-
Notifications
You must be signed in to change notification settings - Fork 4
/
pharmerdb.cpp
1171 lines (1025 loc) · 30.4 KB
/
pharmerdb.cpp
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
/*
Pharmer: Efficient and Exact 3D Pharmacophore Search
Copyright (C) 2011 David Ryan Koes and the University of Pittsburgh
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*/
/*
* pharmerdb.cpp
*
* Created on: Aug 2, 2010
* Author: dkoes
*/
#include "tripletmatching.h"
#include <boost/lexical_cast.hpp>
#include <fcntl.h>
#include <openbabel/mol.h>
#include "CommandLine2/CommandLine.h"
#include "Triplet.h"
#include "BitSetTree.h"
#include "ReadMCMol.h"
#include "Timer.h"
#include "pharmerdb.h"
#include "TripleIndexer.h"
#include "PMol.h"
#ifdef OUTPUTSMINA
#include "SminaConverter.h"
#endif
using namespace boost;
using namespace std;
using namespace OpenBabel;
extern cl::opt<bool> Quiet;
extern cl::opt<unsigned> ReduceConfs;
extern cl::opt<bool> ComputeThresholds;
//location comparison functions for pointdata
bool comparePointDataX(const ThreePointData& lhs, const ThreePointData& rhs)
{
return lhs.l1 < rhs.l1;
}
//location comparison functions for pointdata
bool comparePointDataY(const ThreePointData& lhs, const ThreePointData& rhs)
{
return lhs.l2 < rhs.l2;
}
//location comparison functions for pointdata
bool comparePointDataZ(const ThreePointData& lhs, const ThreePointData& rhs)
{
return lhs.l3 < rhs.l3;
}
//write out to buf, return number of bytes written
unsigned MolDataCreator::ConfCreator::write(unsigned char *buf)
{
unsigned pos = 0;
memcpy(buf, &header, sizeof(header));
pos += sizeof(header);
memcpy(buf + pos, molData.data(), molData.size());
pos += molData.size();
return pos;
}
//initialize data for a conformation
MolDataCreator::ConfCreator::ConfCreator(unsigned mid, float mw, unsigned cid,
OBMol& m) :
header(mid, mw, cid)
{
PMolCreator mol(m, true);
stringstream mdata;
if(mol.writeBinary(mdata))
molData = mdata.str();
else
molData = "";
}
unsigned MolDataCreator::maxIndex = 0;
const char *rotBondsSMART =
"[!$([NH]!@C(=O))&!D1&!$(*#*)]-&!@[!$([NH]!@C(=O))&!D1&!$(*#*)]";
//generate pharma points and database infor for mol
void MolDataCreator::processMol(OBMol& mol, double mWeight, unsigned mid)
{
//identify pharma
vector<vector<PharmaPoint> > mcpoints;
getPharmaPointsMC(pharmas, mol, mcpoints);
double weight = mol.GetMolWt();
//count number of rotatable bounds
OBSmartsPattern rot;
rot.Init(rotBondsSMART);
vector<vector<int> > maplist;
rot.Match(mol, maplist, OBSmartsPattern::AllUnique);
unsigned numrot = maplist.size();
//store each conformation seperately, this results in about 50% more space used,
//but as much as 50% performance improvement when retrieving mols
//generate information for each conformation
for (unsigned c = 0, nc = mol.NumConformers(); c < nc; c++)
{
unsigned confidx = c;
mol.SetConformer(confidx);
confs.push_back(ConfCreator(mid, mWeight, confidx, mol));
ConfCreator& conf = confs.back();
if(!conf.isValid())
{
cerr << mol.GetTitle() << " is too large. Omitting.\n";
confs.pop_back();
continue;
}
//setup pharma points
const vector<PharmaPoint>& points = mcpoints[confidx];
if (maxIndex < points.size())
maxIndex = points.size();
//compute and canonicalize all triple points
unsigned n = points.size();
if (n > (1 << TPD_INDEX_BITS))
{
cerr << "Too many pharma points (" << n << ") in "
<< mol.GetTitle() << "\n";
n = (1 << TPD_INDEX_BITS);
}
for (unsigned i = 0; i < n; i++)
{
for (unsigned j = i + 1; j < n; j++)
{
for (unsigned k = j + 1; k < n; k++)
{
unsigned pos = tindex(points[i].pharma->index,
points[j].pharma->index, points[k].pharma->index);
assert(pos < pdatas.size());
ThreePointData trip(bufferSize, weight, numrot, points, i, j, k);
pdatas[pos].push_back(trip);
}
}
}
confOffsets.push_back(bufferSize);
bufferSize += conf.byteSize();
numConfs++;
}
}
//flatten all the conf info into a buffer
void MolDataCreator::createBuffer()
{
buffer = new unsigned char[bufferSize];
unsigned pos = 0;
for (unsigned i = 0, n = confs.size(); i < n; i++)
{
pos += confs[i].write(&buffer[pos]);
}
assert(pos == bufferSize);
//clear mem
confs.clear();
}
//write out the data with proper offsets to files (assumed positioned properly and single threaded)
unsigned MolDataCreator::write(FILE *molData, vector<PointDataFile >& pointDataFiles)
{
unsigned long location = ftell(molData);
location = ftell(molData);
unsigned mid = location >> (TPD_MOLDATA_BITS - TPD_MOLID_BITS);
//update the point datas
for (unsigned p = 0, np = pdatas.size(); p < np; p++)
{
for (unsigned i = 0, n = pdatas[p].size(); i < n; i++)
{
pdatas[p][i].molPos += location;
}
}
//write out
fwrite(buffer, bufferSize, 1, molData);
for (unsigned p = 0, np = pdatas.size(); p < np; p++)
{
pointDataFiles[p].write(&pdatas[p][0], sizeof(pdatas[p][0]),
pdatas[p].size());
}
//align the start of every mol so the upper MOLID_BITS correspond to an mol id
//(assuming the conformers fit within MOLDATA-MOLID bits)
//we assume (hope) the filesystem has support for holes
location = ftell(molData);
location >>= (TPD_MOLDATA_BITS - TPD_MOLID_BITS);
location++;
location <<= (TPD_MOLDATA_BITS - TPD_MOLID_BITS);
fseek(molData, location, SEEK_SET);
if (ftell(molData) >= (1L << TPD_MOLDATA_BITS))
{
cerr
<< "Input database is too large (moldata more than 1TB). Split and retry.\n";
abort();
}
return mid;
}
void MolData::clear()
{
mol = NULL;
confidx = 0;
mid = 0;
molWeight = 0;
}
bool MolData::read(FILE *molData, PMolReader& areader)
{
clear();
MolDataHeader header;
//get fixed sized header
int ret = fread(&header, sizeof(header), 1, molData);
assert(ret == 1);
//copy to self
confidx = header.confidx;
molWeight = header.molWeight;
mid = header.molID;
//now get mol itself
mol = areader.readPMol(molData);
return mol != NULL;
}
//read in mol data at position location from file molData
bool MolData::read(FILE *molData, unsigned long location, PMolReader& areader)
{
fseek(molData, location, SEEK_SET);
return read(molData, areader);
}
//read in mol data at position location in array molData
bool MolData::read(unsigned char *molData, unsigned long location,
PMolReader& areader)
{
clear();
MolDataHeader header;
//get fixed sized header
unsigned long loc = location;
memcpy(&header, &molData[loc], sizeof(header));
loc += sizeof(header);
//copy to self
confidx = header.confidx;
molWeight = header.molWeight;
mid = header.molID;
//now get mol itself
mol = areader.readPMol(&molData[loc]);
return mol != NULL;
}
//read just metadata, don't parse full mol
void MolData::readDataOnly(unsigned char *molData, unsigned long location)
{
clear();
MolDataHeader header;
//get fixed sized header
unsigned long loc = location;
memcpy(&header, &molData[loc], sizeof(header));
loc += sizeof(header);
//copy to self
confidx = header.confidx;
molWeight = header.molWeight;
mid = header.molID;
}
void PharmerDatabaseCreator::initializeDatabases()
{
//initialize self-managed flat file databases
//open for writing
//info
filesystem::path ipath = dbpath;
ipath /= "info";
assert(!filesystem::exists(ipath));
info = fopen(ipath.string().c_str(), "w+"); //create
assert(info);
//human readable pharmacophore classes
filesystem::path lpath = dbpath;
lpath /= "lookup";
lookup = fopen(lpath.string().c_str(), "w"); //create
assert(lookup);
//mid index (to better support lots of conformers)
filesystem::path mpath = dbpath;
mpath /= "mids";
midList = fopen(mpath.string().c_str(), "w"); //create
assert(midList);
//moldata
filesystem::path mdpath = dbpath;
mdpath /= "molData";
molData = fopen(mdpath.string().c_str(), "w+");
assert(molData);
//smina data
#ifdef OUTPUTSMINA
filesystem::path smipath = dbpath / "sminaIndex";
sminaIndex = fopen(smipath.string().c_str(), "w+");
assert(sminaIndex);
filesystem::path smdpath = dbpath / "sminaData";
sminaData = fopen(smdpath.string().c_str(), "w+");
assert(sminaData);
#endif
//bincnts
filesystem::path binpath = dbpath;
binpath /= "binCnts";
binData = fopen(binpath.string().c_str(), "w+");
assert(binData);
//output pharmas
filesystem::path phpath = dbpath / "pharmas";
ofstream pharmout(phpath.string().c_str());
pharmas.write(pharmout);
//pointData
pointDataFiles.resize(tindex.size());
for (int i = 0, n = tindex.size(); i < n; i++)
{
string pname = string("pointData_") + lexical_cast<string> (i);
filesystem::path pdpath = dbpath / pname;
pointDataFiles[i] = PointDataFile(pdpath.string(), i);
}
//geoData
geoDataFiles.resize(tindex.size(), NULL);
}
//write out index into "correct" mid
void PharmerDatabaseCreator::writeMIDs()
{
if(midList)
{
fseek(midList, 0, SEEK_SET);
unsigned cnt = 0;
for(unsigned i = 0, n = mids.size(); i < n; i++)
{
fwrite(&mids[i],sizeof(unsigned),1,midList);
cnt++;
while(i+1 < n && cnt < mids[i+1])
{
fwrite(&mids[i],sizeof(unsigned),1,midList);
cnt++;
}
}
}
}
void PharmerDatabaseCreator::writeStats()
{
if (info)
{
fseek(info, 0, SEEK_SET);
fwrite(&stats, sizeof(stats), 1, info);
fflush(info);
}
if (lookup)
{
tindex.dump(lookup);
}
writeMIDs();
}
//add a multiconformer mol to the database; weighted is passed in to assure
//weight used for segregration matches reported weight
void PharmerDatabaseCreator::addMolToDatabase(OBMol& mol, double weight)
{
static OBAromaticTyper aromatics;
static OBAtomTyper atyper;
if(mol.NumAtoms() == 0) //skipped
return;
mol.Center();
//ARG!!! these routines cruelly remove conformer data so we need
//a separate OBMol for each conformer so we can add hydrogens
// I wish this were more efficient and less fragile..
unsigned nc = mol.NumConformers();
if (ReduceConfs > 0 && ReduceConfs < nc)
nc = ReduceConfs;
vector<OBMol> confs(nc);
for (unsigned c = 0; c < nc; c++)
{
confs[c] = mol;
confs[c].SetConformer(c);
confs[c].AddHydrogens();
}
mol.AddHydrogens();
//all the implicitly called perception routines are absolutely not
//thread safe, so do all perception here (hopefully haven't missed anything)
mol.FindRingAtomsAndBonds();
mol.FindChiralCenters();
mol.PerceiveBondOrders();
aromatics.AssignAromaticFlags(mol);
mol.FindSSSR();
atyper.AssignTypes(mol);
atyper.AssignHyb(mol);
//now add back conformer coordinates, create new memory for mol
vector<double*> cdata(confs.size());
unsigned ncoords = mol.NumAtoms() * 3;
for (unsigned i = 0, n = confs.size(); i < n; i++)
{
assert(confs[i].NumAtoms()*3 == ncoords);
cdata[i] = new double[ncoords];
memcpy(cdata[i], confs[i].GetConformer(0), ncoords*sizeof(double));
}
mol.SetConformers(cdata);
//generate moldata
MolDataCreator mdc(pharmas, tindex, mol, weight, stats[NumMols]);
unsigned mid = mdc.write(molData, pointDataFiles);
mids.push_back(mid);
//output smina data here
#ifdef OUTPUTSMINA
SminaConverter::MCMolConverter mcsmina(mol);
const vector<unsigned>& confOffsets = mdc.ConfOffsets();
unsigned long mloc = mid;
mloc <<= (TPD_MOLDATA_BITS - TPD_MOLID_BITS);
for(unsigned i = 0, n = confOffsets.size(); i < n; i++)
{
stringstream data;
mcsmina.convertConformer(i, data);
unsigned sz = data.str().size();
unsigned long pos = mloc + confOffsets[i]; //location in molData
unsigned long smpos = ftell(sminaData); //location in smina dta
//output index mapping
fwrite(&pos,sizeof(pos),1,sminaIndex);
fwrite(&smpos,sizeof(smpos),1,sminaIndex);
//output actual data
fwrite(&sz, sizeof(sz),1, sminaData);
fwrite(data.str().c_str(), sizeof(char), sz, sminaData);
}
#endif
stats[NumMols]++;
stats[NumConfs] += mdc.NumConfs();
stats[NumDbPoints] += mdc.NumPoints();
}
/*
* Add the molecules in infile from start offset by stride to the database.
* Sequential molecules with the same name must be conformers of the same mol.
*
* This is single threaded since bottlenecks in openbabel eliminate any advantages to multithreading
*/
void PharmerDatabaseCreator::addMolsToDatabase(istream& infile,
OBFormat *format, unsigned start, unsigned stride,
unsigned long readBytes, unsigned long lastByte)
{
//add all the molecules
Timer t;
OBMol mol;
ReadMCMol reader(infile, format, stride, start, ReduceConfs);
while (reader.read(mol))
{
addMolToDatabase(mol, mol.GetMolWt());
}
if(ComputeThresholds)
thresholdComputer.printBestThresholds();
writeStats();
if (!Quiet)
{
cout << "\nAdded in " << t.elapsed() << "s (" << t.elapsedProcess()
<< "s)\n";
cout << "\t" << stats[NumMols] << "\tmolecules\t";
unsigned long location = ftell(molData);
location >>= (TPD_MOLDATA_BITS - TPD_MOLID_BITS);
cout << "(" << location << " logical)\n";
cout << "\t" << stats[NumConfs] << "\tconformations\n";
cout << "\t" << stats[NumDbPoints] << "\tpoints\n";
cout << "\t" << MolDataCreator::MaxIndex() << "\tmaxIndex\n";
}
}
//coalesce very small collections and
//mmap pointData files, closing file pointer
void PharmerDatabaseCreator::initPointDataArrays()
{
assert(pointDataArrays == NULL);
unsigned n = pointDataFiles.size();
pointDataArrays = new MMappedRegion<ThreePointData> [n];
for (unsigned i = 0; i < n; i++)
{
if(pointDataFiles[i].isValid())
{
pointDataFiles[i].close();
pointDataArrays[i].map(pointDataFiles[i].name, false, true);
}
}
}
//split along the the longest access
//fill in split info
static void chooseSplit(const BoundingBox& box, SplitInfo& info,
unsigned long long numPoints, unsigned long long numUnique,
unsigned pos)
{
if (numPoints == 0)
{
info.type = NoSplit;
}
else if ((numUnique <= MAX_UNIQUE_POINTS_PER_LEAF || numPoints
< MAX_POINTDATAS_PER_LEAF))
{
//not worth further dividing
info.type = NoSplit;
}
else if (2 * pos >= SPLITS_PER_GEOPAGE && numPoints
< MIN_POINTDATAS_PER_PAGE)
{
info.type = NoSplit;
//don't create a whole new geo page
}
else //do a split
{
int xdiff = box.maxx - box.minx;
int ydiff = box.maxy - box.miny;
int zdiff = box.maxz - box.minz;
if (xdiff >= ydiff && xdiff >= zdiff)
{
info.type = SplitXFixed;
info.min = box.minx;
info.max = box.maxx;
info.func = comparePointDataX;
}
else if (ydiff >= xdiff && ydiff >= zdiff)
{
info.type = SplitYFixed;
info.min = box.miny;
info.max = box.maxy;
info.func = comparePointDataY;
}
else //z
{
info.type = SplitZFixed;
info.min = box.minz;
info.max = box.maxz;
info.func = comparePointDataZ;
}
}
}
class PivotCompareRnd
{
short pivot;
const SplitInfo& info;
public:
PivotCompareRnd(short p, const SplitInfo& i) :
pivot(p), info(i)
{
}
//flip a coint as to which way equal vals go
bool operator()(const ThreePointData& x)
{
int diff = info.splitVal(x) - pivot;
if (diff < 0)
return true;
if (diff > 0)
return false;
return ::random() % 2;
}
};
//find the median value in one sequential scan using counts
static unsigned short findMedianValue(ThreePointData *start, ThreePointData *end,
const SplitInfo& info, unsigned short min, unsigned short max)
{
unsigned long total = 0;
int left = min;
int right = left + (max - min);
unsigned long counts[right - left + 1];
memset(counts, 0, sizeof(counts));
for (ThreePointData *itr = start; itr != end; itr++)
{
counts[info.splitVal(*itr) - left]++;
total++;
}
unsigned long mid = total / 2;
unsigned long cnt = 0;
for (int i = left; i <= right; i++)
{
cnt += counts[i - left];
if (cnt >= mid)
return i;
}
abort();
}
//create an internal kd tree
//each node splits points data in two along the axis with the largest spread
//every node has an explicit bounding box (so in that sense this is more like an R-tree)
void PharmerDatabaseCreator::doSplitInPage(unsigned pharma, FILE *geoFile, GeoKDPage& page,
unsigned pos, ThreePointData *start, ThreePointData *end, ThreePointData *begin,
unsigned depth)
{
//first compute the bounding box and count points
unsigned long numPoints = end - start;
unsigned long numUnique = 0;
TripleInt last(0, 0, 0);
BoundingBox box;
for (ThreePointData *itr = start; itr != end; itr++)
{
TripleInt coords(itr->l1, itr->l2, itr->l3);
box.update(coords);
if (coords != last)
{
last = coords;
numUnique++;
}
}
//choose splitting dimension
SplitInfo info;
chooseSplit(box, info, numPoints, numUnique, pos);
page.nodes[pos].box = box;
page.nodes[pos].splitType = info.type;
if (info.type == NoSplit)
{
//this is the end
}
else //need to split
{
//choose a median value, points with this value can be on either side
unsigned short medianVal = findMedianValue(start, end, info, info.getMin(box),
info.getMax(box));
PivotCompareRnd rcmp(medianVal, info);
ThreePointData *median = partition(start, end, rcmp);
if (median == end || median == start) //we got unlikely and did not partition well
{
sort(start, end, info.func);
median = start + (end - start) / 2;
}
page.nodes[pos].splitVal = medianVal;
page.nodes[pos].splitData = median - begin;
if (2 * pos + 1 < SPLITS_PER_GEOPAGE) //stay on internal page
{
if ((end - start) < pdatasFitInMemory && tcnt.threadAvailable())
{
thread t(thread_start_doSplitInPage, this, pharma, geoFile, boost::ref(
page), 2 * pos, start, median, begin, depth);
doSplitInPage(pharma, geoFile, page, 2 * pos + 1, median, end, begin, depth);
t.join();
tcnt.reduceThreadCount();
}
else
{
doSplitInPage(pharma, geoFile, page, 2 * pos, start, median, begin, depth);
doSplitInPage(pharma, geoFile, page, 2 * pos + 1, median, end, begin, depth);
}
}
else //children pushed to new page
{
assert(2*pos >= SPLITS_PER_GEOPAGE);
unsigned lpos = 2 * pos - SPLITS_PER_GEOPAGE;
unsigned rpos = lpos + 1;
assert(rpos < SPLITS_PER_GEOPAGE);
page.nextPages[lpos] = doSplitNewPage(pharma, geoFile, start, median, begin, depth);
page.nextPages[rpos] = doSplitNewPage(pharma, geoFile, median, end, begin, depth);
}
}
}
void PharmerDatabaseCreator::thread_start_doSplitInPage(
PharmerDatabaseCreator *pharmdb, unsigned pharma, FILE *geoFile, GeoKDPage& page,
unsigned pos, ThreePointData *start, ThreePointData *end, ThreePointData *begin,
unsigned depth)
{
pharmdb->doSplitInPage(pharma, geoFile, page, pos, start, end, begin, depth);
}
//create a geokdpage that indexes the data between start and end in geoFile
//writes out page to geoFile and returns its location as an index into an array
//of geokdpages
unsigned long PharmerDatabaseCreator::doSplitNewPage(unsigned pharma, FILE *geoFile,
ThreePointData *start, ThreePointData *end, ThreePointData *begin, unsigned depth)
{
GeoKDPage page;
//reserve space
unique_lock<shared_mutex> lock(fileAccessLock);
fseek(geoFile, 0, SEEK_END);
unsigned long location = ftell(geoFile);
assert(location % sizeof(GeoKDPage) == 0);
int ret = fwrite(&page, sizeof(page), 1, geoFile);
assert(ret == 1);
lock.unlock();
depth++;
//create the page
doSplitInPage(pharma, geoFile, page, 1, start, end, begin, depth);
stats[NumInternalPages]++;
//write it out
lock.lock();
fseek(geoFile, location, SEEK_SET);
location /= sizeof(GeoKDPage); //indexify
ret = fwrite(&page, sizeof(page), 1, geoFile);
assert(ret == 1);
return location;
}
//saturated binning
static unsigned lengthbin(unsigned len)
{
unsigned ret = len / LENGTHDIV;
if(ret >= LENGTH_BINS)
ret = LENGTH_BINS-1;
return ret;
}
void PharmerDatabaseCreator::incrementBinCnt(const ThreePointData& t, unsigned pclass)
{
binnedCnts[pclass][lengthbin(t.l1)][lengthbin(t.l2)][lengthbin(t.l3)]++;
}
//create spatial index for all ijk triples
void PharmerDatabaseCreator::createIJKSpatialIndex(int p)
{
//create spatial
ThreePointData *start = pointDataArrays[p].begin();
ThreePointData *begin = start;
ThreePointData *end = pointDataArrays[p].end();
if(start == end)
return;
//count binned frequencies
for (ThreePointData *itr = start; itr != end; itr++)
{
incrementBinCnt(*itr, p);
}
//open
string gname = string("geoData_") + lexical_cast<string> (p);
filesystem::path gpath = dbpath / gname;
FILE * geoFile= fopen(gpath.string().c_str(), "w");
geoDataFiles[p] = geoFile;
assert(geoFile);
//write out blank page, there's nothing at index 0
fseek(geoFile, 0, SEEK_SET);
GeoKDPage blank;
fwrite(&blank, sizeof(blank), 1, geoFile);
//top-down recursively create kd/r tree spatial data structure
doSplitNewPage(p, geoFile, start, end, begin, 0);
}
/* Create spatial index. */
void PharmerDatabaseCreator::createSpatialIndex()
{
Timer t;
cout << "Creating spatial index...\n";
//close and then mmap pointinfo file
initPointDataArrays();
for (unsigned p = 0, n = tindex.size(); p < n; p++)
createIJKSpatialIndex(p);
for(unsigned i = 0, n = binnedCnts.size(); i < n; i ++)
fwrite(binnedCnts[i].c_array(), LENGTH_BINS*LENGTH_BINS*LENGTH_BINS, sizeof(unsigned), binData);
cout << stats[NumConfs] << "\tconformations\n";
cout << stats[NumMols] << "\tmolecules\n";
cout << stats[NumInternalPages] << "\tinternal pages\n";
cout << "Index creation time: " << t.elapsed() << "s ("
<< t.elapsedProcess() << "s)" << "\n";
writeStats();
}
//Searcher
void PharmerDatabaseSearcher::initializeDatabases()
{
//initialize self-managed flat file databases
//open for reading
//info - and read it in
filesystem::path ipath = dbpath;
ipath /= "info";
info = fopen(ipath.string().c_str(), "r");
assert(info);
int read = fread(&stats, sizeof(stats), 1, info);
if (read != 1)
{
cerr << ipath.string() << "\n";
abort();
}
//pharmas
filesystem::path pharmpath = dbpath / "pharmas";
ifstream pharmain(pharmpath.string().c_str());
if (!pharmas.read(pharmain))
{
cerr << "Invalid pharma specification in database.\n";
return;
}
tindex.set(pharmas.size());
//moldata
filesystem::path mdpath = dbpath;
mdpath /= "molData";
molData.map(mdpath.string(), true, true);
//smina index and data - do not need to exist
filesystem::path smIndex = dbpath / "sminaIndex";
if(filesystem::exists(smIndex))
{
sminaIndex.map(smIndex.string(), true, true);
filesystem::path smData = dbpath / "sminaData";
sminaData.map(smData.string(), true, true);
}
//mids
filesystem::path mpath = dbpath;
mpath /= "mids";
if(filesystem::exists(mpath)) //back-wards compat
midList.map(mpath.string(), true, true,true);
//length histogram
filesystem::path binpath = dbpath / "binCnts";
binnedCnts.map(binpath.string(),true,false);
if(binnedCnts.length() == 0)
{
cerr << "Missing binnedCnts " << binpath << "\n";
return;
}
//pointData
unsigned n = tindex.size();
tripletDataArrays = new MMappedRegion<ThreePointData> [n];
for (unsigned i = 0; i < n; i++)
{
string pname = string("pointData_") + lexical_cast<string> (i);
filesystem::path pdpath = dbpath / pname;
if(filesystem::exists(pdpath))
tripletDataArrays[i].map(pdpath.string(), true, true);
}
//geoData
geoDataArrays = new MMappedRegion<GeoKDPage> [n];
for (unsigned i = 0; i < n; i++)
{
string gname = string("geoData_") + lexical_cast<string> (i);
filesystem::path gpath = dbpath / gname;
if(filesystem::exists(gpath))
geoDataArrays[i].map(gpath.string(), true, true);
}
valid = true;
}
unsigned PharmerDatabaseSearcher::getBinCnt(unsigned pclass, unsigned i, unsigned j, unsigned k)
{
unsigned index = pclass;
index *= LENGTH_BINS;
index += i;
index *= LENGTH_BINS;
index += j;
index *= LENGTH_BINS;
index += k;
return binnedCnts[index];
}
//generate ranking and return index of best triplet
//use binnedCnts histogram
unsigned PharmerDatabaseSearcher::rankTriplets(const
vector<QueryTriplet>& triplets, vector<double>& ranking)
{
double minval = HUGE_VAL;
unsigned besttrip = 0;
ranking.resize(triplets.size(), 0);
//add up values from histogram
for (unsigned i = 0, n = triplets.size(); i < n; i++)
{
double val = 0;
const QueryTriplet& trip = triplets[i];
unsigned pclass = tindex(trip.getPharma(0), trip.getPharma(1),
trip.getPharma(2));
for (unsigned x = lengthbin(trip.getRanges()[0].min), mx =
lengthbin(trip.getRanges()[0].max); x <= mx; x++)
for (unsigned y = lengthbin(trip.getRanges()[1].min), my =
lengthbin(trip.getRanges()[1].max); y <= my; y++)
for (unsigned z = lengthbin(trip.getRanges()[2].min),
mz = lengthbin(trip.getRanges()[2].max); z
<= mz; z++)
{
val += getBinCnt(pclass,x,y,z);
}
ranking[i] = val * triplets[i].numOrderings(); //symmetries will increases matches
if (val < minval)
{
minval = val;
besttrip = i;
}
}
return besttrip;
}
//get values along appropriate axis for split
static void getMinMax(const Triplet& t, SplitType split, int& min, int& max)
{
min = max = 0;
switch (split)
{
case SplitXFixed:
min = t.getRanges()[0].min;
max = t.getRanges()[0].max;
break;
case SplitYFixed:
min = t.getRanges()[1].min;
max = t.getRanges()[1].max;
break;
case SplitZFixed:
min = t.getRanges()[2].min;
max = t.getRanges()[2].max;
break;
default:
;
}
}
inline void dump(ostream& out, const ThreePointData& tpd)
{
out << tpd.molPos << ": " << tpd.l1 <<" " << tpd.l2 << " " << tpd.l3 << "\n";
}
void PharmerDatabaseSearcher::queryProcessPoints(QueryInfo& t,
unsigned long startLoc, unsigned long endLoc)
{
const ThreePointData *start = t.data + startLoc;
const ThreePointData *end = t.data + endLoc;
//put the range of point datas into the priorty queue,
//let the PQ filter out bad values
unsigned cnt = 0;