HSP *HSPLoad (MMPool *mmPool, const char *PackedDNAFileName, const char *AnnotationFileName) {
HSP *hsp;
int i;
FILE *annotationFile = NULL;
hsp = MMPoolDispatch(mmPool, sizeof(HSP));
// Load packed DNA
if (PackedDNAFileName != NULL && PackedDNAFileName[0] != '\0' && PackedDNAFileName[0] != '-') {
hsp->packedDNA = DNALoadPacked(PackedDNAFileName, &hsp->dnaLength, TRUE);
} else {
hsp->packedDNA = NULL;
hsp->dnaLength = 0;
}
// Load Annotation
// fprintf(stderr, "%s\n", AnnotationFileName);
if (AnnotationFileName != NULL && (annotationFile = fopen(AnnotationFileName, "r"))){
unsigned int numOfChar = 0;
int FASTARandomSeed =0;
int chrNum =0;
fscanf(annotationFile, "%u\t%d\t%d\n", &numOfChar, &chrNum, &FASTARandomSeed);
hsp->chrNum = chrNum;
hsp->chrName = MMUnitAllocate((chrNum+1)*sizeof(char *));
for (i=0; i<hsp->chrNum; ++i){
unsigned int nameLen = 0;
fscanf(annotationFile, "%u\t", &nameLen);
hsp->chrName[i] = MMUnitAllocate((nameLen+1) * sizeof(char));
fscanf(annotationFile, "%s\n", hsp->chrName[i]);
}
fscanf(annotationFile, "%d\n", &(hsp->numOfBlock));
hsp->blockList = MMUnitAllocate(((hsp->numOfBlock)+1) * sizeof(ChrBlock));
for(i=0;i<hsp->numOfBlock;++i){
ChrBlock *p = hsp->blockList+i;
int chrID, blockStart, blockEnd,ori;
chrID = blockStart = blockEnd = ori = 0;
fscanf(annotationFile, "%d\t%u\t%u\t%u\n", &(p->chrID), &(p->blockStart), &(p->blockEnd), &(p->ori));
}
}
/* hsp->dnaLength = hsp->blockList[hsp->numOfBlock-1].blockEnd;
*/
#ifdef DEBUG
int j = hsp->numOfBlock;
fprintf(stderr, "%d\t%d\t%u\t%u\n", hsp->blockList[j-1].chrID, hsp->blockList[j-1].ori, hsp->blockList[j-1].blockStart, hsp->blockList[j-1].blockEnd);
#endif
fclose(annotationFile);
return hsp;
}
Socket *SocketCreate(const int socketType, const char *socketPath) {
Socket *socketPointer;
struct sockaddr_un local;
int masterSocket;
int length;
if (socketType != LOCAL_SOCKET) {
fprintf(stderr, "Only local socket is supported!\n");
exit(1);
}
if (socketPath == '\0' || socketPath[0] == '\0') {
fprintf(stderr, "Socket path is not specified!\n");
exit(1);
}
masterSocket = socket(AF_UNIX, SOCK_STREAM, 0);
if (masterSocket == -1) {
return NULL;
}
local.sun_family = AF_UNIX;
strncpy(local.sun_path, socketPath, SOCKET_MAX_PATH_LENGTH);
local.sun_path[SOCKET_MAX_PATH_LENGTH] = '\0';
unlink(local.sun_path);
length = strlen(local.sun_path) + sizeof(local.sun_family);
if (bind(masterSocket, (struct sockaddr *)&local, length) == -1) {
return NULL;
}
socketPointer = MMUnitAllocate(sizeof(Socket));
socketPointer->masterSocket = masterSocket;
socketPointer->currentSocket = -1;
socketPointer->socketType = socketType;
socketPointer->socketPath = MMUnitAllocate(strlen(socketPath) + 1);
strcpy(socketPointer->socketPath, socketPath);
return socketPointer;
}
Socket *SocketInitiateConnection(const int socketType, const char *socketPath) {
int length;
struct sockaddr_un remote;
int currentSocket;
Socket *socketPointer;
if (socketType != LOCAL_SOCKET) {
fprintf(stderr, "Only local socket is supported!\n");
exit(1);
}
if (socketPath == '\0' || socketPath[0] == '\0') {
fprintf(stderr, "Socket path is not specified!\n");
exit(1);
}
currentSocket = socket(AF_UNIX, SOCK_STREAM, 0);
if (currentSocket == -1) {
return NULL;
}
remote.sun_family = AF_UNIX;
strncpy(remote.sun_path, socketPath, SOCKET_MAX_PATH_LENGTH);
length = strlen(remote.sun_path) + sizeof(remote.sun_family);
if (connect(currentSocket, (struct sockaddr *)&remote, length) == -1) {
return NULL;
}
socketPointer = MMUnitAllocate(sizeof(Socket));
socketPointer->masterSocket = -1;
socketPointer->currentSocket = currentSocket;
socketPointer->socketType = socketType;
socketPointer->socketPath = MMUnitAllocate(strlen(socketPath) + 1);
strcpy(socketPointer->socketPath, socketPath);
return socketPointer;
}
unsigned int ProcessSearchPattern() {
FILE *searchPatternFile;
char c;
unsigned int i, j, n;
searchPatternFile = fopen64(PatternFileName, "r");
if (searchPatternFile == NULL) {
fprintf(stderr, "Cannot open search pattern file : %s\n", PatternFileName);
exit(1);
}
j = 0;
n = 0;
// Find out the size of search pattern
while (!feof(searchPatternFile)) {
c = (char)fgetc(searchPatternFile);
// Find the first comment line
while (!feof(searchPatternFile) && c != '>') {
c = (char)fgetc(searchPatternFile);
}
while (!feof(searchPatternFile)) {
// Skip comment line
while (!feof(searchPatternFile) && c != '\n') {
c = (char)fgetc(searchPatternFile);
}
// Count no. of characters
while (!feof(searchPatternFile) && c != '>') {
if (c != '\n') {
j++;
}
c = (char)fgetc(searchPatternFile);
}
// Count no. of patterns
n++;
}
}
if (j > searchPatternAllocated) {
if (searchPattern != NULL) {
MMUnitFree(searchPattern, searchPatternAllocated);
}
searchPattern = MMUnitAllocate(j);
searchPatternAllocated = j;
}
if (n + 1 > searchPatternPositionAllocated) {
if (searchPatternPosition != NULL) {
MMUnitFree(searchPatternPosition, searchPatternPositionAllocated);
}
searchPatternPosition = MMUnitAllocate((n + 1) * sizeof(unsigned int));
searchPatternPositionAllocated = n + 1;
}
fseek(searchPatternFile, 0, SEEK_SET);
j = 0;
n = 0;
i = 0;
searchPatternPosition[0] = 0;
// Read pattern into memory
while (!feof(searchPatternFile)) {
c = (char)fgetc(searchPatternFile);
// Find the first comment line
while (!feof(searchPatternFile) && c != '>') {
c = (char)fgetc(searchPatternFile);
}
while (!feof(searchPatternFile)) {
// Skip comment line
while (!feof(searchPatternFile) && c != '\n') {
c = (char)fgetc(searchPatternFile);
}
// Store pattern
while (!feof(searchPatternFile) && c != '>') {
if (c != '\n') {
searchPattern[j] = c;
j++;
}
c = (char)fgetc(searchPatternFile);
}
// Store pointers to pattern
n++;
searchPatternPosition[n] = j;
}
}
fclose(searchPatternFile);
return n;
}
int main(int argc, char** argv) {
char c;
unsigned int i, j, k;
MMPool *mmPool;
dictionary *programInput, *ini;
char argumentText[11] = "argument:0";
double startTime;
double elapsedTime = 0, totalElapsedTime = 0;
BWT *bwt;
HSP *hsp;
unsigned char charMap[256];
unsigned char *convertedKey;
unsigned int *packedKey;
unsigned int found;
unsigned int numberOfPattern, numberOfPatternFound;
unsigned int saIndexLeft, saIndexRight;
SaIndexGroupNew *saIndexGroup;
SaIndexList *tempSaIndex1, *tempSaIndex2;
unsigned int numberOfSaIndexGroup;
HitList *hitList;
unsigned int textPositionMatched, textPositionRetrieved;
unsigned long long totalTextPositionMatched, totalTextPositionRetrieved;
BWTSaRetrievalStatistics BWTSaRetrievalStatistics;
FILE *logFile;
init(textPositionRetrieved); // to avoid compiler warning only
init(textPositionMatched); // to avoid compiler warning only
programInput = ParseInput(argc, argv);
ini = ParseIniFile();
ProcessIni();
ValidateIni();
PrintIni();
if (Confirmation == TRUE) {
printf("BWT Search. Press Y to go or N to cancel. ");
c = (char)getchar();
while (c != 'y' && c != 'Y' && c != 'n' && c!= 'N') {
c = (char)getchar();
}
if (c == 'n' || c == 'N') {
exit(0);
}
}
startTime = setStartTime();
MMMasterInitialize(1, 0, FALSE, NULL);
mmPool = MMPoolCreate(PoolSize);
// Load Database
printf("Loading Database..");
fflush(stdout);
bwt = BWTLoad(mmPool, BWTCodeFileName, BWTOccValueFileName, SaValueFileName, NULL, SaIndexFileName, NULL);
HSPFillCharMap(charMap);
hsp = HSPLoad(mmPool, PackedDNAFileName, AnnotationFileName, AmbiguityFileName, MAX_SEARCH_PATTERN_LENGTH / CHAR_PER_WORD);
if (bwt->textLength != hsp->dnaLength) {
fprintf(stderr, "BWT-BLAST: Database length inconsistent!\n");
exit(1);
}
if (LogFileName[0] != '\0') {
logFile = fopen64(LogFileName, "w");
if (logFile == NULL) {
fprintf(stderr, "Cannot open log file!\n");
exit(1);
}
} else {
logFile = NULL;
}
packedKey = MMPoolDispatch(mmPool, WordPackedLengthFromText(MAX_SEARCH_PATTERN_LENGTH, BIT_PER_CHAR));
convertedKey = MMPoolDispatch(mmPool, MAX_SEARCH_PATTERN_LENGTH);
saIndexGroup = MMUnitAllocate(MaxNumberOfHitGroups * sizeof(SaIndexGroup));
hitList = MMUnitAllocate(MaxNumberOfTextPosition * sizeof(HitList));
tempSaIndex1 = MMUnitAllocate(MaxNumberOfTextPosition * sizeof(SaIndexList));
tempSaIndex2 = MMUnitAllocate(MaxNumberOfTextPosition * sizeof(SaIndexList));
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
// Process search pattern files
for (i=1; i<=NumberOfSearchPatternFile; i++) {
argumentText[9] = '0' + (char)(i + 2);
iniparser_copystring(programInput, argumentText, PatternFileName, PatternFileName, MAX_FILENAME_LEN);
printf("Loading search pattern : %s\n", PatternFileName);
numberOfPattern = ProcessSearchPattern();
printf("Finished loading search pattern.\n");
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Searching for pattern : %s\n", PatternFileName);
}
if (SABinarySearch == TRUE) {
printf("Start forward search with SA index.\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Forward search with SA index.\n");
}
numberOfPatternFound = 0;
totalTextPositionMatched = 0;
totalTextPositionRetrieved = 0;
BWTInitializeSaRetrievalStatistics(&BWTSaRetrievalStatistics);
for (j=0; j<numberOfPattern; j++) {
//ConvertTextToWordPacked(searchPattern + searchPatternPosition[j], packedKey, charMap, ALPHABET_SIZE,
// searchPatternPosition[j+1] - searchPatternPosition[j]);
//found = BWTForwardSearchSaIndex(packedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
// bwt, hsp->packedDNA, &saIndexLeft, &saIndexRight);
ConvertTextToCode(searchPattern + searchPatternPosition[j], convertedKey, charMap,
searchPatternPosition[j+1] - searchPatternPosition[j]);
found = BWTSaBinarySearch(convertedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
bwt, hsp->packedDNA, &saIndexLeft, &saIndexRight, packedKey);
if (found) {
numberOfPatternFound++;
textPositionMatched = saIndexRight - saIndexLeft + 1;
totalTextPositionMatched += textPositionMatched;
if (FindTextPosition) {
saIndexGroup->startSaIndex = saIndexLeft;
if (textPositionMatched <= MaxNumberOfTextPosition) {
saIndexGroup->numOfMatch = textPositionMatched;
} else {
saIndexGroup->numOfMatch = MaxNumberOfTextPosition;
printf("Max no of text positions reached! Only %u out of %u text positions retrieved.\n", MaxNumberOfTextPosition, textPositionMatched);
}
saIndexGroup->posQuery = 0;
saIndexGroup->info = 0;
textPositionRetrieved = BWTTextPosition(bwt, saIndexGroup, 1, hitList, tempSaIndex1, tempSaIndex2, &BWTSaRetrievalStatistics, FALSE);
totalTextPositionRetrieved += textPositionRetrieved;
}
}
if (LogFileName[0] != '\0') {
if (found) {
fprintf(logFile, "Pattern number %u found. SA Index from %u to %u. ",
j + 1, saIndexLeft, saIndexRight);
if (FindTextPosition) {
fprintf(logFile, "%u matches of %u located. ", textPositionRetrieved, textPositionMatched);
for (k=0; k<textPositionRetrieved; k++) {
fprintf(logFile, "%u ", hitList[k].posText);
}
}
} else {
fprintf(logFile, "Pattern number %u not found.", j + 1);
}
fprintf(logFile, "\n");
}
}
printf("Finished forward search with SA index.\n");
printf("%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
printf("Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Finished forward search with SA index.\n");
fprintf(logFile, "%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
fprintf(logFile, "Elapsed time = ");
printElapsedTime(logFile, FALSE, FALSE, TRUE, 4, elapsedTime);
fprintf(logFile, "\n");
}
}
if (BackwardSearch == TRUE) {
printf("Start backward search.\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Backward search.\n");
}
numberOfPatternFound = 0;
totalTextPositionMatched = 0;
totalTextPositionRetrieved = 0;
BWTInitializeSaRetrievalStatistics(&BWTSaRetrievalStatistics);
for (j=0; j<numberOfPattern; j++) {
ConvertTextToCode(searchPattern + searchPatternPosition[j], convertedKey, charMap,
searchPatternPosition[j+1] - searchPatternPosition[j]);
found = BWTBackwardSearch(convertedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
bwt, &saIndexLeft, &saIndexRight);
if (found) {
numberOfPatternFound++;
textPositionMatched = saIndexRight - saIndexLeft + 1;
totalTextPositionMatched += textPositionMatched;
if (FindTextPosition) {
saIndexGroup->startSaIndex = saIndexLeft;
if (textPositionMatched <= MaxNumberOfTextPosition) {
saIndexGroup->numOfMatch = textPositionMatched;
} else {
saIndexGroup->numOfMatch = MaxNumberOfTextPosition;
printf("Max no of text positions reached! Only %u out of %u text positions retrieved.\n", MaxNumberOfTextPosition, textPositionMatched);
}
saIndexGroup->posQuery = 0;
saIndexGroup->info = 0;
textPositionRetrieved = BWTTextPosition(bwt, saIndexGroup, 1, hitList, tempSaIndex1, tempSaIndex2, &BWTSaRetrievalStatistics, FALSE);
totalTextPositionRetrieved += textPositionRetrieved;
}
}
if (LogFileName[0] != '\0') {
if (found) {
fprintf(logFile, "Pattern number %u found. SA Index from %u to %u. Total %u occurrences ", j + 1, saIndexLeft, saIndexRight, saIndexRight - saIndexLeft + 1);
if (FindTextPosition) {
fprintf(logFile, "%u matches of %u located. ", textPositionRetrieved, textPositionMatched);
for (k=0; k<textPositionRetrieved; k++) {
fprintf(logFile, "%u ", hitList[k].posText);
}
}
} else {
fprintf(logFile, "Pattern number %u not found.", j + 1);
}
fprintf(logFile, "\n");
}
}
printf("Finished backward search.\n");
printf("%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
printf("Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Finished backward search.\n");
fprintf(logFile, "%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
fprintf(logFile, "Elapsed time = ");
printElapsedTime(logFile, FALSE, FALSE, TRUE, 4, elapsedTime);
fprintf(logFile, "\n");
}
}
if (BackwardSearchCheck == TRUE) {
printf("Start backward search with text.\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Backward search with text.\n");
}
numberOfPatternFound = 0;
totalTextPositionMatched = 0;
totalTextPositionRetrieved = 0;
BWTInitializeSaRetrievalStatistics(&BWTSaRetrievalStatistics);
for (j=0; j<numberOfPattern; j++) {
ConvertTextToCode(searchPattern + searchPatternPosition[j], convertedKey, charMap,
searchPatternPosition[j+1] - searchPatternPosition[j]);
ConvertTextToWordPacked(searchPattern + searchPatternPosition[j], packedKey, charMap, ALPHABET_SIZE,
searchPatternPosition[j+1] - searchPatternPosition[j]);
found = BWTBackwardSearchCheckWithText(convertedKey, packedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
bwt, hsp->packedDNA, TextCheckCostFactor, MaxNumberOfTextPosition,
hitList, &saIndexLeft, &saIndexRight);
if (found) {
numberOfPatternFound++;
textPositionMatched = saIndexRight - saIndexLeft + 1;
totalTextPositionMatched += textPositionMatched;
// Find text position if text check not used
if (FindTextPosition && saIndexLeft != 0) {
saIndexGroup->startSaIndex = saIndexLeft;
if (textPositionMatched <= MaxNumberOfTextPosition) {
saIndexGroup->numOfMatch = textPositionMatched;
} else {
saIndexGroup->numOfMatch = MaxNumberOfTextPosition;
printf("Max no of text positions reached! Only %u out of %u text positions retrieved.\n", MaxNumberOfTextPosition, textPositionMatched);
}
saIndexGroup->posQuery = 0;
saIndexGroup->info = 0;
textPositionRetrieved = BWTTextPosition(bwt, saIndexGroup, 1, hitList, tempSaIndex1, tempSaIndex2, &BWTSaRetrievalStatistics, FALSE);
totalTextPositionRetrieved += textPositionRetrieved;
} else {
textPositionRetrieved = textPositionMatched;
totalTextPositionRetrieved += textPositionRetrieved;
}
}
if (LogFileName[0] != '\0') {
if (found) {
fprintf(logFile, "Pattern number %u found. ", j + 1);
if (FindTextPosition) {
fprintf(logFile, "%u matches of %u located. ", textPositionRetrieved, textPositionMatched);
for (k=0; k<textPositionRetrieved; k++) {
fprintf(logFile, "%u ", hitList[k].posText);
}
}
} else {
fprintf(logFile, "Pattern number %u not found.", j + 1);
}
fprintf(logFile, "\n");
}
}
printf("Finished backward search with text.\n");
printf("%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
printf("Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Finished backward search with text.\n");
fprintf(logFile, "%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
fprintf(logFile, "Elapsed time = ");
printElapsedTime(logFile, FALSE, FALSE, TRUE, 4, elapsedTime);
fprintf(logFile, "\n");
}
}
if (HammingDistSearch == TRUE) {
printf("Start hamming distance %u approximate match.\n", MaxErrorAllowed);
if (LogFileName[0] != '\0') {
fprintf(logFile, "Hamming distance %u approximate match.\n", MaxErrorAllowed);
}
numberOfPatternFound = 0;
totalTextPositionMatched = 0;
totalTextPositionRetrieved = 0;
BWTInitializeSaRetrievalStatistics(&BWTSaRetrievalStatistics);
for (j=0; j<numberOfPattern; j++) {
ConvertTextToCode(searchPattern + searchPatternPosition[j], convertedKey, charMap,
searchPatternPosition[j+1] - searchPatternPosition[j]);
numberOfSaIndexGroup = BWTHammingDistMatchOld(convertedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
bwt, MaxErrorAllowed, saIndexGroup, MaxNumberOfHitGroups, 0, 0);
if (numberOfSaIndexGroup) {
numberOfPatternFound++;
textPositionMatched = 0;
for (k=0; k<numberOfSaIndexGroup; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
if (textPositionMatched > MaxNumberOfTextPosition) {
textPositionMatched = 0;
for (k=0; k<numberOfSaIndexGroup && textPositionMatched < MaxNumberOfTextPosition; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
numberOfSaIndexGroup = k - 1;
saIndexGroup[numberOfSaIndexGroup].numOfMatch -= textPositionMatched - MaxNumberOfTextPosition;
for (; k<numberOfSaIndexGroup; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
printf("Max no of text positions reached! Only %u out of %u text positions retrieved.\n", MaxNumberOfTextPosition, textPositionMatched);
}
totalTextPositionMatched += textPositionMatched;
if (FindTextPosition) {
textPositionRetrieved = BWTTextPosition(bwt, saIndexGroup, numberOfSaIndexGroup, hitList, tempSaIndex1, tempSaIndex2, &BWTSaRetrievalStatistics, FALSE);
totalTextPositionRetrieved += textPositionRetrieved;
}
}
if (LogFileName[0] != '\0') {
if (numberOfSaIndexGroup) {
fprintf(logFile, "Pattern number %u found. %u matches", j + 1, textPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "%u matches of %u located. ", textPositionRetrieved, textPositionMatched);
for (k=0; k<textPositionRetrieved; k++) {
fprintf(logFile, "%u ", hitList[k].posText);
}
}
} else {
fprintf(logFile, "Pattern number %u not found.", j + 1);
}
fprintf(logFile, "\n");
}
}
printf("Finished hamming distance search.\n");
printf("%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
printf("Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Finished hamming distance search.\n");
fprintf(logFile, "%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
fprintf(logFile, "Elapsed time = ");
printElapsedTime(logFile, FALSE, FALSE, TRUE, 4, elapsedTime);
fprintf(logFile, "\n");
}
}
if (EditDistSearch == TRUE) {
printf("Start edit distance %u approximate match.\n", MaxErrorAllowed);
if (LogFileName[0] != '\0') {
fprintf(logFile, "Edit distance %u approximate match.\n", MaxErrorAllowed);
}
numberOfPatternFound = 0;
totalTextPositionMatched = 0;
totalTextPositionRetrieved = 0;
BWTInitializeSaRetrievalStatistics(&BWTSaRetrievalStatistics);
for (j=0; j<numberOfPattern; j++) {
ConvertTextToCode(searchPattern + searchPatternPosition[j], convertedKey, charMap,
searchPatternPosition[j+1] - searchPatternPosition[j]);
numberOfSaIndexGroup = BWTEditDistMatchOld(convertedKey, searchPatternPosition[j+1] - searchPatternPosition[j],
bwt, MaxErrorAllowed, (SaIndexGroupWithLengthError*)saIndexGroup, MaxNumberOfHitGroups);
if (numberOfSaIndexGroup > MaxNumberOfHitGroups) {
fprintf(stderr, "numberOfSaIndexGroup > MaxNumberOfHitGroups!\n");
}
if (numberOfSaIndexGroup) {
numberOfPatternFound++;
textPositionMatched = 0;
for (k=0; k<numberOfSaIndexGroup; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
if (textPositionMatched > MaxNumberOfTextPosition) {
textPositionMatched = 0;
for (k=0; k<numberOfSaIndexGroup && textPositionMatched < MaxNumberOfTextPosition; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
numberOfSaIndexGroup = k - 1;
saIndexGroup[numberOfSaIndexGroup].numOfMatch -= textPositionMatched - MaxNumberOfTextPosition;
for (; k<numberOfSaIndexGroup; k++) {
textPositionMatched += saIndexGroup[k].numOfMatch;
}
printf("Max no of text positions reached! Only %u out of %u text positions retrieved.\n", MaxNumberOfTextPosition, textPositionMatched);
}
totalTextPositionMatched += textPositionMatched;
if (FindTextPosition) {
textPositionRetrieved = BWTTextPosition(bwt, saIndexGroup, numberOfSaIndexGroup, hitList, tempSaIndex1, tempSaIndex2, &BWTSaRetrievalStatistics, FALSE);
totalTextPositionRetrieved += textPositionRetrieved;
}
}
if (LogFileName[0] != '\0') {
if (numberOfSaIndexGroup) {
fprintf(logFile, "Pattern number %u found. %u matches. ", j + 1, textPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "%u matches of %u located. ", textPositionRetrieved, textPositionMatched);
for (k=0; k<textPositionRetrieved; k++) {
fprintf(logFile, "%u ", hitList[k].posText);
}
}
} else {
fprintf(logFile, "Pattern number %u not found.", j + 1);
}
fprintf(logFile, "\n");
}
}
printf("Finished edit distance search.\n");
printf("%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
printf("Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
printf("Elapsed time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 4, elapsedTime);
totalElapsedTime += elapsedTime;
printf("\n");
if (LogFileName[0] != '\0') {
fprintf(logFile, "Finished edit distance search.\n");
fprintf(logFile, "%u patterns out of %u found. %llu of %llu matches located.\n", numberOfPatternFound, numberOfPattern, totalTextPositionRetrieved, totalTextPositionMatched);
if (FindTextPosition) {
fprintf(logFile, "Hits: BWT/Cached/Diag/Dup : %u/%u/%u/%u\n",
BWTSaRetrievalStatistics.bwtSaRetrieved, BWTSaRetrievalStatistics.cachedSaRetrieved,
BWTSaRetrievalStatistics.saDiagonalLinked, BWTSaRetrievalStatistics.saDuplicated);
}
fprintf(logFile, "Elapsed time = ");
printElapsedTime(logFile, FALSE, FALSE, TRUE, 4, elapsedTime);
fprintf(logFile, "\n");
}
}
}
MMPoolReturn(mmPool, packedKey, WordPackedLengthFromText(MAX_SEARCH_PATTERN_LENGTH, BIT_PER_CHAR));
MMPoolReturn(mmPool, convertedKey, MAX_SEARCH_PATTERN_LENGTH);
HSPFree(mmPool, hsp, MAX_SEARCH_PATTERN_LENGTH / CHAR_PER_WORD);
BWTFree(mmPool, bwt);
if (searchPattern != NULL) {
MMUnitFree(searchPattern, searchPatternAllocated);
}
if (searchPatternPosition != NULL) {
MMUnitFree(searchPatternPosition, searchPatternPositionAllocated);
}
MMUnitFree(saIndexGroup, MaxNumberOfHitGroups * sizeof(unsigned int));
MMUnitFree(hitList, MaxNumberOfTextPosition * sizeof(unsigned int));
MMPoolFree(mmPool);
iniparser_freedict(programInput);
iniparser_freedict(ini);
return 0;
}
int main(int argc, char** argv) {
unsigned int* __restrict address;
unsigned int* __restrict memory;
unsigned int i, j;
unsigned int t=0;
unsigned int numberOfMemoryLocation;
dictionary *programInput;
double startTime;
double elapsedTime = 0, totalElapsedTime = 0;
double initializationTime, experimentTime;
programInput = ParseInput(argc, argv);
ValidateParameters();
// Initialize memory manager
MMMasterInitialize(0, 0, FALSE, NULL);
numberOfMemoryLocation = MemorySize / sizeof(int);
// Allocate memory
address = MMUnitAllocate((NumberOfAccess + 8) * sizeof(unsigned int));
memory = MMUnitAllocate((numberOfMemoryLocation + 32) * sizeof(unsigned int));
// Set random seed
r250_init(getRandomSeed());
// Set start time
startTime = setStartTime();
printf("Initialize memory pointers with random values..");
for (i=0; i<numberOfMemoryLocation + 32; i++) {
memory[i] = r250();
}
// Initialize address and memory
for (i=0; i<NumberOfAccess + 8; i++) {
address[i] = (r250() % numberOfMemoryLocation) / 4 * 4;
}
printf("finished.\n");
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
totalElapsedTime += elapsedTime;
initializationTime = elapsedTime;
// Test memory speed for read
if (ReadWrite[0] == 'R' || ReadWrite[0] == 'r') {
for (i=0; i<NumberOfIteration; i++) {
for (j=0; j<NumberOfAccess; j++) {
t += memory[address[j]];
}
}
}
// Test memory speed for write
if (ReadWrite[0] == 'W' || ReadWrite[0] == 'w') {
for (i=0; i<NumberOfIteration; i++) {
for (j=0; j<NumberOfAccess; j++) {
memory[address[j]] += address[j];
}
}
}
elapsedTime = getElapsedTime(startTime) - totalElapsedTime;
totalElapsedTime += elapsedTime;
experimentTime = elapsedTime;
// So that compiler does not remove code for variables t and r
if (t==0) {
printf("\n");
}
printf("Experiment completed.\n");
printf("Initialization time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 2, initializationTime);
printf("Experiment time = ");
printElapsedTime(stdout, FALSE, FALSE, TRUE, 2, experimentTime);
MMUnitFree(address, (NumberOfAccess + 8) * sizeof(unsigned int));
MMUnitFree(memory, (numberOfMemoryLocation + 32) * sizeof(unsigned int));
iniparser_freedict(programInput);
return 0;
}