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 insertIntoTree(Kmer * kmer, SplayTree ** T)
{
SplayNode *newNode;
if (*T == NULL) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->left = newNode->right = NULL;
*T = newNode;
return;
}
*T = Splay(kmer, *T);
if (compareKmers(kmer, &((*T)->kmer)) < 0) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->left = (*T)->left;
newNode->right = *T;
(*T)->left = NULL;
*T = newNode;
} else if (compareKmers(&((*T)->kmer), kmer) < 0) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->right = (*T)->right;
newNode->left = *T;
(*T)->right = NULL;
*T = newNode;
}
}
boolean
findOrInsertOccurenceInSplayTree(Kmer * kmer, IDnum * seqID,
Coordinate * position, SplayTree ** T)
{
SplayNode *newNode;
if (*T == NULL) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->seqID = *seqID;
newNode->position = *position;
newNode->left = newNode->right = NULL;
*T = newNode;
return false;
}
*T = Splay(kmer, *T);
if (compareKmers(kmer, &((*T)->kmer)) < 0) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->seqID = *seqID;
newNode->position = *position;
newNode->left = (*T)->left;
newNode->right = *T;
(*T)->left = NULL;
*T = newNode;
printf("1: sequenceID = %d\n", *seqID);
return false;
} else if (compareKmers(kmer, &((*T)->kmer)) > 0) {
newNode = allocateSplayNode();
copyKmers(&(newNode->kmer), kmer);
newNode->seqID = *seqID;
newNode->position = *position;
newNode->right = (*T)->right;
newNode->left = *T;
(*T)->right = NULL;
*T = newNode;
printf("2: sequenceID = %d\n", *seqID);
return false;
} else {
*seqID = (*T)->seqID;
*position = (*T)->position;
printf("3: sequenceID = %d\n", *seqID);
return true;
}
}
int compareKmerOccurences(void const *A, void const *B)
{
KmerOccurence *a = (KmerOccurence *) A;
KmerOccurence *b = (KmerOccurence *) B;
if (compareKmers(&(a->kmer), &(b->kmer)) < 0)
return -1;
else if (compareKmers(&(a->kmer), &(b->kmer)) > 0)
return 1;
else
return 0;
}
KmerOccurence * getNextKmerOccurence(KmerOccurence * current) {
register KmerOccurence * next = current + 1;
if (next->nodeID == 0)
return NULL;
if (compareKmers(¤t->kmer, &next->kmer))
return NULL;
return next;
}
static SplayTree *Splay(Kmer * kmer, SplayTree * T)
{
SplayNode Header;
SplayNode *LeftTreeMax, *RightTreeMin;
if (T == NULL)
return NULL;
Header.left = Header.right = NULL;
LeftTreeMax = RightTreeMin = &Header;
while (compareKmers(kmer, &(T->kmer))) {
if (compareKmers(kmer, &(T->kmer)) < 0) {
if (T->left == NULL)
break;
if (compareKmers(kmer, &(T->left->kmer)) < 0)
T = SingleRotateWithLeft(T);
if (T->left == NULL)
break;
/* Link right */
RightTreeMin->left = T;
RightTreeMin = T;
T = T->left;
} else {
if (T->right == NULL)
break;
if (compareKmers(kmer, &(T->right->kmer)) > 0)
T = SingleRotateWithRight(T);
if (T->right == NULL)
break;
/* Link left */
LeftTreeMax->right = T;
LeftTreeMax = T;
T = T->right;
}
} /* while kmer != T->kmer */
/* Reassemble */
LeftTreeMax->right = T->left;
RightTreeMin->left = T->right;
T->left = Header.right;
T->right = Header.left;
return T;
}
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;
}
}
int compareKmerOccurences(void const *A, void const *B)
{
KmerOccurence *a = (KmerOccurence *) A;
KmerOccurence *b = (KmerOccurence *) B;
return compareKmers(&(a->kmer), &(b->kmer));
}
static void threadSequenceThroughGraph(TightString * tString,
KmerOccurenceTable * kmerTable,
Graph * graph,
IDnum seqID, Category category,
boolean readTracking,
boolean double_strand,
ReferenceMapping * referenceMappings,
Coordinate referenceMappingCount,
IDnum refCount,
Annotation * annotations,
IDnum annotationCount,
boolean second_in_pair)
{
Kmer word;
Kmer antiWord;
Coordinate readNucleotideIndex;
Coordinate kmerIndex;
KmerOccurence *kmerOccurence;
int wordLength = getWordLength(graph);
PassageMarkerI marker = NULL_IDX;
PassageMarkerI previousMarker = NULL_IDX;
Node *node = NULL;
Node *previousNode = NULL;
Coordinate coord = 0;
Coordinate previousCoord = 0;
Nucleotide nucleotide;
boolean reversed;
IDnum refID;
Coordinate refCoord = 0;
ReferenceMapping * refMap;
Annotation * annotation = annotations;
Coordinate index = 0;
Coordinate uniqueIndex = 0;
Coordinate annotIndex = 0;
IDnum annotCount = 0;
SmallNodeList * nodePile = NULL;
// Neglect any string shorter than WORDLENGTH :
if (getLength(tString) < wordLength)
return;
clearKmer(&word);
clearKmer(&antiWord);
// Fill in the initial word :
for (readNucleotideIndex = 0;
readNucleotideIndex < wordLength - 1; readNucleotideIndex++) {
nucleotide = getNucleotide(readNucleotideIndex, tString);
pushNucleotide(&word, nucleotide);
if (double_strand || second_in_pair) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
}
// Go through sequence
while (readNucleotideIndex < getLength(tString)) {
nucleotide = getNucleotide(readNucleotideIndex++, tString);
pushNucleotide(&word, nucleotide);
if (double_strand || second_in_pair) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
// Update annotation if necessary
if (annotCount < annotationCount && annotIndex == getAnnotationLength(annotation)) {
annotation = getNextAnnotation(annotation);
annotCount++;
annotIndex = 0;
}
// Search for reference mapping
if (category == REFERENCE) {
if (referenceMappings)
refMap = findReferenceMapping(seqID, index, referenceMappings, referenceMappingCount);
else
refMap = NULL;
if (refMap) {
node = getNodeInGraph(graph, refMap->nodeID);
if (refMap->nodeID > 0) {
coord = refMap->nodeStart + (index - refMap->referenceStart);
} else {
coord = getNodeLength(node) - refMap->nodeStart - refMap->length + (index - refMap->referenceStart);
}
} else {
node = NULL;
if (previousNode)
break;
}
}
// Search for reference-based mapping
else if (annotCount < annotationCount && uniqueIndex >= getPosition(annotation) && getAnnotSequenceID(annotation) <= refCount && getAnnotSequenceID(annotation) >= -refCount) {
refID = getAnnotSequenceID(annotation);
if (refID > 0)
refCoord = getStart(annotation) + annotIndex;
else
refCoord = getStart(annotation) - annotIndex;
refMap = findReferenceMapping(refID, refCoord, referenceMappings, referenceMappingCount);
// If success
if (refMap) {
if (refID > 0) {
node = getNodeInGraph(graph, refMap->nodeID);
if (refMap->nodeID > 0) {
coord = refMap->nodeStart + (refCoord - refMap->referenceStart);
} else {
coord = getNodeLength(node) - refMap->nodeStart - refMap->length + (refCoord - refMap->referenceStart);
}
} else {
node = getNodeInGraph(graph, -refMap->nodeID);
if (refMap->nodeID > 0) {
coord = getNodeLength(node) - refMap->nodeStart - (refCoord - refMap->referenceStart) - 1;
} else {
coord = refMap->nodeStart + refMap->length - (refCoord - refMap->referenceStart) - 1;
}
}
} else {
node = NULL;
if (previousNode)
break;
}
}
// Search in table
else {
reversed = false;
if (double_strand) {
if (compareKmers(&word, &antiWord) <= 0) {
kmerOccurence =
findKmerInKmerOccurenceTable(&word,
kmerTable);
} else {
kmerOccurence =
findKmerInKmerOccurenceTable(&antiWord,
kmerTable);
reversed = true;
}
} else {
if (!second_in_pair) {
kmerOccurence =
findKmerInKmerOccurenceTable(&word,
kmerTable);
} else {
kmerOccurence =
findKmerInKmerOccurenceTable(&antiWord,
kmerTable);
reversed = true;
}
}
if (kmerOccurence) {
if (!reversed) {
node = getNodeInGraph(graph, getKmerOccurenceNodeID(kmerOccurence));
coord = getKmerOccurencePosition(kmerOccurence);
} else {
node = getNodeInGraph(graph, -getKmerOccurenceNodeID(kmerOccurence));
coord = getNodeLength(node) - getKmerOccurencePosition(kmerOccurence) - 1;
}
} else {
node = NULL;
if (previousNode)
break;
}
}
// Increment positions
if (annotCount < annotationCount && uniqueIndex >= getPosition(annotation))
annotIndex++;
else
uniqueIndex++;
// Fill in graph
if (node)
{
#ifdef OPENMP
lockNode(node);
#endif
kmerIndex = readNucleotideIndex - wordLength;
if (previousNode == node
&& previousCoord == coord - 1) {
if (category / 2 >= CATEGORIES) {
setPassageMarkerFinish(marker,
kmerIndex +
1);
setFinishOffset(marker,
getNodeLength(node)
- coord - 1);
} else {
#ifndef SINGLE_COV_CAT
incrementVirtualCoverage(node, category / 2, 1);
incrementOriginalVirtualCoverage(node, category / 2, 1);
#else
incrementVirtualCoverage(node, 1);
#endif
}
#ifdef OPENMP
unLockNode(node);
#endif
} else {
if (category / 2 >= CATEGORIES) {
marker =
newPassageMarker(seqID,
kmerIndex,
kmerIndex + 1,
coord,
getNodeLength
(node) -
coord - 1);
transposePassageMarker(marker,
node);
connectPassageMarkers
(previousMarker, marker,
graph);
previousMarker = marker;
} else {
if (readTracking) {
if (!isNodeMemorized(node, nodePile)) {
addReadStart(node,
seqID,
coord,
graph,
kmerIndex);
memorizeNode(node, &nodePile);
} else {
blurLastShortReadMarker
(node, graph);
}
}
#ifndef SINGLE_COV_CAT
incrementVirtualCoverage(node, category / 2, 1);
incrementOriginalVirtualCoverage(node, category / 2, 1);
#else
incrementVirtualCoverage(node, 1);
#endif
}
#ifdef OPENMP
lockTwoNodes(node, previousNode);
#endif
createArc(previousNode, node, graph);
#ifdef OPENMP
unLockTwoNodes(node, previousNode);
#endif
}
previousNode = node;
previousCoord = coord;
}
index++;
}
if (readTracking && category / 2 < CATEGORIES)
unMemorizeNodes(&nodePile);
}
static void ghostThreadSequenceThroughGraph(TightString * tString,
KmerOccurenceTable *
kmerTable, Graph * graph,
IDnum seqID, Category category,
boolean readTracking,
boolean double_strand,
ReferenceMapping * referenceMappings,
Coordinate referenceMappingCount,
IDnum refCount,
Annotation * annotations,
IDnum annotationCount,
boolean second_in_pair)
{
Kmer word;
Kmer antiWord;
Coordinate readNucleotideIndex;
KmerOccurence *kmerOccurence;
int wordLength = getWordLength(graph);
Nucleotide nucleotide;
IDnum refID;
Coordinate refCoord;
ReferenceMapping * refMap = NULL;
Coordinate uniqueIndex = 0;
Coordinate annotIndex = 0;
IDnum annotCount = 0;
boolean reversed;
SmallNodeList * nodePile = NULL;
Annotation * annotation = annotations;
Node *node;
Node *previousNode = NULL;
// Neglect any read which will not be short paired
if ((!readTracking && category % 2 == 0)
|| category / 2 >= CATEGORIES)
return;
// Neglect any string shorter than WORDLENGTH :
if (getLength(tString) < wordLength)
return;
// Verify that all short reads are reasonnably short
if (getLength(tString) > USHRT_MAX) {
velvetLog("Short read of length %lli, longer than limit %i\n",
(long long) getLength(tString), SHRT_MAX);
velvetLog("You should better declare this sequence as long, because it genuinely is!\n");
exit(1);
}
clearKmer(&word);
clearKmer(&antiWord);
// Fill in the initial word :
for (readNucleotideIndex = 0;
readNucleotideIndex < wordLength - 1; readNucleotideIndex++) {
nucleotide = getNucleotide(readNucleotideIndex, tString);
pushNucleotide(&word, nucleotide);
if (double_strand || second_in_pair) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
}
// Go through sequence
while (readNucleotideIndex < getLength(tString)) {
// Shift word:
nucleotide = getNucleotide(readNucleotideIndex++, tString);
pushNucleotide(&word, nucleotide);
if (double_strand || second_in_pair) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
// Update annotation if necessary
if (annotCount < annotationCount && annotIndex == getAnnotationLength(annotation)) {
annotation = getNextAnnotation(annotation);
annotCount++;
annotIndex = 0;
}
// Search for reference mapping
if (annotCount < annotationCount && uniqueIndex >= getPosition(annotation) && getAnnotSequenceID(annotation) <= refCount && getAnnotSequenceID(annotation) >= -refCount) {
refID = getAnnotSequenceID(annotation);
if (refID > 0)
refCoord = getStart(annotation) + annotIndex;
else
refCoord = getStart(annotation) - annotIndex;
refMap = findReferenceMapping(refID, refCoord, referenceMappings, referenceMappingCount);
// If success
if (refMap) {
if (refID > 0)
node = getNodeInGraph(graph, refMap->nodeID);
else
node = getNodeInGraph(graph, -refMap->nodeID);
} else {
node = NULL;
if (previousNode)
break;
}
}
// if not.. look in table
else {
reversed = false;
if (double_strand) {
if (compareKmers(&word, &antiWord) <= 0) {
kmerOccurence =
findKmerInKmerOccurenceTable(&word,
kmerTable);
} else {
kmerOccurence =
findKmerInKmerOccurenceTable(&antiWord,
kmerTable);
reversed = true;
}
} else {
if (!second_in_pair) {
kmerOccurence =
findKmerInKmerOccurenceTable(&word,
kmerTable);
} else {
kmerOccurence =
findKmerInKmerOccurenceTable(&antiWord,
kmerTable);
reversed = true;
}
}
if (kmerOccurence) {
if (!reversed)
node = getNodeInGraph(graph, getKmerOccurenceNodeID(kmerOccurence));
else
node = getNodeInGraph(graph, -getKmerOccurenceNodeID(kmerOccurence));
} else {
node = NULL;
if (previousNode)
break;
}
}
if (annotCount < annotationCount && uniqueIndex >= getPosition(annotation))
annotIndex++;
else
uniqueIndex++;
previousNode = node;
// Fill in graph
if (node && !isNodeMemorized(node, nodePile))
{
#ifdef OPENMP
lockNode(node);
#endif
incrementReadStartCount(node, graph);
#ifdef OPENMP
unLockNode(node);
#endif
memorizeNode(node, &nodePile);
}
}
unMemorizeNodes(&nodePile);
}
static KmerOccurenceTable *referenceGraphKmers(char *preGraphFilename,
short int accelerationBits, Graph * graph, boolean double_strand, NodeMask * nodeMasks, Coordinate nodeMaskCount)
{
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;
IDnum index;
IDnum nodeID = 0;
Nucleotide nucleotide;
NodeMask * nodeMask = nodeMasks;
Coordinate nodeMaskIndex = 0;
if (file == NULL)
exitErrorf(EXIT_FAILURE, true, "Could not open %s", preGraphFilename);
// Count kmers
velvetLog("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);
kmerTable = newKmerOccurenceTable(accelerationBits, wordLength);
// 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;
}
velvetLog("%li kmers found\n", (long) kmerCount);
for(nodeMaskIndex = 0; nodeMaskIndex < nodeMaskCount; nodeMaskIndex++) {
kmerCount -= nodeMasks[nodeMaskIndex].finish -
nodeMasks[nodeMaskIndex].start;
}
nodeMaskIndex = 0;
fclose(file);
// Create table
allocateKmerOccurences(kmerCount, kmerTable);
// 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
}
// Update mask if necessary
if (nodeMask) {
if (nodeMask->nodeID < nodeID || (nodeMask->nodeID == nodeID && index >= nodeMask->finish)) {
if (++nodeMaskIndex == nodeMaskCount)
nodeMask = NULL;
else
nodeMask++;
}
}
// Check if not masked!
if (nodeMask) {
if (nodeMask->nodeID == nodeID && index >= nodeMask->start && index < nodeMask->finish) {
index++;
continue;
}
}
if (!double_strand || compareKmers(&word, &antiWord) <= 0)
recordKmerOccurence(&word, nodeID, index, kmerTable);
else
recordKmerOccurence(&antiWord, -nodeID, getNodeLength(getNodeInGraph(graph, nodeID)) - 1 - index, kmerTable);
index++;
}
if (fgets(line, maxline, file) == NULL)
break;
}
fclose(file);
// Sort table
sortKmerOccurenceTable(kmerTable);
return kmerTable;
}
static void threadSequenceThroughGraph(TightString * tString,
KmerOccurenceTable * kmerOccurences,
Graph * graph,
IDnum seqID, Category category,
boolean readTracking,
boolean double_strand)
{
Kmer word;
Kmer antiWord;
Coordinate readNucleotideIndex;
Coordinate kmerIndex;
KmerOccurence *kmerOccurence;
int wordLength = getWordLength(graph);
PassageMarker *marker = NULL;
PassageMarker *previousMarker = NULL;
Node *node;
Node *previousNode = NULL;
Coordinate coord;
Coordinate previousCoord = 0;
Nucleotide nucleotide;
clearKmer(&word);
clearKmer(&antiWord);
// Neglect any string shorter than WORDLENGTH :
if (getLength(tString) < wordLength)
return;
// Fill in the initial word :
for (readNucleotideIndex = 0;
readNucleotideIndex < wordLength - 1; readNucleotideIndex++) {
nucleotide = getNucleotide(readNucleotideIndex, tString);
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
}
// Go through sequence
while (readNucleotideIndex < getLength(tString)) {
nucleotide = getNucleotide(readNucleotideIndex++, tString);
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
// Search in table
if ((!double_strand || compareKmers(&word, &antiWord) <= 0)
&& (kmerOccurence =
findKmerOccurenceInSortedTable(&word,
kmerOccurences))) {
node =
getNodeInGraph(graph, kmerOccurence->nodeID);
coord = kmerOccurence->position;
} else if ((double_strand && compareKmers(&word, &antiWord) > 0)
&& (kmerOccurence =
findKmerOccurenceInSortedTable(&antiWord,
kmerOccurences)))
{
node =
getNodeInGraph(graph, -kmerOccurence->nodeID);
coord =
getNodeLength(node) - kmerOccurence->position -
1;
} else {
node = NULL;
if (previousNode) {
break;
}
}
// Fill in graph
if (node) {
kmerIndex = readNucleotideIndex - wordLength;
if (previousNode == node
&& previousCoord == coord - 1) {
if (category / 2 >= CATEGORIES) {
setPassageMarkerFinish(marker,
kmerIndex +
1);
setFinishOffset(marker,
getNodeLength(node)
- coord - 1);
} else {
incrementVirtualCoverage(node,
category /
2, 1);
incrementOriginalVirtualCoverage
(node, category / 2, 1);
}
} else {
if (category / 2 >= CATEGORIES) {
marker =
newPassageMarker(seqID,
kmerIndex,
kmerIndex + 1,
coord,
getNodeLength
(node) -
coord - 1);
transposePassageMarker(marker,
node);
connectPassageMarkers
(previousMarker, marker,
graph);
previousMarker = marker;
} else {
if (readTracking) {
if (!getNodeStatus(node)) {
addReadStart(node,
seqID,
coord,
graph,
kmerIndex);
setSingleNodeStatus
(node, true);
memorizeNode(node);
} else {
blurLastShortReadMarker
(node, graph);
}
}
incrementVirtualCoverage(node,
category /
2, 1);
incrementOriginalVirtualCoverage
(node, category / 2, 1);
}
createArc(previousNode, node, graph);
}
previousNode = node;
previousCoord = coord;
}
}
unlockMemorizedNodes();
}
static void ghostThreadSequenceThroughGraph(TightString * tString,
KmerOccurenceTable *
kmerOccurences, Graph * graph,
IDnum seqID, Category category,
boolean readTracking,
boolean double_strand)
{
Kmer word;
Kmer antiWord;
Coordinate readNucleotideIndex;
KmerOccurence *kmerOccurence;
int wordLength = getWordLength(graph);
Nucleotide nucleotide;
Node *node;
Node *previousNode = NULL;
clearKmer(&word);
clearKmer(&antiWord);
// Neglect any read which will not be short paired
if ((!readTracking && category % 2 == 0)
|| category / 2 >= CATEGORIES)
return;
// Neglect any string shorter than WORDLENGTH :
if (getLength(tString) < wordLength)
return;
// Verify that all short reads are reasonnably short
if (getLength(tString) > USHRT_MAX) {
printf("Short read of length %lli, longer than limit %i\n",
(long long) getLength(tString), SHRT_MAX);
puts("You should better declare this sequence as long, because it genuinely is!");
exit(1);
}
// Allocate memory for the read pairs
if (!readStartsAreActivated(graph))
activateReadStarts(graph);
// Fill in the initial word :
for (readNucleotideIndex = 0;
readNucleotideIndex < wordLength - 1; readNucleotideIndex++) {
nucleotide = getNucleotide(readNucleotideIndex, tString);
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
}
// Go through sequence
while (readNucleotideIndex < getLength(tString)) {
// Shift word:
nucleotide = getNucleotide(readNucleotideIndex++, tString);
pushNucleotide(&word, nucleotide);
if (double_strand) {
#ifdef COLOR
reversePushNucleotide(&antiWord, nucleotide);
#else
reversePushNucleotide(&antiWord, 3 - nucleotide);
#endif
}
// Search in table
if ((!double_strand || compareKmers(&word, &antiWord) <= 0)
&& (kmerOccurence =
findKmerOccurenceInSortedTable(&word,
kmerOccurences))) {
node =
getNodeInGraph(graph, kmerOccurence->nodeID);
} else if ((double_strand && compareKmers(&word, &antiWord) > 0)
&& (kmerOccurence =
findKmerOccurenceInSortedTable(&antiWord,
kmerOccurences)))
{
node =
getNodeInGraph(graph, -kmerOccurence->nodeID);
} else {
node = NULL;
if (previousNode)
break;
}
previousNode = node;
// Fill in graph
if (node && !getNodeStatus(node)) {
incrementReadStartCount(node, graph);
setSingleNodeStatus(node, true);
memorizeNode(node);
}
}
unlockMemorizedNodes();
}
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;
}
