Connection *createNewConnection(IDnum nodeID, IDnum node2ID,
IDnum direct_count,
IDnum paired_count,
Coordinate distance,
double variance)
{
Node *destination = getNodeInGraph(graph, node2ID);
IDnum nodeIndex = nodeID + nodeCount(graph);
Connection *connect = allocateConnection();
// Fill in
connect->destination = destination;
connect->direct_count = direct_count;
connect->paired_count = paired_count;
connect->distance = distance;
connect->variance = variance;
// Insert in scaffold
connect->previous = NULL;
connect->next = scaffold[nodeIndex];
if (scaffold[nodeIndex] != NULL)
scaffold[nodeIndex]->previous = connect;
scaffold[nodeIndex] = connect;
// Event. pair up to twin
if (getUniqueness(destination))
createTwinConnection(node2ID, nodeID, connect);
else
connect->twin = NULL;
return connect;
}
static void findOppositeNode(Node * node, Node ** oppositeNode,
Coordinate * distance)
{
NodeList *nodeList;
MiniConnection *localConnect;
Node *node2;
IDnum node2ID;
*oppositeNode = NULL;
*distance = 0;
for (nodeList = markedNodes; nodeList != NULL;
nodeList = nodeList->next) {
node2 = nodeList->node;
node2ID = getNodeID(node2);
localConnect = &localScaffold[node2ID + nodeCount(graph)];
if (node2 == node)
continue;
if (!getUniqueness(node2))
continue;
if (localConnect->distance < 0)
continue;
if (*oppositeNode == NULL
|| *distance > localConnect->distance) {
*oppositeNode = node2;
*distance = localConnect->distance;
}
}
}
static void projectFromNode(IDnum nodeID,
ReadOccurence ** readNodes,
IDnum * readNodeCounts,
IDnum * readPairs, Category * cats,
boolean * dubious, Coordinate * lengths)
{
IDnum index;
ShortReadMarker *nodeArray, *shortMarker;
PassageMarker *marker;
Node *node;
IDnum nodeReadCount;
node = getNodeInGraph(graph, nodeID);
if (node == NULL || !getUniqueness(node))
return;
nodeArray = getNodeReads(node, graph);
nodeReadCount = getNodeReadCount(node, graph);
for (index = 0; index < nodeReadCount; index++) {
shortMarker = getShortReadMarkerAtIndex(nodeArray, index);
if (dubious[getShortReadMarkerID(shortMarker) - 1])
continue;
projectFromShortRead(node, shortMarker, readPairs, cats,
readNodes, readNodeCounts, lengths);
}
for (marker = getMarker(node); marker != NULL;
marker = getNextInNode(marker)) {
if (getPassageMarkerSequenceID(marker) > 0)
projectFromLongRead(node, marker, readPairs, cats,
readNodes, readNodeCounts,
lengths);
}
}
void renumberLocusNodes(Locus * locus) {
IDnum index;
Node * node;
IDnum counter = 0;
Node ** newArray;
for (index = 0; index < locus->contigCount; index++) {
node = locus->contigs[index];
if (!getNodeStatus(node)) {
locus->contigs[index] = NULL;
counter++;
if (getUniqueness(node))
locus->longContigCount--;
}
}
if (counter == 0)
return;
newArray = callocOrExit(locus->contigCount - counter, Node *);
counter = 0;
for (index = 0; index < locus->contigCount; index++) {
node = locus->contigs[index];
if (node == NULL)
counter++;
else
newArray[index - counter] = node;
}
free(locus->contigs);
locus->contigs = newArray;
locus->contigCount -= counter;
}
static void clipTipsVeryHardLocally()
{
NodeList *nodeList, *next;
Node *current, *twin;
boolean modified = true;
//velvetLog("Clipping short tips off graph HARD\n");
while (modified) {
modified = false;
for (nodeList = getMarkedNodeList(); nodeList != NULL;
nodeList = next) {
next = nodeList->next;
current = nodeList->node;
if (current == NULL || getNodeStatus(current) != 1)
continue;
if (getUniqueness(current))
continue;
//velvetLog("Checking node HARD %li %i\n", (long)getNodeID(current), simpleArcCount(current));
twin = getTwinNode(current);
if( isLocalDeadEnd(current) || isLocalTwinDeadEnd(current) ){
//velvetLog("Found tip at node %li\n", (long)getNodeID(current));
handicapNode(current);
modified = true;
}
}
}
}
static void trimLongReadTips()
{
IDnum index;
Node *node;
PassageMarkerI marker, next;
velvetLog("Trimming read tips\n");
for (index = 1; index <= nodeCount(graph); index++) {
node = getNodeInGraph(graph, index);
if (getUniqueness(node))
continue;
for (marker = getMarker(node); marker != NULL_IDX;
marker = next) {
next = getNextInNode(marker);
if (!isInitial(marker) && !isTerminal(marker))
continue;
if (isTerminal(marker))
marker = getTwinMarker(marker);
while (!getUniqueness(getNode(marker))) {
if (next != NULL_IDX
&& (marker == next
|| marker == getTwinMarker(next)))
next = getNextInNode(next);
if (getNextInSequence(marker) != NULL_IDX) {
marker = getNextInSequence(marker);
destroyPassageMarker
(getPreviousInSequence
(marker));
} else {
destroyPassageMarker(marker);
break;
}
}
}
}
}
void readCoherentGraph(Graph * inGraph, boolean(*isUnique) (Node * node),
double coverage, ReadSet * reads)
{
IDnum nodeIndex;
Node *node;
IDnum previousNodeCount = 0;
graph = inGraph;
listMemory = newRecycleBin(sizeof(PassageMarkerList), 100000);
expected_coverage = coverage;
sequences = reads->tSequences;
velvetLog("Read coherency...\n");
resetNodeStatus(graph);
identifyUniqueNodes(isUnique);
trimLongReadTips();
previousNodeCount = 0;
while (previousNodeCount != nodeCount(graph)) {
previousNodeCount = nodeCount(graph);
for (nodeIndex = 1; nodeIndex <= nodeCount(graph);
nodeIndex++) {
node = getNodeInGraph(graph, nodeIndex);
if (node == NULL || !getUniqueness(node))
continue;
while (uniqueNodesConnect(node))
node = bypass();
node = getTwinNode(node);
while (uniqueNodesConnect(node))
node = bypass();
}
renumberNodes(graph);
}
destroyRecycleBin(listMemory);
destroyRecycleBin(nodeListMemory);
velvetLog("Confronted to %li multiple hits and %li null over %li\n",
(long) multCounter, (long) nullCounter, (long) dbgCounter);
velvetLog("Read coherency over!\n");
}
static void computeLocalNodeToNodeMappingsFromConnections(Connection *
connect,
Connection *
connect2)
{
Node *node1 = getTwinNode(getConnectionDestination(connect));
Node *node2 = getTwinNode(getConnectionDestination(connect2));
IDnum nodeID1 = getNodeID(node1);
IDnum nodeID2 = getNodeID(node2);
Coordinate distance =
getNodeLength(node1)/2 + getNodeLength(node2)/2;
Arc *arc;
if (getUniqueness(node1) || getUniqueness(node2))
return;
if ((arc = getArcBetweenNodes(node1, node2, graph))
&& !getConnectionBetweenNodes(node1, getTwinNode(node2))) {
createConnection(nodeID1, -nodeID2, getMultiplicity(arc),
0, distance,
1 / (double) getMultiplicity(arc));
incrementConnectionWeight(getConnectionBetweenNodes
(node1, getTwinNode(node2)),
getMultiplicity(arc));
}
if ((arc = getArcBetweenNodes(node2, node1, graph))
&& !getConnectionBetweenNodes(node2, getTwinNode(node1))) {
createConnection(nodeID2, -nodeID1, getMultiplicity(arc),
0, distance,
1 / (double) getMultiplicity(arc));
incrementConnectionWeight(getConnectionBetweenNodes
(node2, getTwinNode(node1)),
getMultiplicity(arc));
}
}
static void propagateComponent(Node * node)
{
Connection *connect;
if (getNodeStatus(node) || !getUniqueness(node))
return;
setNodeStatus(node, true);
for (connect = getConnection(node); connect != NULL;
connect = getNextConnection(connect))
propagateComponent(getConnectionDestination(connect));
for (connect = getConnection(getTwinNode(node)); connect != NULL;
connect = getNextConnection(connect))
propagateComponent(getConnectionDestination(connect));
}
static void computeLocalNodeToNodeMappings()
{
IDnum index;
Node *node;
puts("Computing local connections");
activateArcLookupTable(graph);
for (index = -nodeCount(graph); index <= nodeCount(graph); index++) {
node = getNodeInGraph(graph, index);
if (node && getUniqueness(node))
computeLocalNodeToNodeMappingsFromNode(node);
}
deactivateArcLookupTable(graph);
}
static boolean goesToNode(PassageMarkerI marker, Node * node)
{
PassageMarkerI current;
Node * start = getNode(marker);
for (current = getNextInSequence(marker); current != NULL_IDX;
current = getNextInSequence(current)) {
if (getNode(current) == node)
return true;
else if (getNode(current) == start)
continue;
else if (getUniqueness(getNode(current)))
return false;
}
return false;
}
static void fillUpComponent(Node * node)
{
Connection *connect;
if (getNodeStatus(node) || !getUniqueness(node))
return;
setSingleNodeStatus(node, true);
recordNode(node);
for (connect = getConnection(node); connect != NULL;
connect = getNextConnection(connect))
fillUpComponent(getTwinNode
(getConnectionDestination(connect)));
for (connect = getConnection(getTwinNode(node)); connect != NULL;
connect = getNextConnection(connect))
fillUpComponent(getConnectionDestination(connect));
}
static boolean comesFromNode(PassageMarkerI marker, Node * node)
{
Node *source = getNode(getTwinMarker(marker));
Node *target = getTwinNode(node);
PassageMarkerI current;
for (current = getNextInSequence(getTwinMarker(marker)); current != NULL_IDX;
current = getNextInSequence(current)) {
if (getNode(current) == target)
return true;
else if (getNode(current) == source)
continue;
else if (getUniqueness(getNode(current)))
return false;
}
return false;
}
static IDnum countConnectedComponents(Graph * graph)
{
IDnum index;
IDnum count = 0;
Node *node;
resetNodeStatus(graph);
for (index = 1; index <= nodeCount(graph); index++) {
node = getNodeInGraph(graph, index);
if (!getNodeStatus(node) && getUniqueness(node)) {
count++;
propagateComponent(node);
}
}
return count;
}
static Locus *extractConnectedComponents(IDnum locusCount)
{
Locus *loci = allocateLocusArray(locusCount);
Locus *locus;
IDnum index;
IDnum locusIndex = 0;
IDnum nodeIndex;
Node *node;
resetNodeStatus(graph);
for (index = 1; index <= nodeCount(graph); index++) {
node = getNodeInGraph(graph, index);
if (!getNodeStatus(node) && getUniqueness(node)) {
locus = getLocus(loci, locusIndex++);
clearLocus(locus);
// Long contigs
fillUpComponent(node);
setLongContigCount(locus, countMarkedNodes());
while (existsMarkedNode())
addContig(locus, popNodeRecord());
// Secondary contigs
extendComponent(locus);
setContigCount(locus, getLongContigCount(locus) + countMarkedNodes());
while (existsMarkedNode())
addContig(locus, popNodeRecord());
// Mark primary nodes so that their twins are not reused
for (nodeIndex = 0;
nodeIndex < getLongContigCount(locus);
nodeIndex++)
setNodeStatus(getContig(locus, nodeIndex), true);
// Unmark secondary nodes so that they are available to other loci
for (nodeIndex = getLongContigCount(locus);
nodeIndex < getContigCount(locus); nodeIndex++)
setNodeStatus(getContig(locus, nodeIndex), false);
}
}
return loci;
}
static void projectFromReadPair(Node * node, ReadOccurence * readOccurence,
Coordinate position, Coordinate offset,
Coordinate insertLength,
double insertVariance)
{
Coordinate distance = insertLength;
Coordinate variance = insertVariance;
Node *target = getNodeInGraph(graph, readOccurence->nodeID);
if (target == getTwinNode(node) || target == node)
return;
if (getUniqueness(target) && getNodeID(target) < getNodeID(node))
return;
if (position < 0) {
variance += getNodeLength(node) * getNodeLength(node) / 16;
// distance += 0;
} else {
// variance += 0;
distance += position - offset - getNodeLength(node) / 2;
}
if (readOccurence->position < 0) {
variance +=
getNodeLength(target) * getNodeLength(target) / 16;
//distance += 0;
} else {
// variance += 0;
distance +=
readOccurence->position - readOccurence->offset -
getNodeLength(target) / 2;
}
if (distance - getNodeLength(node)/2 - getNodeLength(target)/2 < -6 * sqrt(insertVariance))
return;
else if (distance < getNodeLength(node)/2 + getNodeLength(target)/2)
distance = getNodeLength(node)/2 + getNodeLength(target)/2;
createConnection(getNodeID(node), getNodeID(target), 0, 1,
distance, variance);
}
static boolean goesToNode(PassageMarkerI marker, Node * node)
{
PassageMarkerI current;
Node * start = getNode(marker);
Node * twinStart = getTwinNode(start);
Node * currentNode;
for (current = getNextInSequence(marker); current != NULL_IDX;
current = getNextInSequence(current)) {
currentNode = getNode(current);
if (currentNode == start || currentNode == twinStart)
return false;
else if (currentNode == node)
return true;
else if (getUniqueness(currentNode))
return false;
}
return false;
}
static boolean expandLongNodes(boolean force_jumps)
{
IDnum nodeID;
Node *node;
boolean modified = false;
for (nodeID = 1; nodeID <= nodeCount(graph); nodeID++) {
node = getNodeInGraph(graph, nodeID);
if (node != NULL && getUniqueness(node)) {
modified = expandLongNode(node, force_jumps)
|| modified;
modified =
expandLongNode(getTwinNode(node), force_jumps)
|| modified;
}
}
return modified;
}
static void updateMembers(Node * bypass, Node * nextNode)
{
PassageMarkerI marker, next, tmp;
Coordinate nextLength = getNodeLength(nextNode);
// Update marker + arc info
for (marker = getMarker(bypass); marker != NULL_IDX; marker = tmp) {
tmp = getNextInNode(marker);
if (!isTerminal(marker)
&& getNode(getNextInSequence(marker)) == nextNode) {
// Marker steps right into target
next = getNextInSequence(marker);
disconnectNextPassageMarker(marker, graph);
destroyPassageMarker(next);
} else if (getUniqueness(nextNode)
&& goesToNode(marker, nextNode)) {
// Marker goes indirectly to target
while (getNode(getNextInSequence(marker)) !=
nextNode) {
next = getNextInSequence(marker);
disconnectNextPassageMarker(marker, graph);
destroyPassageMarker(next);
}
next = getNextInSequence(marker);
disconnectNextPassageMarker(marker, graph);
destroyPassageMarker(next);
} else if (!isTerminal(marker)
&& getFinishOffset(marker) == 0) {
// Marker goes somewhere else than to target
next = getNextInSequence(marker);
incrementFinishOffset(marker, nextLength);
} else {
// Marker goes nowhere
incrementFinishOffset(marker, nextLength);
}
}
}
static boolean testConnection(IDnum IDA, Connection * connect,
IDnum ** counts)
{
IDnum total = 0;
Category cat;
// Spare unique -> undetermined node connections
if (!getUniqueness(connect->destination))
return true;
// Destroy tenuous connections
if (connect->paired_count + connect->direct_count <
UNRELIABLE_CONNECTION_CUTOFF)
return false;
for (cat = 0; cat <= CATEGORIES; cat++)
total +=
expectedNumberOfConnections(IDA, connect, counts, cat);
// Remove inconsistent connections
return connect->paired_count >= total / 10;
}
static void identifyUniqueNodes(boolean(*isUniqueFunction) (Node *))
{
IDnum index;
Node *node;
IDnum counter = 0;
velvetLog("Identifying unique nodes\n");
for (index = 1; index <= nodeCount(graph); index++) {
node = getNodeInGraph(graph, index);
if (node == NULL)
continue;
setUniqueness(node, isUniqueFunction(node));
if (getUniqueness(node))
counter++;
}
velvetLog("Done, %li unique nodes counted\n", (long) counter);
}
static Node *bypass()
{
Node *bypass = getNode(path);
Node *next = NULL;
Arc *arc;
PassageMarkerI nextMarker;
// Remove unwanted arcs
while (getArc(bypass) != NULL)
destroyArc(getArc(bypass), graph);
// Update extensive variables (length + descriptors + passage markers)
while (!isTerminal(path)) {
nextMarker = getNextInSequence(path);
next = getNode(nextMarker);
while (next == bypass) {
disconnectNextPassageMarker(path, graph);
destroyPassageMarker(nextMarker);
nextMarker = getNextInSequence(path);
next = getNode(nextMarker);
}
if (next == NULL)
return bypass;
// Overall node update
if (!getUniqueness(next)) {
adjustShortReads(bypass, getNextInSequence(path));
appendSequence(bypass, sequences,
getNextInSequence(path), graph);
} else {
concatenateReadStarts(bypass, next, graph);
#ifndef SINGLE_COV_CAT
Category cat;
for (cat = 0; cat < CATEGORIES; cat++) {
// Update virtual coverage
incrementVirtualCoverage(bypass, cat,
getVirtualCoverage(next, cat));
// Update original virtual coverage
incrementOriginalVirtualCoverage(bypass, cat,
getOriginalVirtualCoverage(next, cat));
}
#else
incrementVirtualCoverage(bypass, getVirtualCoverage(next));
#endif
appendDescriptors(bypass, next);
}
// Members
updateMembers(bypass, next);
// Termination
if (isTerminal(path) || getUniqueness(next))
break;
}
// Remove unique groupies from arrival
admitGroupies(next, bypass);
// Copy destination arcs
for (arc = getArc(next); arc != NULL; arc = getNextArc(arc)) {
if (getDestination(arc) == next)
continue;
else if (getDestination(arc) == getTwinNode(next))
createAnalogousArc(bypass, getTwinNode(bypass),
arc, graph);
else
createAnalogousArc(bypass, getDestination(arc),
arc, graph);
}
destroyNode(next, graph);
return bypass;
}
static boolean pushNeighbours(Node * node, Node * oppositeNode,
Coordinate distance, boolean force_jumps)
{
Node *candidate;
Coordinate oldLength = getNodeLength(node);
MiniConnection *localConnect;
NodeList *path, *tmp;
if ((path = pathIsClear(node, oppositeNode, distance))) {
while (path) {
candidate = path->node;
tmp = path->next;
deallocateNodeList(path);
path = tmp;
///////////////////////////////////////
// Stepping forward to destination //
///////////////////////////////////////
if (getUniqueness(candidate)) {
concatenateReadStarts(node, candidate,
graph);
concatenateLongReads(node, candidate,
graph);
absorbExtension(node, candidate);
// Scaffold changes
recenterNode(node, oldLength);
recenterLocalScaffold(node, oldLength);
absorbExtensionInScaffold(node, candidate);
// Read coverage
#ifndef SINGLE_COV_CAT
Category cat;
for (cat = 0; cat < CATEGORIES; cat++) {
incrementVirtualCoverage(node, cat,
getVirtualCoverage(candidate, cat));
incrementOriginalVirtualCoverage(node, cat,
getOriginalVirtualCoverage(candidate, cat));
}
#else
incrementVirtualCoverage(node, getVirtualCoverage(candidate));
#endif
if (getNodeStatus(candidate)) {
localConnect =
&localScaffold[getNodeID
(candidate) +
nodeCount
(graph)];
if (localConnect->frontReference) {
destroyConnection
(localConnect->
frontReference,
getNodeID(node));
localConnect->
frontReference = NULL;
}
if (localConnect->backReference) {
destroyConnection
(localConnect->
backReference,
-getNodeID(node));
localConnect->
backReference = NULL;
}
unmarkNode(candidate,
localConnect);
}
if (getNodeStatus(getTwinNode(candidate))) {
localConnect =
&localScaffold[-getNodeID
(candidate) +
nodeCount
(graph)];
if (localConnect->frontReference) {
destroyConnection
(localConnect->
frontReference,
getNodeID(node));
localConnect->
frontReference = NULL;
}
if (localConnect->backReference) {
destroyConnection
(localConnect->
backReference,
-getNodeID(node));
localConnect->
backReference = NULL;
}
unmarkNode(getTwinNode(candidate),
localConnect);
}
destroyNode(candidate, graph);
return true;
} else {
adjustShortReads(node, candidate);
adjustLongReads(node, getNodeLength(candidate));
absorbExtension(node, candidate);
}
}
}
if (force_jumps && oppositeNode
&& abs_ID(getNodeID(oppositeNode)) < abs_ID(getNodeID(node))) {
distance -= getNodeLength(node) / 2;
distance -= getNodeLength(oppositeNode) / 2;
if (distance > 10) {
adjustShortReadsByLength(node, distance);
adjustLongReads(node, distance);
appendGap(node, distance, graph);
} else {
adjustShortReadsByLength(node, 10);
adjustLongReads(node, 10);
appendGap(node, 10, graph);
}
concatenateReadStarts(node, oppositeNode, graph);
concatenateLongReads(node, oppositeNode, graph);
absorbExtension(node, oppositeNode);
// Scaffold changes
recenterNode(node, oldLength);
recenterLocalScaffold(node, oldLength);
absorbExtensionInScaffold(node, oppositeNode);
// Read coverage
#ifndef SINGLE_COV_CAT
Category cat;
for (cat = 0; cat < CATEGORIES; cat++)
incrementVirtualCoverage(node, cat,
getVirtualCoverage(oppositeNode, cat));
#else
incrementVirtualCoverage(node, getVirtualCoverage(oppositeNode));
#endif
if (getNodeStatus(oppositeNode)) {
localConnect =
&localScaffold[getNodeID(oppositeNode) +
nodeCount(graph)];
if (localConnect->frontReference) {
destroyConnection(localConnect->
frontReference,
getNodeID(node));
localConnect->frontReference = NULL;
}
if (localConnect->backReference) {
destroyConnection(localConnect->
backReference,
-getNodeID(node));
localConnect->backReference = NULL;
}
unmarkNode(oppositeNode, localConnect);
}
if (getNodeStatus(getTwinNode(oppositeNode))) {
localConnect =
&localScaffold[-getNodeID(oppositeNode) +
nodeCount(graph)];
if (localConnect->frontReference) {
destroyConnection(localConnect->
frontReference,
getNodeID(node));
localConnect->frontReference = NULL;
}
if (localConnect->backReference) {
destroyConnection(localConnect->
backReference,
-getNodeID(node));
localConnect->backReference = NULL;
}
unmarkNode(getTwinNode(oppositeNode),
localConnect);
}
destroyNode(oppositeNode, graph);
}
return false;
}
static NodeList *pathIsClear(Node * node, Node * oppositeNode,
Coordinate distance)
{
Arc *arc;
Node *candidate, *dest, *current;
Coordinate extension_distance = 0;
boolean maxRepeat = 1;
Node *repeatEntrance = NULL;
IDnum counter = 0;
NodeList *path = NULL;
NodeList *tail = path;
setSingleNodeStatus(node, 2);
current = node;
while (true) {
//////////////////////////////////
// Selecting destination //
//////////////////////////////////
candidate = NULL;
// First round for priority nodes
for (arc = getArc(current); arc != NULL;
arc = getNextArc(arc)) {
dest = getDestination(arc);
if (dest == node || dest == getTwinNode(node))
continue;
if (getNodeStatus(dest) <= 0)
continue;
if (candidate == NULL
|| getNodeStatus(candidate) >
getNodeStatus(dest)
|| (getNodeStatus(candidate) ==
getNodeStatus(dest)
&& extension_distance >
localScaffold[getNodeID(dest) +
nodeCount(graph)].
distance - getNodeLength(dest) / 2)) {
extension_distance =
localScaffold[getNodeID(dest) +
nodeCount(graph)].
distance - getNodeLength(dest) / 2;
candidate = dest;
}
}
// In case of failure
if (candidate == NULL) {
for (arc = getArc(current); arc != NULL;
arc = getNextArc(arc)) {
dest = getDestination(arc);
if (getNodeStatus(dest) == 0)
continue;
if (dest == node
|| dest == getTwinNode(node))
continue;
if (candidate == NULL
|| getNodeStatus(candidate) <
getNodeStatus(dest)
|| (getNodeStatus(candidate) ==
getNodeStatus(dest)
&& extension_distance <
localScaffold[getNodeID(dest) +
nodeCount(graph)].
distance -
getNodeLength(dest) / 2)) {
extension_distance =
localScaffold[getNodeID(dest) +
nodeCount
(graph)].
distance -
getNodeLength(dest) / 2;
candidate = dest;
}
}
}
if (candidate == NULL) {
while (path) {
tail = path->next;
deallocateNodeList(path);
path = tail;
}
return false;
}
// Loop detection
if (candidate == repeatEntrance
&& abs_bool(getNodeStatus(candidate)) ==
maxRepeat + 1) {
while (path) {
tail = path->next;
deallocateNodeList(path);
path = tail;
}
return false;
} else if (abs_bool(getNodeStatus(candidate)) > maxRepeat) {
maxRepeat = abs_bool(getNodeStatus(candidate));
repeatEntrance = candidate;
} else if (abs_bool(getNodeStatus(candidate)) == 1) {
maxRepeat = 1;
repeatEntrance = NULL;
}
if (getNodeStatus(candidate) > 0)
setSingleNodeStatus(candidate,
getNodeStatus(candidate) + 1);
else
setSingleNodeStatus(candidate,
getNodeStatus(candidate) - 1);
if (abs_bool(getNodeStatus(candidate)) > 100
|| counter > nodeCount(graph)) {
while (path) {
tail = path->next;
deallocateNodeList(path);
path = tail;
}
return false;
}
// Missassembly detection
if (getUniqueness(candidate) && oppositeNode
&& candidate != oppositeNode
&& extension_distance > distance) {
while (path) {
tail = path->next;
deallocateNodeList(path);
path = tail;
}
return false;
}
if (path == NULL) {
path = allocateNodeList();
path->next = NULL;
path->node = candidate;
tail = path;
} else {
tail->next = allocateNodeList();
tail = tail->next;
tail->node = candidate;
tail->next = NULL;
}
if (getUniqueness(candidate))
return path;
current = candidate;
}
}
static void projectFromSingleRead(Node * node,
ReadOccurence * readOccurence,
Coordinate position,
Coordinate offset, Coordinate length)
{
Coordinate distance = 0;
Node *target = getNodeInGraph(graph, -readOccurence->nodeID);
double variance = 1;
if (target == getTwinNode(node) || target == node)
return;
if (getUniqueness(target) && getNodeID(target) < getNodeID(node))
return;
if (position < 0) {
variance += getNodeLength(node) * getNodeLength(node) / 16;
// distance += 0;
} else {
// variance += 0;
distance += position - getNodeLength(node) / 2;
}
if (readOccurence->position < 0) {
variance +=
getNodeLength(target) * getNodeLength(target) / 16;
//distance += 0;
} else {
// variance += 0;
distance +=
-readOccurence->position + getNodeLength(target) / 2;
}
if (readOccurence->offset < 0 || offset < 0) {
variance += length * length / 16;
//distance += 0;
} else {
// variance += 0;
distance += readOccurence->offset - offset;
}
// Relative ordering
if (offset > 0 && readOccurence->offset > 0) {
if (offset < readOccurence->offset) {
if (distance - getNodeLength(node)/2 - getNodeLength(target)/2 < -10)
;
else if (distance < getNodeLength(node)/2 + getNodeLength(target)/2)
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
else
createConnection(getNodeID(node), getNodeID(target), 1, 0,
distance, variance);
} else if (offset > readOccurence->offset) {
if (-distance - getNodeLength(node)/2 - getNodeLength(target)/2 < -10)
;
else if (-distance < getNodeLength(node)/2 + getNodeLength(target)/2)
createConnection(-getNodeID(node), -getNodeID(target), 1,
0, getNodeLength(node)/2 + getNodeLength(target)/2 , variance);
else
createConnection(-getNodeID(node), -getNodeID(target), 1,
0, -distance, variance);
}
} else if (offset > 0 && position > 0) {
if (distance - offset > -getNodeLength(node)/2 && distance - offset + length > getNodeLength(node)/2)
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
else if (distance - offset < -getNodeLength(node)/2 && distance - offset + length < getNodeLength(node)/2)
createConnection(-getNodeID(node), -getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
else {
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
createConnection(-getNodeID(node), -getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
}
} else if (readOccurence->offset > 0 && readOccurence->position > 0) {
if (-distance - readOccurence->offset > -getNodeLength(target)/2 && -distance - readOccurence->offset + length > getNodeLength(target)/2)
createConnection(-getNodeID(node), -getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
if (-distance - readOccurence->offset < -getNodeLength(target)/2 && -distance - readOccurence->offset + length < getNodeLength(target)/2)
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
else {
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
createConnection(-getNodeID(node), -getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
}
} else {
createConnection(getNodeID(node), getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
createConnection(-getNodeID(node), -getNodeID(target), 1, 0,
getNodeLength(node)/2 + getNodeLength(target)/2, variance);
}
}
static void projectFromReadPair(Node * node, ReadOccurence * readOccurence,
Coordinate position, Coordinate offset,
Coordinate insertLength,
double insertVariance, boolean weight)
{
Coordinate distance = insertLength;
Coordinate variance = insertVariance;
Node *target = getNodeInGraph(graph, readOccurence->nodeID);
Connection *connect;
double score;
// Filter for useless reads:
if (readOccurence->position == -1 && readOccurence->offset == -1)
return;
if (target == getTwinNode(node) || target == node)
return;
if (getUniqueness(target) && getNodeID(target) < getNodeID(node))
return;
if (weight) {
if (position > 0 && readOccurence->position > 0
&& (connect =
getConnectionBetweenNodes(node, target))) {
distance = getConnectionDistance(connect);
distance -=
position - offset - getNodeLength(node) / 2;
distance -=
readOccurence->position -
readOccurence->offset -
getNodeLength(target) / 2;
score =
K *
exp((insertLength - distance) * (distance -
insertLength)
/ (2 * insertVariance));
incrementConnectionWeight(connect, score);
}
return;
}
if (position < 0) {
variance += getNodeLength(node) * getNodeLength(node) / 16;
// distance += 0;
} else {
// variance += 0;
distance += position - offset - getNodeLength(node) / 2;
}
if (readOccurence->position < 0) {
variance +=
getNodeLength(target) * getNodeLength(target) / 16;
//distance += 0;
} else {
// variance += 0;
distance +=
readOccurence->position - readOccurence->offset -
getNodeLength(target) / 2;
}
if (distance - getNodeLength(node) / 2 -
getNodeLength(target) / 2 < -6 * sqrt(insertVariance))
return;
createConnection(getNodeID(node), getNodeID(target), 0, 1,
distance, variance);
}
// Replaces two consecutive nodes into a single equivalent node
// The extra memory is freed
void concatenateStringOfNodes(Node * nodeA, Graph * graph)
{
Node *twinA = getTwinNode(nodeA);
Node * nodeB = nodeA;
Node * twinB;
Node *currentNode, *nextNode;
Coordinate totalLength = 0;
PassageMarkerI marker, tmpMarker;
Arc *arc;
Category cat;
while (simpleArcCount(nodeB) == 1
&&
simpleArcCount(getTwinNode
(getDestination(getArc(nodeB)))) ==
1
&& getDestination(getArc(nodeB)) != getTwinNode(nodeB)
&& getDestination(getArc(nodeB)) != nodeA) {
totalLength += getNodeLength(nodeB);
nodeB = getDestination(getArc(nodeB));
}
twinB = getTwinNode(nodeB);
totalLength += getNodeLength(nodeB);
reallocateNodeDescriptor(nodeA, totalLength);
currentNode = nodeA;
while (currentNode != nodeB) {
currentNode = getDestination(getArc(currentNode));
// Passage marker management in node A:
for (marker = getMarker(nodeA); marker != NULL_IDX;
marker = getNextInNode(marker))
if (getNode(getNextInSequence(marker)) != currentNode)
incrementFinishOffset(marker,
getNodeLength(currentNode));
// Swapping new born passageMarkers from B to A
for (marker = getMarker(currentNode); marker != NULL_IDX; marker = tmpMarker) {
tmpMarker = getNextInNode(marker);
if (isInitial(marker)
|| getNode(getPreviousInSequence(marker)) != nodeA) {
extractPassageMarker(marker);
transposePassageMarker(marker, nodeA);
incrementFinishOffset(getTwinMarker(marker),
getNodeLength(nodeA));
} else
disconnectNextPassageMarker(getPreviousInSequence
(marker), graph);
}
// Read starts
concatenateReadStarts(nodeA, currentNode, graph);
// Gaps
appendNodeGaps(nodeA, currentNode, graph);
// Update uniqueness:
setUniqueness(nodeA, getUniqueness(nodeA) || getUniqueness(currentNode));
// Update virtual coverage
for (cat = 0; cat < CATEGORIES; cat++)
incrementVirtualCoverage(nodeA, cat,
getVirtualCoverage(currentNode, cat));
// Update original virtual coverage
for (cat = 0; cat < CATEGORIES; cat++)
incrementOriginalVirtualCoverage(nodeA, cat,
getOriginalVirtualCoverage
(currentNode, cat));
// Descriptor management (node)
directlyAppendDescriptors(nodeA, currentNode, totalLength);
}
// Correct arcs
for (arc = getArc(nodeB); arc != NULL; arc = getNextArc(arc)) {
if (getDestination(arc) != twinB)
createAnalogousArc(nodeA, getDestination(arc),
arc, graph);
else
createAnalogousArc(nodeA, twinA, arc, graph);
}
// Freeing gobbled nodes
currentNode = getTwinNode(nodeB);
while (currentNode != getTwinNode(nodeA)) {
arc = getArc(currentNode);
nextNode = getDestination(arc);
destroyNode(currentNode, graph);
currentNode = nextNode;
}
}
static void integrateDerivativeDistances(Connection * connect,
Coordinate min_distance,
boolean direction)
{
Node *reference = getConnectionDestination(connect);
Node *destination;
IDnum destinationID;
Coordinate distance, baseDistance;
double variance, baseVariance;
Connection *connect2;
MiniConnection *localConnect;
// debug
IDnum counter = 0;
if (!getUniqueness(reference))
return;
//velvetLog("Opposite node %li length %li at %li ± %f\n", getNodeID(reference), getNodeLength(reference), getConnectionDistance(connect), getConnectionVariance(connect));
baseDistance = getConnectionDistance(connect);
baseVariance = getConnectionVariance(connect);
for (connect2 = getConnection(reference);
connect2 != NULL; connect2 = getNextConnection(connect2)) {
// Avoid null derivative
if (connect2 == getTwinConnection(connect))
continue;
destination = getConnectionDestination(connect2);
// Beware of directionality
if (!direction)
destination = getTwinNode(destination);
// Derivate values
destinationID = getNodeID(destination);
// Beware of directionality (bis)
if (direction)
distance = baseDistance - getConnectionDistance(connect2);
else
distance = getConnectionDistance(connect2) - baseDistance;
variance = getConnectionVariance(connect2) + baseVariance;
localConnect =
&localScaffold[destinationID + nodeCount(graph)];
// Avoid over-projection
if (distance < min_distance) {
//velvetLog("Node %li not at distance %li± %f (min %li)\n", destinationID, distance, variance, min_distance);
continue;
}
counter++;
if (getNodeStatus(destination)) {
readjustMiniConnection(destination, localConnect,
distance, min_distance,
variance, NULL, NULL);
} else
resetMiniConnection(destination, localConnect,
distance, variance, NULL, NULL,
true);
//velvetLog("Node %li now at distance %li\n", destinationID, localConnect->distance);
}
//velvetLog("%li secondary distances added\n", counter);
}
// Replaces two consecutive nodes into a single equivalent node
// The extra memory is freed
void concatenateNodes(Node * nodeA, Node * nodeB, Graph * graph)
{
PassageMarkerI marker, tmpMarker;
Node *twinA = getTwinNode(nodeA);
Node *twinB = getTwinNode(nodeB);
Arc *arc;
Category cat;
// Arc management:
// Freeing useless arcs
while (getArc(nodeA) != NULL)
destroyArc(getArc(nodeA), graph);
// Correct arcs
for (arc = getArc(nodeB); arc != NULL; arc = getNextArc(arc)) {
if (getDestination(arc) != twinB)
createAnalogousArc(nodeA, getDestination(arc),
arc, graph);
else
createAnalogousArc(nodeA, twinA, arc, graph);
}
// Passage marker management in node A:
for (marker = getMarker(nodeA); marker != NULL_IDX;
marker = getNextInNode(marker))
if (isTerminal(marker))
incrementFinishOffset(marker,
getNodeLength(nodeB));
// Swapping new born passageMarkers from B to A
for (marker = getMarker(nodeB); marker != NULL_IDX; marker = tmpMarker) {
tmpMarker = getNextInNode(marker);
if (isInitial(marker)
|| getNode(getPreviousInSequence(marker)) != nodeA) {
extractPassageMarker(marker);
transposePassageMarker(marker, nodeA);
incrementFinishOffset(getTwinMarker(marker),
getNodeLength(nodeA));
} else
disconnectNextPassageMarker(getPreviousInSequence
(marker), graph);
}
// Read starts
concatenateReadStarts(nodeA, nodeB, graph);
// Gaps
appendNodeGaps(nodeA, nodeB, graph);
// Descriptor management (node)
appendDescriptors(nodeA, nodeB);
// Update uniqueness:
setUniqueness(nodeA, getUniqueness(nodeA) || getUniqueness(nodeB));
// Update virtual coverage
for (cat = 0; cat < CATEGORIES; cat++)
incrementVirtualCoverage(nodeA, cat,
getVirtualCoverage(nodeB, cat));
// Update original virtual coverage
for (cat = 0; cat < CATEGORIES; cat++)
incrementOriginalVirtualCoverage(nodeA, cat,
getOriginalVirtualCoverage
(nodeB, cat));
// Freeing gobbled node
destroyNode(nodeB, graph);
}
static boolean uniqueNodesConnect(Node * startingNode)
{
Node *destination = NULL;
PassageMarkerI startMarker, currentMarker;
RBConnection *newList;
RBConnection *list = NULL;
boolean multipleHits = false;
if (arcCount(startingNode) == 0)
return false;
if (getMarker(startingNode) == NULL_IDX)
return false;
dbgCounter++;
// Checking for multiple destinations
for (startMarker = getMarker(startingNode); startMarker != NULL_IDX;
startMarker = getNextInNode(startMarker)) {
if (getFinishOffset(startMarker) >
2 * getWordLength(graph))
continue;
for (currentMarker = getNextInSequence(startMarker);
currentMarker != NULL_IDX;
currentMarker = getNextInSequence(currentMarker)) {
if (!getUniqueness(getNode(currentMarker))) {
continue;
} else if (getNodeStatus(getNode(currentMarker))) {
if (getStartOffset(currentMarker) >
2 * getWordLength(graph))
break;
for (newList = list; newList != NULL;
newList = newList->next) {
if (newList->node ==
getNode(currentMarker)) {
newList->multiplicity++;
break;
}
}
if (newList == NULL)
abort();
break;
} else {
if (getStartOffset(currentMarker) >
2 * getWordLength(graph))
break;
setSingleNodeStatus(getNode(currentMarker),
true);
newList = allocateRBConnection();
newList->node = getNode(currentMarker);
newList->multiplicity = 1;
newList->marker = startMarker;
newList->next = list;
list = newList;
break;
}
}
}
while (list != NULL) {
newList = list;
list = newList->next;
setSingleNodeStatus(newList->node, false);
if (newList->multiplicity >= MULTIPLICITY_CUTOFF) {
if (destination == NULL) {
destination = newList->node;
path = newList->marker;
} else if (destination != newList->node)
multipleHits = true;
}
deallocateRBConnection(newList);
}
if (multipleHits) {
multCounter++;
setUniqueness(startingNode, false);
return false;
}
if (destination == NULL || destination == startingNode
|| destination == getTwinNode(startingNode)) {
nullCounter++;
return false;
}
// Check for reciprocity
for (startMarker = getMarker(getTwinNode(destination));
startMarker != NULL_IDX;
startMarker = getNextInNode(startMarker)) {
if (getFinishOffset(startMarker) >
2 * getWordLength(graph))
continue;
for (currentMarker = getNextInSequence(startMarker);
currentMarker != NULL_IDX;
currentMarker = getNextInSequence(currentMarker)) {
if (!getUniqueness(getNode(currentMarker))) {
continue;
} else if (getNodeStatus(getNode(currentMarker))) {
if (getStartOffset(currentMarker) >
2 * getWordLength(graph))
break;
for (newList = list; newList != NULL;
newList = newList->next) {
if (newList->node ==
getNode(currentMarker)) {
newList->multiplicity++;
break;
}
}
if (newList == NULL)
abort();
break;
} else {
if (getStartOffset(currentMarker) >
2 * getWordLength(graph))
break;
setSingleNodeStatus(getNode(currentMarker),
true);
newList = allocateRBConnection();
newList->node = getNode(currentMarker);
newList->multiplicity = 1;
newList->next = list;
list = newList;
break;
}
}
}
while (list != NULL) {
newList = list;
list = newList->next;
setSingleNodeStatus(newList->node, false);
if (newList->multiplicity >= MULTIPLICITY_CUTOFF
&& newList->node != getTwinNode(startingNode))
multipleHits = true;
deallocateRBConnection(newList);
}
if (multipleHits) {
multCounter++;
setUniqueness(destination, false);
return false;
}
// Aligning long reads to each other:
// TODO
// Merge pairwise alignments and produce consensus
// TODO
return true;
}