static KmerOccurence *findKmerOccurenceInSortedTable(Kmer * kmer,
KmerOccurenceTable *
table)
{
KmerOccurence *array = table->kmerTable;
KmerKey key = keyInAccelerationTable(kmer, table);
Coordinate leftIndex, rightIndex, middleIndex;
if (table->accelerationTable != NULL) {
leftIndex = table->accelerationTable[key];
rightIndex = table->accelerationTable[key + 1];
} else {
leftIndex = 0;
rightIndex = table->kmerTableSize;
}
while (true) {
middleIndex = (rightIndex + leftIndex) / 2;
if (leftIndex >= rightIndex)
return NULL;
else if (compareKmers(&(array[middleIndex]).kmer, kmer) == 0)
return &(array[middleIndex]);
else if (leftIndex == middleIndex)
return NULL;
else if (compareKmers(&(array[middleIndex]).kmer, kmer) > 0)
rightIndex = middleIndex;
else
leftIndex = middleIndex;
}
}
void sortKmerOccurenceTable(KmerOccurenceTable * table) {
KmerKey lastHeader = 0;
KmerKey header;
IDnum *accelPtr = NULL;
IDnum kmerOccurenceIndex;
velvetLog("Sorting kmer occurence table ... \n");
qsort(table->kmerTable, table->kmerTableSize, sizeof(KmerOccurence),
compareKmerOccurences);
velvetLog("Sorting done.\n");
// Fill up acceleration table
if (table->accelerationTable != NULL) {
accelPtr = table->accelerationTable;
*accelPtr = (IDnum) 0;
for (kmerOccurenceIndex = 0;
kmerOccurenceIndex < table->kmerTableSize;
kmerOccurenceIndex++) {
header =
keyInAccelerationTable(&table->kmerTable
[kmerOccurenceIndex].
kmer, table);
while (lastHeader < header) {
lastHeader++;
accelPtr++;
*accelPtr = kmerOccurenceIndex;
}
}
while (lastHeader < (KmerKey) 1 << table->accelerationBits) {
lastHeader++;
accelPtr++;
*accelPtr = table->kmerTableSize;
}
}
}
KmerOccurence *findKmerInKmerOccurenceTable(Kmer * kmer,
KmerOccurenceTable *
table)
{
KmerOccurence *array = table->kmerTable;
KmerKey key = keyInAccelerationTable(kmer, table);
Coordinate leftIndex, rightIndex, middleIndex;
int diff;
if (table->accelerationTable != NULL) {
leftIndex = table->accelerationTable[key];
rightIndex = table->accelerationTable[key + 1];
} else {
leftIndex = 0;
rightIndex = table->kmerTableSize;
}
while (true) {
middleIndex = (rightIndex + leftIndex) / 2;
if (leftIndex >= rightIndex)
return NULL;
diff = compareKmers(&(array[middleIndex].kmer), kmer);
if (diff == 0) {
while (middleIndex > 0 && compareKmers(&(array[middleIndex - 1].kmer), kmer) == 0)
middleIndex--;
return &(array[middleIndex]);
} else if (leftIndex == middleIndex)
return NULL;
else if (diff > 0)
rightIndex = middleIndex;
else
leftIndex = middleIndex;
}
}
static KmerOccurenceTable *referenceGraphKmers(char *preGraphFilename,
short int accelerationBits, Graph * graph, boolean double_strand)
{
FILE *file = fopen(preGraphFilename, "r");
const int maxline = MAXLINE;
char line[MAXLINE];
char c;
int wordLength;
Coordinate lineLength, kmerCount;
Kmer word;
Kmer antiWord;
KmerOccurenceTable *kmerTable = NULL;
KmerOccurence *kmerOccurences, *kmerOccurencePtr;
Coordinate kmerOccurenceIndex;
IDnum index;
IDnum nodeID = 0;
IDnum *accelPtr = NULL;
KmerKey lastHeader = 0;
KmerKey header;
Nucleotide nucleotide;
if (file == NULL)
exitErrorf(EXIT_FAILURE, true, "Could not open %s", preGraphFilename);
// Count kmers
printf("Scanning pre-graph file %s for k-mers\n",
preGraphFilename);
// First line
if (!fgets(line, maxline, file))
exitErrorf(EXIT_FAILURE, true, "PreGraph file incomplete");
sscanf(line, "%*i\t%*i\t%i\n", &wordLength);
// Initialize kmer occurence table:
kmerTable = mallocOrExit(1, KmerOccurenceTable);
if (accelerationBits > 2 * wordLength)
accelerationBits = 2 * wordLength;
if (accelerationBits > 32)
accelerationBits = 32;
if (accelerationBits > 0) {
kmerTable->accelerationBits = accelerationBits;
kmerTable->accelerationTable =
callocOrExit((((size_t) 1) << accelerationBits) + 1,
IDnum);
accelPtr = kmerTable->accelerationTable;
kmerTable->accelerationShift =
(short int) 2 *wordLength - accelerationBits;
} else {
kmerTable->accelerationBits = 0;
kmerTable->accelerationTable = NULL;
kmerTable->accelerationShift = 0;
}
// Read nodes
if (!fgets(line, maxline, file))
exitErrorf(EXIT_FAILURE, true, "PreGraph file incomplete");
kmerCount = 0;
while (line[0] == 'N') {
lineLength = 0;
while ((c = getc(file)) != EOF && c != '\n')
lineLength++;
kmerCount += lineLength - wordLength + 1;
if (fgets(line, maxline, file) == NULL)
break;
}
fclose(file);
// Create table
printf("%li kmers found\n", (long) kmerCount);
kmerOccurences = callocOrExit(kmerCount, KmerOccurence);
kmerOccurencePtr = kmerOccurences;
kmerOccurenceIndex = 0;
kmerTable->kmerTable = kmerOccurences;
kmerTable->kmerTableSize = kmerCount;
// Fill table
file = fopen(preGraphFilename, "r");
if (file == NULL)
exitErrorf(EXIT_FAILURE, true, "Could not open %s", preGraphFilename);
if (!fgets(line, maxline, file))
exitErrorf(EXIT_FAILURE, true, "PreGraph file incomplete");
// Read nodes
if (!fgets(line, maxline, file))
exitErrorf(EXIT_FAILURE, true, "PreGraph file incomplete");
while (line[0] == 'N') {
nodeID++;
// Fill in the initial word :
clearKmer(&word);
clearKmer(&antiWord);
for (index = 0; index < wordLength - 1; index++) {
c = getc(file);
if (c == 'A')
nucleotide = ADENINE;
else if (c == 'C')
nucleotide = CYTOSINE;
else if (c == 'G')
nucleotide = GUANINE;
else if (c == 'T')
nucleotide = THYMINE;
else if (c == '\n')
exitErrorf(EXIT_FAILURE, true, "PreGraph file incomplete");
else
nucleotide = ADENINE;
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
}
// Scan through node
index = 0;
while((c = getc(file)) != '\n' && c != EOF) {
if (c == 'A')
nucleotide = ADENINE;
else if (c == 'C')
nucleotide = CYTOSINE;
else if (c == 'G')
nucleotide = GUANINE;
else if (c == 'T')
nucleotide = THYMINE;
else
nucleotide = ADENINE;
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
if (!double_strand || compareKmers(&word, &antiWord) <= 0) {
copyKmers(&kmerOccurencePtr->kmer, &word);
kmerOccurencePtr->nodeID = nodeID;
kmerOccurencePtr->position =
index;
} else {
copyKmers(&kmerOccurencePtr->kmer, &antiWord);
kmerOccurencePtr->nodeID = -nodeID;
kmerOccurencePtr->position =
getNodeLength(getNodeInGraph(graph, nodeID)) - 1 - index;
}
kmerOccurencePtr++;
kmerOccurenceIndex++;
index++;
}
if (fgets(line, maxline, file) == NULL)
break;
}
fclose(file);
// Sort table
qsort(kmerOccurences, kmerCount, sizeof(KmerOccurence),
compareKmerOccurences);
// Fill up acceleration table
if (kmerTable->accelerationTable != NULL) {
*accelPtr = (IDnum) 0;
for (kmerOccurenceIndex = 0;
kmerOccurenceIndex < kmerCount;
kmerOccurenceIndex++) {
header =
keyInAccelerationTable(&kmerOccurences
[kmerOccurenceIndex].
kmer, kmerTable);
while (lastHeader < header) {
lastHeader++;
accelPtr++;
*accelPtr = kmerOccurenceIndex;
}
}
while (lastHeader < (KmerKey) 1 << accelerationBits) {
lastHeader++;
accelPtr++;
*accelPtr = kmerCount;
}
}
return kmerTable;
}
