sp_param['used_albacore_version'] = 1;
elif used_version < LooseVersion("2.0"): sp_param['used_albacore_version'] = 1;
elif used_version >= LooseVersion("2.0"): sp_param['used_albacore_version'] = 2;
except: # default verion is 1 now
sp_param['used_albacore_version'] = 1;
# not used now.
def get_kmer_corrected_info(moptions):
if ('kmer_model_file' not in moptions) or moptions['kmer_model_file']==None or (not os.path.isfile(moptions['kmer_model_file'])): return;
fr = open(moptions['kmer_model_file'], 'r')
moptions['kmer_model_dict'] = defaultdict()
line = fr.readline();
while line:
line = string.strip(line);
if len(line)>0 and (not line[0]=='#'):
try:
c_kmer, c_level_mean, c_level_stdv = line.split()[:3]
c_level_mean, c_level_stdv = float(c_level_mean), float(c_level_stdv)
moptions['kmer_model_dict'][c_kmer] = (c_level_mean, 1/(c_level_stdv*c_level_stdv))
except:
pass;
line = fr.readline();
fr.close();
# not used now
# get shift and scale values for normalization
#
def get_cur_shift_scale(moptions, sp_param):
if not sp_param['f5status']=="": return;
if "kmer_model_dict" not in moptions: return;
event_key = 'm_event'
try:
cur_model = np.array([moptions['kmer_model_dict'][c_model_state] for c_model_state in sp_param[event_key]['model_state']], dtype=[('level_mean', np.float), ('level_stdv', np.float)]);
c_mean_stdv = cur_model['level_mean']*cur_model['level_stdv']
c_mean_stdv_sum = c_mean_stdv.sum()
model_coef_matrix = np.array(( (cur_model['level_stdv'].sum(), c_mean_stdv_sum), \
(c_mean_stdv_sum, (c_mean_stdv*cur_model['level_mean']).sum()) \
))
c_event_stdv = sp_param[event_key]['mean'] * cur_model['level_stdv']
c_event_stdv_mean = c_event_stdv * cur_model['level_mean']
dependent_array = np.array((c_event_stdv.sum(), c_event_stdv_mean.sum()));
sp_param['shift_scale'] = {}
sp_param['shift_scale']['cal_shift'], sp_param['shift_scale']['cal_scale'] = np.linalg.solve(model_coef_matrix, dependent_array)
sp_param['shift_scale']['chn_shift'], sp_param['shift_scale']['chn_scale'] = -sp_param["channel_info"]['offset'], sp_param["channel_info"]['digitisation']/sp_param["channel_info"]['range']
sp_param['shift_scale']['shift']=sp_param['shift_scale']['chn_shift']+sp_param['shift_scale']['chn_scale']*sp_param['shift_scale']['cal_shift']
sp_param['shift_scale']['scale']=sp_param['shift_scale']['chn_scale']*sp_param['shift_scale']['cal_scale']
sp_param['raw_signals'] = np.round(sp_param['raw_signals']/sp_param['shift_scale']['cal_scale'] - sp_param['shift_scale']['cal_shift']/sp_param['shift_scale']['cal_scale'], 6)
except:
raiseError('Cannot nanopore correction', sp_param, "Cannot nanopore correction")
#
# get events from a fast5 file
#
def getEvent(moptions, sp_param):
if not sp_param['f5status']=="": return;
# If use move tables intead of event tables
if moptions['move']:
try: # get events from a fast5 file'
mv_str = '/'.join(['', 'Analyses', moptions['basecall_1d'], moptions['basecall_2strand'], 'Move'])
move_data = sp_param['f5reader'][mv_str][()]
sp_param['events_data'] = move_data
except:
raiseError('No move data', sp_param, "No move data")
return;
m_event = MoveTable.getMove_Info(moptions, sp_param, move_data)
sp_param['m_event'] = m_event
# get sequence from events
sp_param['m_event_basecall'] = sp_param['fq_seq']
sp_param['left_right_skip'] = (0, 0)
return
# End the part of getting move tables
try: # get events from a fast5 file
event_path = ''.join([fast5_analysis, '/', moptions['basecall_1d'], '/', moptions['basecall_2strand'], '/', fast5_events])
events_data = sp_param['f5reader'][event_path].value
except:
raiseError('No events data', sp_param, "No events data")
return;
convertError = False;
if sp_param['f5status'] == "":
sp_param['events_data'] = events_data
if sp_param['used_albacore_version']==1:
move0_left = 0; move0_right = len(events_data)-1;
while move0_left<move0_right: # get the first non-stay event at the left tail
if events_data['move'][move0_left]==0: move0_left += 1;
else: break;
if move0_left>move0_right-20:
raiseError(("Too many move0 at 3'(l%d, r%d)" % (move0_left, move0_right)), sp_param, "Remove too many bases on left")
return;
while move0_right>move0_left: # get the last non-stay event at the right tail
if events_data['move'][move0_right]==0: move0_right -= 1
else: break;
if move0_right<move0_left+20:
raiseError(("Too many move0 at 5'(l%d, r%d)" % (move0_left, move0_right)), sp_param, 'Remove too many bases on right')
return
# get the starting time
based_ind = events_data['start'][move0_left].astype(np.float64)*sp_param["channel_info"]["sampling_rate"] - sp_param['raw_attributes']['start_time']
first_base_index_in_raw_signal = np.round(events_data['start'][move0_left].astype(np.float64)*sp_param["channel_info"]["sampling_rate"]).astype(np.int64) - sp_param['raw_attributes']['start_time']
# get the potential error of the starting time
if first_base_index_in_raw_signal<-2:
raiseError(('The index of the first base is less than -2(%d=%.6f*%d-%d)' % (first_base_index_in_raw_signal, events_data['start'][move0_left].astype(np.float64), sp_param["channel_info"]["sampling_rate"], sp_param['raw_attributes']['start_time'])), sp_param, "The index of the first base is less than -2")
return;
elif first_base_index_in_raw_signal<0:
first_base_index_in_raw_signal = 0
if moptions['outLevel']<=myCom.OUTPUT_INFO: print ('Warning!!! first_base_index_in_raw_signal less than 0 ' + sp_param['mfile_path'])
first_base_index_in_raw_signal = np.uint64(first_base_index_in_raw_signal)
m_event = []; pre_i = move0_left;
cur_length=(events_data['length'][pre_i]*sp_param["channel_info"]["sampling_rate"]).astype('uint64');
for i in range(move0_left+1, move0_right+1):
if events_data['move'][i]>0: # for non-stay event
if pre_i==move0_left:
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), first_base_index_in_raw_signal, cur_length, events_data['model_state'][pre_i].upper()))
else: # calculate starting index in raw signal
# calculated position
cal_st = (events_data['start'][pre_i]-events_data['start'][move0_left])*sp_param["channel_info"]["sampling_rate"]+based_ind
if cal_st<0: print("Wanging Less than 0")
if cal_st>0 and cal_st - (m_event[-1][2]+ m_event[-1][3])>0 and (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')>0:
if (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')>2: #
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64'), events_data['model_state'][pre_i].upper()))
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), cal_st.astype('uint64'), cur_length, events_data['model_state'][pre_i].upper()))
else: # for a normal event
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')+ cur_length, events_data['model_state'][pre_i].upper()))
else:
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], cur_length, events_data['model_state'][pre_i].upper()))
if m_event[-1][2]>np.iinfo(np.int64).max-2 or m_event[-1][2]<0:
if not convertError:
print ('ex: %.7f*%d=%.0f' % (events_data['start'][move0_left].astype(np.float64), sp_param["channel_info"]["sampling_rate"], events_data['start'][move0_left].astype(np.float64)*sp_param["channel_info"]["sampling_rate"])), sp_param['raw_attributes']['start_time'], sp_param['mfile_path'], m_event[-1][2], m_event[-1][3]
convertError = True;
pre_i = i;
cur_length=(events_data['length'][i]*sp_param["channel_info"]["sampling_rate"]).astype('uint64');
else: # for stay event
cur_length += (events_data['length'][i]*sp_param["channel_info"]["sampling_rate"]).astype('uint64')
if sp_param['f5status'] == "": # for the last event
# calculated position
cal_st = (events_data['start'][pre_i]-events_data['start'][move0_left])*sp_param["channel_info"]["sampling_rate"]+based_ind
if cal_st<0: print("Wanging Less than 0")
if cal_st>0 and cal_st - (m_event[-1][2]+ m_event[-1][3])>0 and (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')>0:
if (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')>2:
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64'), events_data['model_state'][pre_i].upper()))
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), cal_st.astype('uint64'), cur_length, events_data['model_state'][pre_i].upper()))
else:
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], (cal_st - (m_event[-1][2]+ m_event[-1][3])).astype('uint64')+ cur_length, events_data['model_state'][pre_i].upper()))
else:
m_event.append((round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), m_event[-1][2]+ m_event[-1][3], cur_length, events_data['model_state'][pre_i].upper()))
# decode
m_event = np.array(m_event, dtype=[('mean', '<f4'), ('stdv', '<f4'), ('start', np.uint64), ('length', np.uint64), ('model_state', 'U5')]) #'S5')]
sp_param['m_event'] = m_event
# get sequence from events
sp_param['m_event_basecall'] = ''.join([event_model_state[2] for event_model_state in m_event['model_state']]);
sp_param['left_right_skip'] = (move0_left, len(events_data)-move0_right-1)
elif sp_param['used_albacore_version']==2:
if moptions['SignalGroup']=='simple':
m_event = [];
pre_i = 0; pre_length = events_data['length'][pre_i].astype('uint64');
for cur_i in range(1, len(events_data)):
if events_data['move'][cur_i]>0: # non-stay vents
m_event.append( (round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), events_data['start'][pre_i], pre_length, events_data['model_state'][pre_i]) )
pre_i = cur_i; pre_length = events_data['length'][pre_i].astype('uint64');
else: # for stay events
pre_length += events_data['length'][cur_i].astype('uint64');
m_event.append( (round(events_data['mean'][pre_i],3), round(events_data['stdv'][pre_i],3), events_data['start'][pre_i], pre_length, events_data['model_state'][pre_i]) )
# format events
m_event = np.array(m_event, dtype=[('mean', '<f4'), ('stdv', '<f4'), ('start', np.uint64), ('length', np.uint64), ('model_state', 'U5')])
else:
m_event = EventTable.getEvent_Info(moptions, sp_param, events_data)
sp_param['m_event'] = m_event
# get sequence from events
sp_param['m_event_basecall'] = ''.join([event_model_state[2] for event_model_state in m_event['model_state']]);
sp_param['left_right_skip'] = (0, 0)
else:
raise RuntimeError ("This version of Albacore is not supported. Please use the version of Albacore 1.x or 2.x")
#
# normalize raw signals
#
def mnormalized(moptions, sp_param):
if not sp_param['m_event']['start'][0] < (sp_param['m_event']['start'][-1]+sp_param['m_event']['length'][-1]):
print ('Fatal error signal start position is less than the end position', sp_param['mfile_path'], sp_param['m_event']['start'][0], sp_param['m_event']['start'][-1], sp_param['m_event']['length'][-1])
# get shift and scale
mshift = np.median(sp_param['raw_signals'][sp_param['m_event']['start'][0]:(sp_param['m_event']['start'][-1]+sp_param['m_event']['length'][-1])])
mscale = np.median(np.abs(sp_param['raw_signals'][sp_param['m_event']['start'][0]:(sp_param['m_event']['start'][-1]+sp_param['m_event']['length'][-1])]-mshift));
# standardize
sp_param['raw_signals'] = (sp_param['raw_signals'] - mshift)/mscale
# get meand
read_med = np.median(sp_param['raw_signals'][sp_param['m_event']['start'][0]:(sp_param['m_event']['start'][-1]+sp_param['m_event']['length'][-1])])
read_mad = np.median(np.abs(sp_param['raw_signals'][sp_param['m_event']['start'][0]:(sp_param['m_event']['start'][-1]+sp_param['m_event']['length'][-1])] - read_med))
lower_lim = read_med - (read_mad * 5)
upper_lim = read_med + (read_mad * 5)
# normalize as nanoraw did.
sp_param['raw_signals'] = np.round(np.array([upper_lim if sp_param['raw_signals'][i]>upper_lim else (lower_lim if sp_param['raw_signals'][i]<lower_lim else sp_param['raw_signals'][i]) for i in range(np.size(sp_param['raw_signals']))]), 3)
#
# get Signal from a fast5 file
#
def getRawInfo(moptions, sp_param):
if not sp_param['f5status']=="": return;
try:
# get attribute of raw signals
for raw_data in sp_param['f5reader'][fast5_rawReads].values(): pass;
sp_param['raw_attributes'] = dict(raw_data.attrs.items())
sp_param['raw_signals'] = raw_data['Signal'][()]
except:
raiseError(("No Raw_reads/Signal data %s" % (fast5_rawReads)), sp_param, "No Raw_reads/Signal")
#
# get channel_info, AlbacoreVersion, read_id, Raw Signals, Event from a fast5 file
#
def getFast5Info(moptions, sp_param):
# get channel info
get_channel_info(moptions, sp_param)
if "channel_info" not in sp_param:
raiseError(("Channel information could not be found in %s " % fast5_channel_id), sp_param, "Channel information could not be found")
return;
# get albacore version
getAlbacoreVersion(moptions, sp_param)
if 'used_albacore_version' not in sp_param:
return
try:
# get fastq attribute
fq_path = ''.join([fast5_analysis,'/',moptions['basecall_1d'],'/',moptions['basecall_2strand'],'/',fast5_basecall_fq])
fq_data = sp_param['f5reader'][fq_path][()]
except:
raiseError('No Fastq data', sp_param, "No Fastq data")
if sp_param['f5status']=="":
fq_data = (fq_data.decode(encoding="utf-8")).split('\n')
sp_param['read_id'] = (fq_data[0][1:] if fq_data[0][0]=='@' else fq_data[0]).replace(" ", ":::").replace("\t", "|||")
sp_param['fq_seq'] = fq_data[1];
# get raw signals
getRawInfo(moptions, sp_param)
# get events
if sp_param['f5status']=="":
getEvent(moptions, sp_param)
# normalize signals.
if sp_param['f5status']=="":
mnormalized(moptions, sp_param)
if sp_param['f5status']=="":
# get mean, std for each event
for i in range(len(sp_param['m_event'])):
if (len(sp_param['raw_signals'][sp_param['m_event']['start'][i]:sp_param['m_event']['start'][i]+sp_param['m_event']['length'][i]])==0):
print ('Signal out of range {}: {}-{} {};{} for {}'.format(i, sp_param['m_event']['start'][i], sp_param['m_event']['length'][i], len(sp_param['m_event']), len(sp_param['raw_signals']), sp_param['mfile_path']))
if i>500:
sp_param['m_event'] = sp_param['m_event'][:i-1]
else:
sp_param['f5status']=="Less event"
break;
sp_param['m_event']['mean'][i] = round(np.mean(sp_param['raw_signals'][sp_param['m_event']['start'][i]:sp_param['m_event']['start'][i]+sp_param['m_event']['length'][i]]), 3)
sp_param['m_event']['stdv'][i] = round(np.std(sp_param['raw_signals'][sp_param['m_event']['start'][i]:sp_param['m_event']['start'][i]+sp_param['m_event']['length'][i]]), 3)
#
# associate signals for each event in a fast5 file
#
def get_Event_Signals(moptions, sp_options, f5files):
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
start_time = time.time(); runnum = 0;
f5data = {}
sp_options["Error"] = defaultdict(list)
sp_options["get_albacore_version"] = defaultdict(int)
# for each fast5 file
for f5f in f5files:
try:
with h5py.File(f5f, 'r') as mf5:
sp_param = {}
sp_param['mfile_path'] = f5f
sp_param['f5reader'] = mf5
sp_param['f5status'] = "";
getFast5Info(moptions, sp_param)
if 'get_albacore_version' in sp_param:
sp_options["get_albacore_version"][str(sp_param['get_albacore_version'])] += 1
if sp_param['f5status'] == "":
if sp_param['read_id'] in f5data:
print ('Duplicate id', sp_param['read_id'], f5f)
f5data[sp_param['read_id']] = (sp_param['m_event_basecall'], sp_param['m_event'], sp_param['raw_signals'], f5f, sp_param['left_right_skip'])
else:
sp_options["Error"][sp_param['f5status']].append(f5f)
# for outputing progress
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
runnum += 1;
if runnum%500==0:
end_time = time.time();
print ("%d consuming time %d" % (runnum, end_time-start_time))
except:
sp_options["Error"]["Cannot open fast5 or other errors"].append(f5f)
print("Cannot open fast5 or other errors: {}".format(f5f))
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
end_time = time.time();
print ("All consuming time %d" % (end_time-start_time))
return f5data;
#
# get signals of events
# map bases from events to a reference genome
#
def mDetect1(moptions, sp_options, f5files):
f5data = get_Event_Signals(moptions, sp_options, f5files)
if moptions['outLevel']<=myCom.OUTPUT_DEBUG: start_time = time.time();
# for fa files of base sequences from events
temp_fa = tempfile.NamedTemporaryFile(suffix='.fa', mode='w')
f5keys = sorted(f5data.keys()); #f5keys.sort()
for f5k in f5keys:
temp_fa.write(''.join(['>', f5k, '\n', f5data[f5k][0], '\n']))
temp_fa.flush();
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
end_time = time.time();
print ("Write consuming time %d" % (end_time-start_time))
# run alignmen tools of bwa-mem or minimap2
temp_sam = tempfile.NamedTemporaryFile()
if moptions['alignStr']=='bwa':
cmd_opt = ['mem', '-x', 'ont2d', '-v', '1', '-t', '1', moptions['Ref'], temp_fa.name]
else:
cmd_opt = ['-ax', 'map-ont', moptions['Ref'], temp_fa.name]
returncode = subprocess.call([moptions['alignStr'],]+cmd_opt, stdout=temp_sam)
if not returncode==0:
print ('Fatal Error!!! returncode is non-zero(%d) for "%s"' % (returncode, curcmd))
errkey = "Cannot running aligment"
for f5k in f5keys:
sp_options["Error"][errkey].append(f5data[f5k][3])
return;
temp_fa.close();
temp_sam.seek(0);
# get content from aligned results
align_info = temp_sam.readlines()
align_info = [str(align_info[i], 'utf-8').strip() for i in range(len(align_info))]
temp_sam.close();
sp_param = defaultdict();
sp_param['f5data'] = f5data
f5align = defaultdict()
f5keydict = defaultdict();
sp_param['ref_info'] = defaultdict()
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:start_time = time.time();
ilid = 0;
# get alignment records
while ilid < len(align_info):
if len(align_info[ilid])==0 or align_info[ilid][0]=='@':
ilid += 1
continue;
sp_param['f5status'] = "";
sp_param['line'] = align_info[ilid]
qname = handle_line(moptions, sp_param, f5align)
if sp_param['f5status'] == "":
f5keydict[qname] = True;
ilid += 1
# get unmapped reads
for f5k in f5keys:
if f5k not in f5keydict:
sp_options["Error"]["Not in alignment sam"].append(f5data[f5k][3])
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
end_time = time.time();
print ("Get BAM consuming time %d" % (end_time-start_time))
# get features, prediction for each fast5 files
sp_param['f5status']= ""
sp_param['line'] = ""
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:start_time = time.time();
handle_record(moptions, sp_options, sp_param, f5align, f5data)
if moptions['outLevel']<=myCom.OUTPUT_DEBUG:
end_time = time.time();
print ("Analyze & annotate & save consuming time %d" % (end_time-start_time))
#
# get reference sequenceng from a reference genome
#
def getRefSeq(moptions, sp_param, rname):
temp_seq = tempfile.NamedTemporaryFile()
# command to be run
cmd_opt = ['faidx', moptions['Ref'], rname]
returncode = subprocess.call(['samtools',]+cmd_opt, stdout=temp_seq)
if not returncode==0:
print ('Fatal Error!!! cannot find the chrosome sequence %s' % rname)
else:
temp_seq.seek(0);
seqinfo = temp_seq.readlines()
seqinfo = [str(seqinfo[i], 'utf-8').strip() for i in range(len(seqinfo))]
temp_seq.close();
sp_param['ref_info'][rname] = ''.join(seqinfo[1:]).strip().upper()
#
# get mapping information and associate it with events/signals.
#
def handle_record(moptions, sp_options, sp_param, f5align, f5data):
alignkeys = list(f5align.keys());
numreg = re.compile('\d+')
mdireg = re.compile('[MIDNSHPX=]{1}')
# for each alignment record
for readk_ind in range(len(alignkeys)):
sp_param['f5status']= ""
readk = alignkeys[readk_ind]
sp_param['mfile_path'] = f5data[readk][3]
mapq, flag, rname, pos, cigar, readseq = f5align[readk]
# output chromosome of interest
if (not moptions['ConUnk']) and ((not rname.find('_')==-1) or (not rname.find('-')==-1) or (not rname.find('/')==-1) or (not rname.find(':')==-1)):
continue;
isinreg = False;
# check the region of interest
for cur_mr in moptions['region']:
if (cur_mr[0] in ['', None, rname]):
isinreg = True;
break;
if not isinreg:
continue;
# get reference information
if rname not in sp_param['ref_info']:
getRefSeq(moptions, sp_param, rname)
refseq = sp_param['ref_info'][rname]
# mapped position and strand
pos = pos - 1
forward_reverse = '-' if flag&0x10 else '+'
numinfo = numreg.findall(cigar);
mdiinfo = mdireg.findall(cigar)
numinfo = [int(numinfo[i]) for i in range(len(numinfo))] #map(int, numinfo)
leftclip = 0; rightclip = 0;
# remove left clip
while mdiinfo[0] in ['I', 'D', 'N', 'S', 'H', 'P', 'X']:
if mdiinfo[0] in ['I', 'S', 'X']:
leftclip += numinfo[0]; readseq = readseq[numinfo[0]:]
if mdiinfo[0] in ['H']: leftclip += numinfo[0]
if mdiinfo[0] in ['D', 'N', 'X']:
pos += numinfo[0]
numinfo = numinfo[1:]; mdiinfo = mdiinfo[1:]
# remove right clip
while mdiinfo[-1] in ['I', 'D', 'N', 'S', 'H', 'P', 'X']:
if mdiinfo[-1] in ['I', 'S', 'X']:
rightclip += numinfo[-1]; readseq = readseq[:-numinfo[-1]]
if mdiinfo[-1] in ['H']: rightclip += numinfo[-1]
numinfo = numinfo[:-1]; mdiinfo = mdiinfo[:-1]
if forward_reverse=='+': # remove clipped events
if rightclip>0: m_event = f5data[readk][1][leftclip:-rightclip]
else: m_event = f5data[readk][1][leftclip:]
else:
if leftclip>0: m_event = f5data[readk][1][rightclip:-leftclip]
else: m_event = f5data[readk][1][rightclip:]
isinreg = False;
# for specify regions
for cur_mr in moptions['region']:
if (cur_mr[0] in ['', None, rname]) and \
(cur_mr[1] in ['', None] or pos>cur_mr[1]) and \
(cur_mr[2] in ['', None] or pos+len(m_event)<cur_mr[2]):
isinreg = True;
break;
### for check consistency
if not isinreg:
continue;
lastmatch = None; firstmatch = None;
first_match_pos = None; last_match_pos = None
last_al_match = None; first_al_match = None
lasmtind = 0;
base_map_info = []; #indel_groups = defaultdict()
nummismatch = 0; numinsert = 0; numdel = 0;
read_ind = 0;
# get map detail: reference base, read base, ref position, read-position
for n1ind in range(len(numinfo)):
mdi = mdiinfo[n1ind];
for n1i in range(numinfo[n1ind]):
# for each cigar type
if mdi=='M':
base_map_info.append((refseq[pos], readseq[read_ind], pos, read_ind, 0))
if refseq[pos]==readseq[read_ind]:
if firstmatch==None: firstmatch = read_ind
if lastmatch==None or lastmatch<read_ind: lastmatch = read_ind; lasmtind=n1ind
if first_al_match==None: first_al_match=len(base_map_info)-1
if last_al_match==None or last_al_match<len(base_map_info): last_al_match=len(base_map_info)-1
if first_match_pos==None: first_match_pos = pos
if last_match_pos==None or last_match_pos<pos: last_match_pos = pos
else: nummismatch += 1
pos += 1; read_ind += 1;
elif mdi =='I':
base_map_info.append(('-', readseq[read_ind], pos, read_ind, 0))
read_ind += 1;
numinsert += 1
elif mdi == 'D':
base_map_info.append((refseq[pos], '-', pos, read_ind, 0))
pos += 1;
numdel += 1
elif mdi == 'N':
base_map_info.append((refseq[pos], '-', pos, read_ind, 0))
pos += 1;
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ('CIGAR-Error N exist', f5data[readk][3])
elif mdi == 'S':
read_ind += 1;
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ('CIGAR-Error!!! S in the middle of the sequence', f5data[readk][3])
elif mdi == 'H':
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ('CIGAR-Error!!! H in the middle of the sequence', f5data[readk][3])
elif mdi == 'P':
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ('CIGAR-Error!!! P exist', f5data[readk][3])
elif mdi == '=':
base_map_info.append((refseq[pos], readseq[read_ind], pos, read_ind, 0))
if first_match_pos==None: first_match_pos = pos
if last_match_pos==None or last_match_pos<pos: last_match_pos = pos
pos += 1; read_ind += 1;
if firstmatch==None: firstmatch = read_ind - 1
if lastmatch==None or lastmatch<read_ind-1: lastmatch = read_ind - 1; lasmtind=n1ind
if last_al_match==None or last_al_match<len(base_map_info): last_al_match=len(base_map_info)-1
if first_al_match==None: first_al_match=len(base_map_info)-1
elif mdi == 'X':
base_map_info.append((refseq[pos], readseq[read_ind], pos, read_ind, 0))
pos += 1; read_ind += 1;
nummismatch += 1
else:
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ('CIGAR-Error!!!', 'Warning unknow CIGAR element ' + str(numinfo[n1ind]) + ' ' + mdi, f5data[readk][3])
if firstmatch==None or lastmatch==None or firstmatch<0 or lastmatch<0:
if moptions['outLevel']<=myCom.OUTPUT_WARNING:
print ("Errorfast5 "+f5data[readk][3])
print('match-Error!!! no first and/or last match',f5data[readk][3],('firstmatch=%d' % firstmatch) if not (firstmatch==None) else "N", ('lastmatch%d' % lastmatch) if not (lastmatch==None) else "N", str(flag), rname, str(pos));
print('\tf=%d, chr=%s, p=%d, c=%s, s=%s' % (flag, rname, pos, cigar, readseq))
continue;
# remove more un-matched events
if forward_reverse=='+':
if not firstmatch==None: leftclip += firstmatch
if (not lastmatch==None) and len(m_event)-lastmatch>1: rightclip += len(m_event)-lastmatch-1
else:
if not firstmatch==None: rightclip += firstmatch
if (not lastmatch==None) and len(m_event)-lastmatch>1: leftclip += len(m_event)-lastmatch-1
if forward_reverse=='+':
if len(m_event)-lastmatch>1:
m_event = m_event[firstmatch:(lastmatch+1-len(m_event))]
elif firstmatch>0: m_event = m_event[firstmatch:]
else:
if firstmatch>0: m_event = m_event[(len(m_event)-1-lastmatch):-firstmatch]
elif len(m_event)-lastmatch>1: m_event = m_event[(len(m_event)-1-lastmatch):]
# check potential error
if firstmatch>0 or len(base_map_info)-last_al_match>1:
if moptions['outLevel']<=myCom.OUTPUT_WARNING and ((firstmatch>0) or (len(base_map_info)-last_al_match>1 and refseq[last_match_pos+1] not in ['N'])):
print ("Errorfast5"+f5data[readk][3])
print ('Warning!!! first not match', firstmatch, lastmatch, first_al_match, last_al_match, len(base_map_info), numinfo[lasmtind-2:(lasmtind+5)], mdiinfo[lasmtind-2:(lasmtind+5)], lasmtind, len(numinfo))
print('\tref='+refseq[last_match_pos:last_match_pos+20]+"\n\tred="+readseq[lastmatch:lastmatch+20])
if firstmatch>0:
print('\tref='+refseq[(first_match_pos-20 if first_match_pos-20>0 else 0):first_match_pos]+"\n\tred="+readseq[(firstmatch-20 if firstmatch-20>0 else 0):firstmatch])
print('\tf=%d, chr=%s, p=%d, c=%s, s=%s' % (flag, rname, pos, cigar, readseq)) # flag, rname, pos, cigar, readseq
if len(base_map_info)-last_al_match>1:
base_map_info = base_map_info[first_al_match:(last_al_match+1-len(base_map_info))]
elif first_al_match>0:
base_map_info = base_map_info[first_al_match:]
# format base
base_map_info = np.array(base_map_info, dtype=[('refbase', 'U1'), ('readbase', 'U1'), ('refbasei', np.uint64), ('readbasei', np.uint64), ('mod_pred', np.int)])
if forward_reverse=='-':
base_map_info = np.flipud(base_map_info)
for bmii in range(len(base_map_info)):
base_map_info['refbase'][bmii] = get_complement(base_map_info['refbase'][bmii])
base_map_info['readbase'][bmii] = get_complement(base_map_info['readbase'][bmii])
leftclip, rightclip = rightclip, leftclip
if False: #True: # for test base_map_info ### for check consistency
ref_align_key = '/Analyses/NanomoCorrected_000/BaseCalled_template/Alignment/genome_alignment'
read_align_key = '/Analyses/NanomoCorrected_000/BaseCalled_template/Alignment/read_alignment'
with h5py.File(f5data[readk][3], 'r') as mf5:
read_align_list = [bt.decode(encoding="utf-8") for bt in mf5[read_align_key]]
ref_align_list = [bt.decode(encoding="utf-8") for bt in mf5[ref_align_key]]
for rali in range(len(read_align_list)):
if not read_align_list[rali]==base_map_info['readbase'][rali]:
print ("Error not equal1! %s %s %d %s" % (read_align_list[rali], base_map_info['readbase'][rali], rali, f5data[readk][3]))
if not ref_align_list[rali]==base_map_info['refbase'][rali]:
print ("Error not equal2! %s %s %d %s" % (ref_align_list[rali], base_map_info['refbase'][rali], rali, f5data[readk][3]))
# ## for check consistency
if True:
# for special alignment
# CCG or CGG
# C-G C-G
for ali in range(len(base_map_info)):
if base_map_info['refbase'][ali]=='C' and base_map_info['readbase'][ali]=='C':
if ali+1<len(base_map_info) and base_map_info['readbase'][ali+1]=='-' and base_map_info['refbase'][ali+1]=='G':
addali = 2;
while ali + addali < len(base_map_info):
if base_map_info['readbase'][ali+addali]=='-' and base_map_info['refbase'][ali+addali]=='G': addali += 1;
else: break;
if ali + addali < len(base_map_info) and base_map_info['readbase'][ali+addali]=='G' and base_map_info['refbase'][ali+addali]=='G':
base_map_info['readbase'][ali+1], base_map_info['readbase'][ali+addali] = base_map_info['readbase'][ali+addali], base_map_info['readbase'][ali+1]
if base_map_info['refbase'][ali]=='G' and base_map_info['readbase'][ali]=='G':
if ali-1>-1 and base_map_info['readbase'][ali-1]=='-' and base_map_info['refbase'][ali-1]=='C':
addali = 2;
while ali - addali >-1:
if base_map_info['readbase'][ali-addali]=='-' and base_map_info['refbase'][ali-addali]=='C': addali += 1;
else: break;
if ali - addali>-1 and base_map_info['readbase'][ali-addali]=='C' and base_map_info['refbase'][ali-addali]=='C':
base_map_info['readbase'][ali-1], base_map_info['readbase'][ali-addali] = base_map_info['readbase'][ali-addali], base_map_info['readbase'][ali-1]
if len(m_event)<50:
raiseError("Less Event", sp_param, "Less Event");
sp_options["Error"]["Less Event"].append(f5data[readk][3])
continue;
# get feature
mfeatures,isdif = get_Feature(moptions, sp_options, sp_param, f5align, f5data, readk, leftclip, rightclip, base_map_info, forward_reverse, rname, first_match_pos, numinsert, numdel)
if isdif and moptions['outLevel']<=myCom.OUTPUT_WARNING:
print("Dif is true")
print([lastmatch, firstmatch, first_match_pos, last_match_pos, first_al_match, last_al_match, lasmtind, len(base_map_info), nummismatch, numinsert, numdel, len(base_map_info)-nummismatch-numinsert-numdel])
if not sp_param['f5status']=="": continue
# generate/save prediction information
pred_mod_num = mPredict1(moptions, sp_options, sp_param, mfeatures, base_map_info, readk, leftclip, rightclip)
predfile = (sp_options['ctfolder'] if sp_options['ctfolder'][-1] not in ['/', '\\'] else sp_options['ctfolder'][:-1])+'/rnn.pred.detail.fast5'+'.'+str(sp_options['batchid'])
pred_f5_key = 'pred_'+str(readk_ind)
sp_options['Mod'].append([rname, forward_reverse, f5align[readk][3]-1, pred_f5_key, f5data[readk][3][len(moptions['wrkBase'])+1:], predfile[len(moptions['outFolder']+moptions['FileID'])+1:]])
# save the prediction
with h5py.File(predfile, 'a') as save_data:
if "pred" in save_data: base_group = save_data["pred"]
else: base_group = save_data.create_group("pred")
if pred_f5_key in base_group:
del base_group[pred_f5_key]
save_data.flush()
pred_group = base_group.create_group(pred_f5_key)
# save mapped chr, strand, positions
pred_group.attrs['mapped_chr'] = rname
pred_group.attrs['mapped_strand'] = forward_reverse
pred_group.attrs['mapped_start'] = base_map_info['refbasei'][0] if forward_reverse=='+' else base_map_info['refbasei'][-1]
pred_group.attrs['mapped_end'] = base_map_info['refbasei'][-1] if forward_reverse=='+' else base_map_info['refbasei'][0]
if forward_reverse=='+':
pred_group.attrs['clipped_bases_start'] = leftclip
pred_group.attrs['clipped_bases_end'] = rightclip
else:
pred_group.attrs['clipped_bases_start'] = rightclip
pred_group.attrs['clipped_bases_end'] = leftclip
# save indel, matches/mismatches
pred_group.attrs['num_insertions'] = numinsert
pred_group.attrs['num_deletions'] = numdel
pred_group.attrs['num_matches'] = len(base_map_info)-nummismatch-numinsert-numdel
pred_group.attrs['num_mismatches'] = nummismatch
# save the fast5 file and prediction
pred_group.attrs['pred_mod_num'] = pred_mod_num
pred_group.attrs['f5file'] = f5data[readk][3]
pred_group.attrs['readk'] = readk
base_map_info = np.array(base_map_info, dtype=[('refbase', 'S1'), ('readbase', 'S1'), ('refbasei', np.uint64), ('readbasei', np.uint64), ('mod_pred', np.int)])
pred_group.create_dataset('predetail', data=base_map_info, compression="gzip")
try:
save_data.flush();
save_data.close();
except:
sp_options["Error"]['Cannot save data'].append(f5data[readk][3])
print ('Error!!! %s in %s' % ("Cannot save data", f5data[readk][3]))
# save index information
sp_options['Mod'] = sorted(sp_options['Mod'])
# index file
pred_ind_file = (sp_options['ctfolder'] if sp_options['ctfolder'][-1] not in ['/', '\\'] else sp_options['ctfolder'][:-1])+'/%s.' + pre_base_str + '.' + str(sp_options['batchid'])
if len(sp_options['Mod'])>0:
cur_chr = None; cur_writer = None;
for mfi in sp_options['Mod']:
if cur_chr==None or (not cur_chr == mfi[0]):
if not cur_chr==None:
cur_writer.flush();
cur_writer.close()
cur_chr = mfi[0]
cur_writer = open((pred_ind_file % cur_chr), 'w')
cur_m_f = []
for mfidetail in mfi:
cur_m_f.append(str(mfidetail))
cur_m_f.append('\n')
cur_writer.write(' '.join(cur_m_f))
if not cur_writer==None:
cur_writer.flush();
cur_writer.close()
#
# make modificatoin prediction for a long read
#
def mPredict1(moptions, sp_options, sp_param, mfeatures, base_map_info, readk, start_clip, end_clip):
#
modevents = sp_param['f5data'][readk][1]
# get features. labels might be all zero
t0, ty, tx = np.split(mfeatures, [1,3], axis=1);
t0 = t0.astype(int)
m_data = []; m_y = [];
for ie in range(start_clip-100, len(modevents)-end_clip+100):
mind = ie - (start_clip-100)
if ie>=start_clip and ie<len(modevents)-end_clip:
m_y.append(ty[mind])
# format to input with windoe size
m_data.append(tx[(mind-int(moptions['windowsize']/2)):(mind+int(moptions['windowsize']/2)+1)])
# for input
test_feature = np.reshape(m_data, (len(m_data), len(m_data[0]), len(m_data[0][0])))
test_label = np.reshape(m_y, (len(m_y), len(m_y[0]))).astype(int)
sp_options['rnn'][0].run(sp_options['rnn'][3])
# split into small group
if len(test_feature) > rnn_pred_batch_size*1.2:
x_sub_group = np.array_split(test_feature, int(len(test_feature)/rnn_pred_batch_size))
y_sub_group = np.array_split(test_label, int(len(test_feature)/rnn_pred_batch_size))
else:
x_sub_group = [test_feature]; y_sub_group = [test_label]
# make prediction on each small groups
for subi in range(len(x_sub_group)):
if subi==0:
mfpred_output = (sp_options['rnn'][0].run([sp_options['rnn'][4]], \
feed_dict={sp_options['rnn'][1]:x_sub_group[subi], sp_options['rnn'][2]:y_sub_group[subi]}))[0];
else:
mfpred_output = np.concatenate((mfpred_output, (sp_options['rnn'][0].run([sp_options['rnn'][4]], \
feed_dict={sp_options['rnn'][1]:x_sub_group[subi], sp_options['rnn'][2]:y_sub_group[subi]}))[0]), axis=0);
# associate the prediction with reference positions and read positions
modevents = sp_param['f5data'][readk][1]
aligni = 0; pred_mod_num = 0;
for ie in range(start_clip, len(modevents)-end_clip):
while base_map_info['readbase'][aligni]=='-': aligni += 1
if not base_map_info['readbase'][aligni] == modevents['model_state'][ie][2]:
print ('Error Does not match', base_map_info['readbase'][aligni], modevents['model_state'][ie][2], aligni, ie)
if mfpred_output[ie-start_clip]==1:
base_map_info['mod_pred'][aligni] = 1;
pred_mod_num += 1;
aligni += 1
return pred_mod_num
#
# get feature for a long read
#
def get_Feature(moptions, sp_options, sp_param, f5align, f5data, readk, start_clip, end_clip, base_map_info, forward_reverse, rname, mapped_start_pos, num_insertions, num_deletions):
modevents = sp_param['f5data'][readk][1]
# class num, bin num, and bin size
clnum = 2; binnum = 50; binlen = 0.2;
if forward_reverse=='+':
align_ref_pos = mapped_start_pos
else:
align_ref_pos = mapped_start_pos + len(base_map_info) - num_insertions - 1
# initialize feature matrix
if moptions['fnum']==57:
mfeatures = np.zeros((len(modevents)-end_clip+100-(start_clip-100), (binnum+3+3+4)));
else: mfeatures = np.zeros((len(modevents)-end_clip+100-(start_clip-100), (3+3+4)));
aligni = 0; isdif = False;
# get feature for each event; each event is a row
for ie in range(start_clip-100, len(modevents)-end_clip+100):
cur_row_num = ie - (start_clip-100); cur_base = ''
if ie>=start_clip and ie<len(modevents)-end_clip:
if align_ref_pos<mapped_start_pos:
print ('ERRRR align_ref_pos(%d)<mapped_start_pos(%d)' % (align_ref_pos, mapped_start_pos))
# get non-indel events
while base_map_info['readbase'][aligni]=='-':
if not align_ref_pos==base_map_info['refbasei'][aligni]:
print ('ERRRR align_ref_pos(%d) not equal to %d' % (align_ref_pos, base_map_info['refbasei'][aligni] ))
if not base_map_info['refbase'][aligni]=='-':
if forward_reverse=='+': align_ref_pos += 1
else: align_ref_pos -= 1
aligni += 1
if not base_map_info['readbase'][aligni] == modevents['model_state'][ie][2]:
print ('Error Does not match', base_map_info['readbase'][aligni], modevents['model_state'][ie][2], aligni, ie)
sp_param['f5status']= "Error Does not match"
if f5data[readk][3] not in sp_options["Error"]['Error Does not match']:
sp_options["Error"]['Error Does not match'].append(f5data[readk][3])
if aligni>50: break;
isdif = True;
mfeatures[cur_row_num][0] = align_ref_pos
cur_base = base_map_info['refbase'][aligni]
# both positive/negative labels is zero
mfeatures[cur_row_num][1] = 0; mfeatures[cur_row_num][2] = 0
if not base_map_info['refbase'][aligni]=='-':
if forward_reverse=='+': align_ref_pos += 1
else: align_ref_pos -= 1
aligni += 1
# for bin features
if ie>=0 and ie<len(modevents) and moptions['fnum']==57:
for currs in sp_param['f5data'][readk][2][modevents['start'][ie]:int(modevents['start'][ie]+int(modevents['length'][ie]+0.5))]:
if currs>10 or currs<-10: print ('Error raw signal', currs, ie, modevents['start'][ie], modevents['length'][ie])
curbin = int((currs+5)/binlen)
if curbin<0: curbin = 0
elif not curbin<binnum: curbin = binnum-1
mfeatures[cur_row_num][curbin+3] += 1
if ie>=0 and ie<len(modevents):
# for reference base type
if cur_base in myCom.g_ACGT:
mfeatures[cur_row_num][moptions['fnum']-3+3-4+myCom.g_ACGT.index(cur_base)] = 1
cur_index_add = moptions['fnum'] - 3 + 3
# for mean, std and length of signals
mfeatures[cur_row_num][cur_index_add + 0] = modevents["mean"][ie]
mfeatures[cur_row_num][cur_index_add + 1] = modevents["stdv"][ie]
mfeatures[cur_row_num][cur_index_add + 2] = modevents["length"][ie]
return (mfeatures, isdif)
#
# get complementary base of a given base
#
def get_complement(na):
if na in myCom.acgt: return myCom.na_bp[na]
else: return na;
#
# get mean/std of signals
#
def calculate_mean_std(m_event, event_ind, forward_reverse, raw_pv, moptions, sp_param):
if forward_reverse=='-':
pvsignal = raw_pv[m_event[-event_ind-1][2]:(m_event[-event_ind-1][2]+m_event[-event_ind-1][3])]
else:
pvsignal = raw_pv[m_event[event_ind][2]:(m_event[event_ind][2]+m_event[event_ind][3])]
# get mean/std for an event
c_mean = round(np.mean(pvsignal), 3)
c_std = round(np.std(pvsignal), 3)
return (c_mean, c_std)
#
# get required information from a mapped record
#
def handle_line(moptions, sp_param, f5align):
lsp = sp_param['line'].split('\t')
qname, flag, rname, pos, mapq, cigar, _, _, _, seq, _ = lsp[:11]
# check query name, map quality, reference position, cigar and reference name
if qname=='*': sp_param['f5status'] = "qname is *"
elif int(mapq)==255: sp_param['f5status'] = "mapq is 255"
elif int(pos)==0: sp_param['f5status'] = "pos is 0"
elif cigar=='*': sp_param['f5status'] = "cigar is *"
elif rname=='*': sp_param['f5status'] = "rname is *"
if not sp_param['f5status']=="": return qname
if (qname not in f5align) or f5align[qname][0]<int(mapq):
f5align[qname] = (int(mapq), int(flag), rname, int(pos), cigar, seq)
return qname
#
# the worker of the detection in a multiprocessing way
#
def detect_handler(moptions, h5files_Q, failed_Q, file_map_info_q):
_, init_l, _, _, _, X, Y, _, _, _, _, mfpred = myMultiBiRNN.mCreateSession(moptions['fnum'], moptions['hidden'], moptions['windowsize'], moptions)
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.Session(config=config)
# load module
new_saver = tf.train.import_meta_graph(moptions['modfile'][0]+'.meta')
new_saver.restore(sess,tf.train.latest_checkpoint(moptions['modfile'][1]))
while not h5files_Q.empty():
cur_start_time = time.time()
try:
# get fast5 file
f5files, ctfolderid, batchid = h5files_Q.get(block=False)
except:
break;
sp_options = defaultdict();
sp_options['ctfolderid'] = ctfolderid
sp_options['ctfolder'] = moptions['outFolder']+moptions['FileID']+'/'+str(ctfolderid)
if not os.path.isdir(sp_options['ctfolder']):
os.system('mkdir '+sp_options['ctfolder'])
#if moptions['testrnn']:
sp_options['rnn'] = (sess, X, Y, init_l, mfpred)
sp_options['batchid'] = batchid
sp_options['Mod'] = [];
# make modification prediction for each fast5
mDetect1(moptions, sp_options, f5files)
# outputing errors
for errtype, errfiles in sp_options["Error"].items():
failed_Q.put((errtype, errfiles));
print ("Cur Prediction consuming time %d for %d %d" % (time.time() - cur_start_time, ctfolderid, batchid))
sess.close()
#
# read index files for each output prediction file
#
def read_file_list(cur_cif, cur_chr, cur_strand, sp_options):
cur_list = []
with open(cur_cif, 'r') as mr:
line = mr.readline();
while line: # get where is prediction file from each line
line = line.strip();
if len(line)>0:
lsp = line.split();
if line[0]=='#':
if lsp[1][0] not in ['/', '\\']:
lsp[1] = lsp[1] + '/'
if lsp[0]=='#base_folder_fast5': sp_options['base_folder_fast5'] = lsp[1];
elif lsp[0]=='#base_folder_output': sp_options['base_folder_output'] = lsp[1];
else:
if lsp[1] == cur_strand:
cur_list.append(lsp)
if not lsp[0]==cur_chr:
print ('Warning!!! The chr should be %s but % is found.' % (cur_chr, lsp[0]))
line = mr.readline();
sp_options['handlingList'] = cur_list
#
# get prediction detail
#
def read_pred_detail(moptions, sp_options, f5info):
f5pred_file = sp_options['base_folder_output'] + '/' + f5info[5]
f5_pred_key = ('/pred/%s/predetail' % f5info[3])
# get prediction detail from saved prediction file
# each file contains predictions for multiple fast5
with h5py.File(f5pred_file, 'r') as mr:
m_pred = mr[f5_pred_key].value;
mapped_chrom = mr['/pred/%s' % f5info[3]].attrs['mapped_chr'] #.decode(encoding="utf-8")
mapped_strand= mr['/pred/%s' % f5info[3]].attrs['mapped_strand'] #.decode(encoding="utf-8")
m_pred = np.array(m_pred, dtype=[('refbase', 'U1'), ('readbase', 'U1'), ('refbasei', np.uint64), ('readbasei', np.uint64), ('mod_pred', np.int)])
return (m_pred, mapped_chrom, mapped_strand)
#
# summarize modification for each genome position of interest
#
def sum_handler(moptions, chr_strand_Q):
while not chr_strand_Q.empty():
try:
# get setting for summarization of predicted modifications
cur_cif, cur_chr, cur_strand = chr_strand_Q.get(block=False)
except:
break;
sp_options = {}
# get prediction files
read_file_list(cur_cif, cur_chr, cur_strand, sp_options)
sp_options['4NA'] = {moptions['Base']:defaultdict()}
sp_options['4NAfile'] = {}
for nak in sp_options['4NA']:
if not moptions['mod_cluster']:
sp_options['4NAfile'][nak] = ('%s/mod_pos.%s%s.%s.bed' % (moptions['outFolder'], cur_chr, cur_strand, nak))
else:
sp_options['4NAfile'][nak] = ('%s/cluster_mod_pos.%s%s.%s.bed' % (moptions['outFolder'], cur_chr, cur_strand, nak))
cur_start_time = time.time(); hlnum = 0;
for hl in sp_options['handlingList']:
# read prediction detail for each fast5
m_pred, mapped_chrom, mapped_strand = read_pred_detail(moptions, sp_options, hl)
if not (mapped_chrom==cur_chr and mapped_strand==cur_strand):
print("ERRoR not the same chr (real=%s vs expect=%s) and strand (real=%s VS expect=%s)" % (mapped_chrom, cur_chr, mapped_strand, cur_strand))
#####################################################
if moptions['mod_cluster']: # revised; should not used now
from numpy.lib.recfunctions import append_fields
m_pred = append_fields(m_pred, 'mod_pred2', m_pred['mod_pred']+0, usemask=False)
for mi in range(len(m_pred)):
if not m_pred['mod_pred2'][mi] == m_pred['mod_pred'][mi]:
print('mod_pred_Error %d %d %d' % (mi, m_pred['mod_pred2'][mi], m_pred['mod_pred'][mi]))
if m_pred['mod_pred2'][mi]==1 or m_pred['refbase'][mi] not in ['C']: continue;
m_3 = []
m_5 = []
mj = mi-1
while mj>-1 and len(m_3)<12:
if m_pred['refbase'][mj] in ['N', 'n']: break;
if m_pred['refbase'][mj] not in ['-']:
m_3.append((m_pred['refbase'][mj], m_pred['mod_pred2'][mj]))
mj -= 1;
if len(m_3)>0: m_3 = m_3[::-1]
mj = mi + 1;
while mj < len(m_pred) and len(m_5)<12:
if m_pred['refbase'][mj] in ['N', 'n']: break;
if m_pred['refbase'][mj] not in ['-']:
m_5.append((m_pred['refbase'][mj], m_pred['mod_pred2'][mj]))
mj += 1
cpgnum = 0; meth_cpgnum = 0;
m_3and5 = [m_3, m_5]
for m_53 in m_3and5:
for mj in range(len(m_53)-1):
if m_53[mj][0]=='C' and m_53[mj+1][0]=='G':
cpgnum +=1
if -0.1 < m_53[mj][1]-1 < 0.1:
meth_cpgnum += 1
if cpgnum>0 and (meth_cpgnum>0 and meth_cpgnum/float(cpgnum)>0.5):
m_pred['mod_pred'][mi] = 1
#####################################################################################
for mi in range(len(m_pred)):
# get prediction for each base type
if m_pred['refbase'][mi] not in sp_options['4NA']: continue;
if m_pred['refbase'][mi] in ['-', 'N', 'n']: continue;
if (cur_chr, cur_strand, m_pred['refbasei'][mi]) not in sp_options['4NA'][m_pred['refbase'][mi]]:
sp_options['4NA'][m_pred['refbase'][mi]][(cur_chr, cur_strand, int(m_pred['refbasei'][mi]) )] = [0, 0, m_pred['refbase'][mi]]
if not (m_pred['refbase'][mi] == sp_options['4NA'][m_pred['refbase'][mi]][(cur_chr, cur_strand, int(m_pred['refbasei'][mi]) )][2]):
print ('Error !!!! NA not equal %s == %s' % (m_pred['refbase'][mi], sp_options['4NA'][m_pred['refbase'][mi]][(cur_chr, cur_strand, int(m_pred['refbasei'][mi]) )][2]))
if not m_pred['readbase'][mi]=='-':
sp_options['4NA'][m_pred['refbase'][mi]][(cur_chr, cur_strand, int(m_pred['refbasei'][mi]) )][0] += 1
if -0.1 < m_pred['mod_pred'][mi]-1 < 0.1:
sp_options['4NA'][m_pred['refbase'][mi]][(cur_chr, cur_strand, int(m_pred['refbasei'][mi]) )][1] += 1
hlnum += 1
if hlnum % 1000==0:
print ("\tCurrent time consuming %d for %d" % (time.time() - cur_start_time, hlnum))
cur_start_time = time.time()
print ('====sum done! To save')
for nak in sp_options['4NA']:
print ('\tSave %s' % sp_options['4NAfile'][nak])
if len(sp_options['4NA'][nak])>0:
with open(sp_options['4NAfile'][nak], 'w') as mw:
# save the summary for each base type
pos_keys = sp_options['4NA'][nak].keys();
pos_keys = sorted(pos_keys);
for pk in pos_keys:
neighna = [sp_options['4NA'][nak][pk][2]]
mw.write(' '.join([ pk[0], str(pk[2]), str(pk[2]+1), ''.join(neighna), \
str(1000 if sp_options['4NA'][nak][pk][0]>1000 else sp_options['4NA'][nak][pk][0]), \
pk[1], str(pk[2]), str(pk[2]+1), '0,0,0', str(sp_options['4NA'][nak][pk][0]), \
('%d' % (100*sp_options['4NA'][nak][pk][1]/(sp_options['4NA'][nak][pk][0] if sp_options['4NA'][nak][pk][0]>0 else 1))), \
str(sp_options['4NA'][nak][pk][1]), '\n' ]))
#
# prediction manager of a multiprocess process
#
def mDetect_manager(moptions):
pmanager = multiprocessing.Manager();
# get input folder
while (not moptions['wrkBase']==None) and len(moptions['wrkBase'])>0 and moptions['wrkBase'][-1] in ['/', '\\']:
moptions['wrkBase'] = moptions['wrkBase'][:-1]
# need to make prediction of modification
if moptions['predDet']==1:
# get well-trained model
if moptions['modfile'].rfind('/')==-1:
moptions['modfile'] = [moptions['modfile'], './']
else:
moptions['modfile'] = [moptions['modfile'], moptions['modfile'][:moptions['modfile'].rfind('/')+1]]
start_time = time.time();
# get fast5 files in a recurisive way
f5files = glob.glob(os.path.join(moptions['wrkBase'],"*.fast5" ))
if moptions['recursive']==1:
f5files.extend(glob.glob(os.path.join(moptions['wrkBase'],"*/*.fast5" )))
f5files.extend(glob.glob(os.path.join(moptions['wrkBase'],"*/*/*.fast5" )))
f5files.extend(glob.glob(os.path.join(moptions['wrkBase'],"*/*/*/*.fast5" )))
print('Total files=%d' % len(f5files))
# output folder
if not os.path.isdir(moptions['outFolder']+moptions['FileID']):
os.system('mkdir -p '+moptions['outFolder']+moptions['FileID'])
# prepare multiprocessing
h5files_Q = pmanager.Queue();
file_map_info_q = pmanager.Queue();
failed_Q = pmanager.Queue()
# spliting fast5 files into different lists
h5_batch = []; h5batchind = 0;
sub_folder_size = 100; sub_folder_id = 0;
for f5f in f5files:
h5_batch.append(f5f);
if len(h5_batch)==moptions['files_per_thread']:
# each batch
h5files_Q.put((h5_batch, sub_folder_id, h5batchind))
h5_batch = []; h5batchind += 1
if h5batchind % sub_folder_size ==0:
sub_folder_id += 1
if len(h5_batch)>0:
h5files_Q.put((h5_batch, sub_folder_id, h5batchind))
h5_batch = []; h5batchind += 1
# start multiprocessing
share_var = (moptions, h5files_Q, failed_Q, file_map_info_q)
handlers = []
for hid in range(moptions['threads']):
p = multiprocessing.Process(target=detect_handler, args=share_var);
p.start();
handlers.append(p);
# check potential errors
failed_files = defaultdict(list);
while any(p.is_alive() for p in handlers):
try:
errk, fns = failed_Q.get(block=False);
failed_files[errk].extend(fns)
except:
time.sleep(1);
continue;
# prepare modificatoin summary for reference positions of interest
moptions['predpath'] = moptions['outFolder'] + '/'+moptions['FileID']
pred_ind_pref = moptions['outFolder'] + '/'+moptions['FileID']+'/'+pre_base_str
pred_chr_files = glob.glob(os.path.join(moptions['outFolder']+moptions['FileID'], '*/*.'+pre_base_str+'.*'))
chr_dict = defaultdict(list);
for pcf in pred_chr_files:
chr_dict[ pcf.split('/')[-1].split('.'+pre_base_str)[0] ].append(pcf)
chr_keys = chr_dict.keys();
for ck in chr_keys:
cur_ind_pred_f = pred_ind_pref + '.' + ck;
cur_list = [ ['#base_folder_fast5', moptions['wrkBase']], ['#base_folder_output', os.path.abspath(moptions['outFolder']+moptions['FileID'])] ]
for sub_c_f in chr_dict[ck]:
with open(sub_c_f, 'r') as mr:
line = mr.readline()
while line:
line = line.strip()
if len(line)>0:
lsp = line.split();
lsp[2] = int(lsp[2])
cur_list.append(lsp)
line = mr.readline()
cur_list = sorted(cur_list)
with open(cur_ind_pred_f, 'w') as indf_writer:
for mfi in cur_list:
cur_m_f = []
for mfidetail in mfi:
cur_m_f.append(str(mfidetail))
cur_m_f.append('\n')
indf_writer.write(' '.join(cur_m_f))
# error info
if len(failed_files)>0:
print ('Error information for different fast5 files:')
for errtype, errfiles in failed_files.items():
print ('\t'+errtype, len(errfiles))
moptions['outFolder'] = moptions['outFolder']+moptions['FileID']
end_time = time.time();
print ("Per-read Prediction consuming time %d" % (end_time-start_time))
### for summarizing modificatoin prediction
start_time = time.time();
# get all index files of prediction
all_chr_ind_files = glob.glob(os.path.join(moptions['predpath'], pre_base_str+'.*'))
print('Find: %s %d %s' % (moptions['predpath'], len(all_chr_ind_files), pre_base_str))
print (all_chr_ind_files)
# for each chromosome, a thread will be initialized for multiprocessing summarization of modifications
chr_strand_Q = pmanager.Queue(); jobnum = 0;
for cur_cif in all_chr_ind_files:
chr_strand_Q.put((cur_cif, cur_cif.split(pre_base_str)[-1][1:], '+'))
chr_strand_Q.put((cur_cif, cur_cif.split(pre_base_str)[-1][1:], '-'))
jobnum +=2
# star to summarize modificaiton prediction of reference genomes of interest
share_var = (moptions, chr_strand_Q)
handlers = []
for hid in range(moptions['threads'] if moptions['threads']<jobnum else jobnum):
p = multiprocessing.Process(target=sum_handler, args=share_var);
p.start();
handlers.append(p);
while any(p.is_alive() for p in handlers):
try:
time.sleep(1);
except:
time.sleep(1);
continue;
end_time = time.time();
print ("Genomic-position Detection consuming time %d" % (end_time-start_time))
os.system('touch '+moptions['outFolder']+'.done')
# for independent testing of code
if __name__=='__main__':
# if len(sys.argv)>4:
moptions = {}
moptions['basecall_1d'] = 'Basecall_1D_000'
moptions['basecall_1d'] = ['Basecall_1D_000']
moptions['basecall_2strand'] = 'BaseCalled_template'
moptions['outLevel'] = myCom.OUTPUT_WARNING
moptions['outLevel'] = myCom.OUTPUT_INFO
moptions['modfile'] = '../../mod_output/train1/2/mod_train'
moptions['fnum'] = 3;
moptions['hidden'] = 100;
moptions['windowsize'] = 21;
moptions['threads'] = 8
moptions['threads'] = 1
moptions['files_per_thread'] = 500
mDetect_manager(moptions)