/*
* Recursive function which fills a givenlist with the
* connected nodes within a module
*/
static void buildFaces_fillTopNodeList(Cap * cap, stList *list,
stHash *liftedEdgesTable) {
stList *liftedEdges;
int64_t index;
// Limit of recursion
if (stList_contains(list, cap))
return;
// Actual filling
st_logInfo("Adding cap %p to face\n", cap);
stList_append(list, cap);
// Recursion through lifted edges
if ((liftedEdges = stHash_search(liftedEdgesTable, cap)))
for (index = 0; index < stList_length(liftedEdges); index++)
buildFaces_fillTopNodeList(
((LiftedEdge *) stList_get(liftedEdges, index))->destination,
list, liftedEdgesTable);
// Recursion through adjacency
if (cap_getAdjacency(cap))
buildFaces_fillTopNodeList(cap_getAdjacency(cap),list,
liftedEdgesTable);
}
bool getCapGetAtEndOfPath(Cap *cap, Cap **pathEndCap,
int64_t *pathLength, int64_t *nCount, stList *haplotypeEventStrings, stList *contaminationEventStrings) {
//Account for length of adjacency
*pathLength += getTerminalAdjacencyLength(cap);
*nCount += getNumberOfNsInAdjacency(cap);
Segment *segment = getAdjacentCapsSegment(cap);
if (segment == NULL) {
*pathEndCap = cap_getAdjacency(getTerminalCap(cap));
assert(*pathEndCap != NULL);
return 0;
}
Cap *adjacentCap = cap_getSide(cap) ? segment_get3Cap(segment)
: segment_get5Cap(segment);
assert(
cap_getName(adjacentCap) == cap_getName(
cap_getAdjacency(getTerminalCap(cap))));
End *adjacentEnd = cap_getEnd(adjacentCap);
if (hasCapInEvents(adjacentEnd, contaminationEventStrings) || hasCapInEvents(adjacentEnd, haplotypeEventStrings)) { //hasCapNotInEvent(adjacentEnd, event_getHeader(cap_getEvent(cap)))) { //isContaminationEnd(adjacentEnd) || isHaplotypeEnd(adjacentEnd)) {
*pathEndCap = adjacentCap;
return 1;
}
*pathLength += segment_getLength(segment);
*nCount += getNumberOfNsInSegment(segment);
return getCapGetAtEndOfPath(cap_getOtherSegmentCap(adjacentCap),
pathEndCap, pathLength, nCount, haplotypeEventStrings, contaminationEventStrings);
}
bool trueAdjacency(Cap *cap, stList *eventStrings) {
if(getTerminalAdjacencyLength(cap) > 0) {
return 0;
}
cap = getTerminalCap(cap);
assert(cap != NULL);
Cap *otherCap = cap_getAdjacency(cap);
assert(otherCap != NULL);
assert(cap_getAdjacency(otherCap) == cap);
//So is the adjacency present in one of the haplotypes? That's what we're going to answer..
End *otherEnd = end_getPositiveOrientation(cap_getEnd(otherCap));
End_InstanceIterator *endInstanceIt = end_getInstanceIterator(cap_getEnd(cap));
Cap *cap2;
while ((cap2 = end_getNext(endInstanceIt)) != NULL) {
Cap *otherCap2 = cap_getAdjacency(cap2);
assert(otherCap2 != NULL);
if (otherEnd == end_getPositiveOrientation(cap_getEnd(otherCap2))) {
//const char *eventName = event_getHeader(cap_getEvent(cap2));
assert(event_getHeader(cap_getEvent(cap2)) == event_getHeader(
cap_getEvent(otherCap2)));
if (capHasGivenEvents(cap2, eventStrings)) { //strcmp(eventName, "hapA1") == 0 || strcmp(eventName, "hapA2") == 0) {
if(getTerminalAdjacencyLength(cap2) == 0) {
end_destructInstanceIterator(endInstanceIt);
return 1;
}
}
}
}
end_destructInstanceIterator(endInstanceIt);
return 0;
}
/*
* Constructs a face from a given Cap
*/
static void buildFaces_constructFromCap(Cap * startingCap,
stHash *liftedEdgesTable, Flower * flower) {
Face *face = face_construct(flower);
stList *topNodes = stList_construct3(16, NULL);
stList *liftedEdges;
Cap *cap, *bottomNode, *ancestor;
int64_t index, index2;
printf("Constructing new face");
// Establishlist of top nodes
buildFaces_fillTopNodeList(startingCap, topNodes, liftedEdgesTable);
#ifndef NDEBUG
// What, no top nodes!?
if (stList_length(topNodes) == 0)
abort();
#endif
// Initialize data structure
face_allocateSpace(face, stList_length(topNodes));
// For every top node
for (index = 0; index < stList_length(topNodes); index++) {
cap = stList_get(topNodes, index);
face_setTopNode(face, index, cap);
liftedEdges = stHash_search(liftedEdgesTable, cap);
if (!liftedEdges) {
face_setBottomNodeNumber(face, index, 0);
continue;
}
face_setBottomNodeNumber(face, index, stList_length(liftedEdges));
// For every bottom node of that top node
for (index2 = 0; index2 < stList_length(liftedEdges); index2++) {
bottomNode
= ((LiftedEdge *) stList_get(liftedEdges, index2))->bottomNode;
face_addBottomNode(face, index, bottomNode);
ancestor = cap_getTopCap(cap_getPositiveOrientation(
cap_getAdjacency(bottomNode)));
if (cap_getAdjacency(cap) != ancestor)
face_setDerivedDestination(face, index, index2, ancestor);
else
face_setDerivedDestination(face, index, index2, NULL);
#ifndef NDEBUG
// If bottom nodes part of top nodes
assert(!stList_contains(topNodes, cap_getPositiveOrientation(
((LiftedEdge*) stList_get(liftedEdges, index2))->bottomNode)));
#endif
}
}
// Clean up
stList_destruct(topNodes);
}
static int64_t setCoordinates(Flower *flower, MetaSequence *metaSequence, Cap *cap, int64_t coordinate) {
/*
* Sets the coordinates of the reference thread and sets the bases of the actual sequence
* that of the consensus.
*/
Sequence *sequence = flower_getSequence(flower, metaSequence_getName(metaSequence));
if (sequence == NULL) {
sequence = sequence_construct(metaSequence, flower);
}
while (1) {
assert(cap_getStrand(cap));
assert(!cap_getSide(cap));
Cap *adjacentCap = cap_getAdjacency(cap);
assert(adjacentCap != NULL);
assert(cap_getStrand(adjacentCap));
assert(cap_getSide(adjacentCap));
int64_t adjacencyLength = cap_getCoordinate(cap);
assert(adjacencyLength == cap_getCoordinate(adjacentCap));
assert(adjacencyLength != INT64_MAX);
assert(adjacencyLength >= 0);
cap_setCoordinates(cap, coordinate, 1, sequence);
coordinate += adjacencyLength + 1;
cap_setCoordinates(adjacentCap, coordinate, 1, sequence);
//Traverse any block..
if ((cap = cap_getOtherSegmentCap(adjacentCap)) == NULL) {
break;
}
coordinate += segment_getLength(cap_getSegment(adjacentCap)) - 1;
}
return coordinate;
}
static enum CapCode getHaplotypeSwitchCode(Cap *cap, stList *eventStrings) {
Cap *adjacentCap = cap_getAdjacency(getTerminalCap(cap));
assert(adjacentCap != NULL);
End *end = cap_getEnd(cap);
End *adjacentEnd = cap_getEnd(adjacentCap);
stSortedSet *eventStringsForEnd1 = getEventStrings(end, eventStrings);
stSortedSet *eventStringsForEnd2 = getEventStrings(adjacentEnd, eventStrings);
assert(stSortedSet_size(eventStringsForEnd1) > 0);
assert(stSortedSet_size(eventStringsForEnd2) > 0);
stSortedSet *intersectionOfEventStrings = stSortedSet_getIntersection(
eventStringsForEnd1, eventStringsForEnd2);
enum CapCode code1 = (stSortedSet_size(intersectionOfEventStrings)
!= stSortedSet_size(eventStringsForEnd1) || stSortedSet_size(
intersectionOfEventStrings)
!= stSortedSet_size(eventStringsForEnd2)) ? HAP_SWITCH
: HAP_NOTHING;
stSortedSet_destruct(eventStringsForEnd1);
stSortedSet_destruct(eventStringsForEnd2);
stSortedSet_destruct(intersectionOfEventStrings);
return code1;
}
static int64_t traceThreadLength(Cap *cap, Cap **terminatingCap) {
/*
* Gets the length in bases of the thread in the flower, starting from a given attached stub cap.
* The thread length includes the lengths of adjacencies that it contains.
*
* Terminating cap is initialised with the final cap on the thread from cap.
*/
assert(end_isStubEnd(cap_getEnd(cap)));
int64_t threadLength = 0;
while (1) {
assert(cap_getCoordinate(cap) != INT64_MAX);
int64_t adjacencyLength = cap_getCoordinate(cap);
threadLength += adjacencyLength;
Cap *adjacentCap = cap_getAdjacency(cap);
assert(adjacentCap != NULL);
assert(adjacencyLength == cap_getCoordinate(adjacentCap));
//Traverse any block..
if (cap_getSegment(adjacentCap) != NULL) {
threadLength += segment_getLength(cap_getSegment(adjacentCap));
cap = cap_getOtherSegmentCap(adjacentCap);
assert(cap != NULL);
} else {
assert(end_isStubEnd(cap_getEnd(adjacentCap)));
*terminatingCap = adjacentCap;
return threadLength;
}
}
return 1;
}
/*
* Fill uplist with bottom nodes for top node
*/
static void buildFaces_computeLiftedEdgesAtTopNode(Cap * cap, stList * liftedEdges) {
int64_t childIndex;
LiftedEdge * liftedEdge;
// Termination
if (cap_getAdjacency(cap)) {
liftedEdge = st_malloc(sizeof(LiftedEdge));
liftedEdge->bottomNode = cap_getPositiveOrientation(cap);
liftedEdge->destination = cap_getPositiveOrientation(cap_getTopCap(cap_getAdjacency(cap)));
stList_append(liftedEdges, liftedEdge);
return;
}
// Recursion through children
for (childIndex = 0; childIndex < cap_getChildNumber(cap); childIndex++)
buildFaces_computeLiftedEdgesAtTopNode(cap_getChild(cap, childIndex), liftedEdges);
}
/*
* Fill in a hashtable which to every node associates
* alist of lifted edges
*/
static stHash *buildFaces_computeLiftedEdges(Flower * flower) {
stHash *liftedEdgesTable = stHash_construct3(buildFaces_hashfunction,
buildFaces_key_eq_fn, NULL, buildFaces_destructValue);
Flower_CapIterator *iter = flower_getCapIterator(flower);
Cap *cap, *attachedAncestor;
Cap *adjacency, *adjacencyAncestor;
stList *liftedEdges;
LiftedEdge *liftedEdge;
// Iterate through potential bottom nodes
while ((cap = flower_getNextCap(iter))) {
// ... check if connected
if ((adjacency = cap_getAdjacency(cap))) {
// ... lift
attachedAncestor = cap_getTopCap(cap);
adjacencyAncestor = cap_getTopCap(cap_getPositiveOrientation(
adjacency));
#ifndef NDEBUG
assert((attachedAncestor && adjacencyAncestor) || (!attachedAncestor && !adjacencyAncestor));
#endif
// If root node
if (attachedAncestor == NULL)
continue;
// ... create lifted edge
liftedEdge = st_malloc(sizeof(LiftedEdge));
liftedEdge->destination = adjacencyAncestor;
liftedEdge->bottomNode = cap;
#ifndef NDEBUG
// Self loop
if (adjacencyAncestor == attachedAncestor)
abort();
#endif
// ... add it to the hashtable
if ((liftedEdges
= stHash_search(liftedEdgesTable, attachedAncestor))) {
stList_append(liftedEdges, liftedEdge);
} else {
liftedEdges = stList_construct3(2,
buildFaces_stList_destructElem);
stList_append(liftedEdges, liftedEdge);
stHash_insert(liftedEdgesTable, attachedAncestor, liftedEdges);
}
}
}
flower_destructCapIterator(iter);
return liftedEdgesTable;
}
static void setAdjacencyLengthsAndRecoverNewCapsAndBrokenAdjacencies(Cap *cap, stList *recoveredCaps) {
/*
* Sets the coordinates of the caps to be equal to the length of the adjacency sequence between them.
* Used to build the reference sequence bottom up.
*
* One complexity is that a reference thread between the two caps
* in each flower f may be broken into two in the children of f.
* Therefore, for each flower f first identify attached stub ends present in the children of f that are
* not present in f and copy them into f, reattaching the reference caps as needed.
*/
while (1) {
Cap *adjacentCap = cap_getAdjacency(cap);
assert(adjacentCap != NULL);
assert(cap_getCoordinate(cap) == INT64_MAX);
assert(cap_getCoordinate(adjacentCap) == INT64_MAX);
assert(cap_getStrand(cap) == cap_getStrand(adjacentCap));
assert(cap_getSide(cap) != cap_getSide(adjacentCap));
Group *group = end_getGroup(cap_getEnd(cap));
assert(group != NULL);
if (!group_isLeaf(group)) { //Adjacency is not terminal, so establish its sequence.
Flower *nestedFlower = group_getNestedFlower(group);
Cap *nestedCap = flower_getCap(nestedFlower, cap_getName(cap));
assert(nestedCap != NULL);
Cap *nestedAdjacentCap = flower_getCap(nestedFlower, cap_getName(adjacentCap));
assert(nestedAdjacentCap != NULL);
Cap *breakerCap;
int64_t adjacencyLength = traceThreadLength(nestedCap, &breakerCap);
assert(cap_getOrientation(nestedAdjacentCap));
if (cap_getPositiveOrientation(breakerCap) != nestedAdjacentCap) { //The thread is broken at the lower level.
//Copy cap into higher level graph.
breakerCap = copyCapToParent(breakerCap, recoveredCaps);
assert(cap_getSide(breakerCap));
cap_makeAdjacent(cap, breakerCap);
setAdjacencyLength(cap, breakerCap, adjacencyLength);
adjacencyLength = traceThreadLength(nestedAdjacentCap, &breakerCap);
assert(cap_getPositiveOrientation(breakerCap) != cap);
breakerCap = copyCapToParent(breakerCap, recoveredCaps);
assert(!cap_getSide(breakerCap));
cap_makeAdjacent(breakerCap, adjacentCap);
setAdjacencyLength(adjacentCap, breakerCap, adjacencyLength);
} else { //The thread is not broken at the lower level
setAdjacencyLength(cap, adjacentCap, adjacencyLength);
}
} else {
//Set the coordinates of the caps to the adjacency size
setAdjacencyLength(cap, adjacentCap, 0);
}
if ((cap = cap_getOtherSegmentCap(adjacentCap)) == NULL) {
break;
}
}
}
/*
* Recursive function which fills a givenlist with the
* connected nodes within a module and fills their lifted
* edges in the same pass
*/
static void buildFaces_fillTopNodeList2(Cap * cap, stList *list,
stHash *liftedEdgesTable) {
stList *liftedEdges = stList_construct3(2,
buildFaces_stList_destructElem);
int64_t index;
// Orientation check
cap = cap_getPositiveOrientation(cap);
// Limit of recursion
if (stList_contains(list, cap))
return;
// Actual filling
st_logInfo("Adding cap %p to face\n", cap);
stList_append(list, cap);
// Compute lifted edges
for (index = 0; index < cap_getChildNumber(cap); index++)
buildFaces_computeLiftedEdgesAtTopNode(cap_getChild(cap, index), liftedEdges);
// If emptylist...
if (stList_length(liftedEdges) == 0)
stList_destruct(liftedEdges);
// Recursion through lifted edges
else {
stHash_insert(liftedEdgesTable, cap, liftedEdges);
for (index = 0; index < stList_length(liftedEdges); index++)
buildFaces_fillTopNodeList2(
((LiftedEdge *) stList_get(liftedEdges, index))->destination,
list, liftedEdgesTable);
}
// Recursion through adjacency
if (cap_getAdjacency(cap))
buildFaces_fillTopNodeList2(cap_getAdjacency(cap),list,
liftedEdgesTable);
}
int64_t flower_getTotalBaseLength(Flower *flower) {
/*
* The implementation of this function is very like that in group_getTotalBaseLength, with a few differences. Consider merging them.
*/
Flower_EndIterator *endIterator = flower_getEndIterator(flower);
End *end;
int64_t totalLength = 0;
while ((end = flower_getNextEnd(endIterator)) != NULL) {
if (!end_isBlockEnd(end)) {
End_InstanceIterator *instanceIterator = end_getInstanceIterator(end);
Cap *cap;
while ((cap = end_getNext(instanceIterator)) != NULL) {
cap = cap_getStrand(cap) ? cap : cap_getReverse(cap);
if (!cap_getSide(cap) && cap_getSequence(cap) != NULL) {
Cap *cap2 = cap_getAdjacency(cap);
assert(cap2 != NULL);
while (end_isBlockEnd(cap_getEnd(cap2))) {
Segment *segment = cap_getSegment(cap2);
assert(segment != NULL);
assert(segment_get5Cap(segment) == cap2);
cap2 = cap_getAdjacency(segment_get3Cap(segment));
assert(cap2 != NULL);
assert(cap_getStrand(cap2));
assert(cap_getSide(cap2));
}
assert(cap_getStrand(cap2));
assert(cap_getSide(cap2));
int64_t length = cap_getCoordinate(cap2) - cap_getCoordinate(cap) - 1;
assert(length >= 0);
totalLength += length;
}
}
end_destructInstanceIterator(instanceIterator);
}
}
flower_destructEndIterator(endIterator);
return totalLength;
}
Segment *block_splitP(Segment *segment,
Block *leftBlock, Block *rightBlock) {
Segment *leftSegment = segment_getSequence(segment) != NULL
? segment_construct2(leftBlock, segment_getStart(segment), segment_getStrand(segment), segment_getSequence(segment))
: segment_construct(leftBlock, segment_getEvent(segment));
Segment *rightSegment = segment_getSequence(segment) != NULL
? segment_construct2(rightBlock, segment_getStart(segment) + block_getLength(leftBlock), segment_getStrand(segment), segment_getSequence(segment))
: segment_construct(rightBlock, segment_getEvent(segment));
//link together.
cap_makeAdjacent(segment_get3Cap(leftSegment), segment_get5Cap(rightSegment));
//update adjacencies.
Cap *_5Cap = segment_get5Cap(segment);
Cap *new5Cap = segment_get5Cap(leftSegment);
Cap *_3Cap = segment_get3Cap(segment);
Cap *new3Cap = segment_get3Cap(rightSegment);
if(cap_getAdjacency(_5Cap) != NULL) {
cap_makeAdjacent(cap_getAdjacency(_5Cap), new5Cap);
}
if(cap_getAdjacency(_3Cap) != NULL) {
cap_makeAdjacent(cap_getAdjacency(_3Cap), new3Cap);
}
return leftSegment;
}
void topDown(Flower *flower, Name referenceEventName) {
/*
* Run on each flower, top down. Sets the coordinates of each reference cap to the correct
* sequence, and sets the bases of the reference sequence to be consensus bases.
*/
Flower_EndIterator *endIt = flower_getEndIterator(flower);
End *end;
while ((end = flower_getNextEnd(endIt)) != NULL) {
Cap *cap = getCapForReferenceEvent(end, referenceEventName); //The cap in the reference
if (cap != NULL) {
cap = cap_getStrand(cap) ? cap : cap_getReverse(cap);
if (!cap_getSide(cap)) {
assert(cap_getCoordinate(cap) != INT64_MAX);
Sequence *sequence = cap_getSequence(cap);
assert(sequence != NULL);
Group *group = end_getGroup(end);
if (!group_isLeaf(group)) {
Flower *nestedFlower = group_getNestedFlower(group);
Cap *nestedCap = flower_getCap(nestedFlower, cap_getName(cap));
assert(nestedCap != NULL);
nestedCap = cap_getStrand(nestedCap) ? nestedCap : cap_getReverse(nestedCap);
assert(cap_getStrand(nestedCap));
assert(!cap_getSide(nestedCap));
int64_t endCoordinate = setCoordinates(nestedFlower, sequence_getMetaSequence(sequence),
nestedCap, cap_getCoordinate(cap));
(void) endCoordinate;
assert(endCoordinate == cap_getCoordinate(cap_getAdjacency(cap)));
assert(endCoordinate
== cap_getCoordinate(
flower_getCap(nestedFlower, cap_getName(cap_getAdjacency(cap)))));
}
}
}
}
flower_destructEndIterator(endIt);
}
void testCap_adjacent(CuTest* testCase) {
cactusCapTestSetup();
CuAssertTrue(testCase, cap_getAdjacency(leaf1Cap) == NULL);
CuAssertTrue(testCase, cap_getAdjacency(leaf3Cap) == NULL);
cap_makeAdjacent(leaf1Cap, leaf3Cap);
CuAssertTrue(testCase, cap_getAdjacency(leaf1Cap) == cap_getReverse(leaf3Cap));
CuAssertTrue(testCase, cap_getAdjacency(leaf3Cap) == cap_getReverse(leaf1Cap));
CuAssertTrue(testCase, cap_getAdjacency(cap_getReverse(leaf1Cap)) == leaf3Cap);
CuAssertTrue(testCase, cap_getAdjacency(cap_getReverse(leaf3Cap)) == leaf1Cap);
cactusCapTestTeardown();
}
stList *getContigPaths(Flower *flower, const char *eventString, stList *eventStrings) {
stList *maximalHaplotypePaths = stList_construct3(0,
(void(*)(void *)) stList_destruct);
stSortedSet *segmentSet = stSortedSet_construct();
getMaximalHaplotypePathsP(flower, maximalHaplotypePaths, segmentSet, eventString, eventStrings);
//Do some debug checks..
st_logDebug("We have %" PRIi64 " maximal haplotype paths\n", stList_length(
maximalHaplotypePaths));
getMaximalHaplotypePathsCheck(flower, segmentSet, eventString, eventStrings);
for (int64_t i = 0; i < stList_length(maximalHaplotypePaths); i++) {
stList *maximalHaplotypePath = stList_get(maximalHaplotypePaths, i);
st_logDebug("We have a maximal haplotype path with length %" PRIi64 "\n",
stList_length(maximalHaplotypePath));
assert(stList_length(maximalHaplotypePath) > 0);
Segment *_5Segment = stList_get(maximalHaplotypePath, 0);
Segment *_3Segment = stList_get(maximalHaplotypePath, stList_length(
maximalHaplotypePath) - 1);
if (getAdjacentCapsSegment(segment_get5Cap(_5Segment)) != NULL) {
assert(!trueAdjacency(segment_get5Cap(_5Segment), eventStrings));
}
if (getAdjacentCapsSegment(segment_get3Cap(_3Segment)) != NULL) {
assert(!trueAdjacency(segment_get3Cap(_3Segment), eventStrings));
}
for (int64_t j = 0; j < stList_length(maximalHaplotypePath) - 1; j++) {
_5Segment = stList_get(maximalHaplotypePath, j);
_3Segment = stList_get(maximalHaplotypePath, j + 1);
assert(trueAdjacency(segment_get3Cap(_5Segment), eventStrings));
assert(trueAdjacency(segment_get5Cap(_3Segment), eventStrings));
assert(cap_getAdjacency(getTerminalCap(segment_get3Cap(_5Segment)))
== getTerminalCap(segment_get5Cap(_3Segment)));
assert(strcmp(event_getHeader(segment_getEvent(_5Segment)),
eventString) == 0);
assert(strcmp(event_getHeader(segment_getEvent(_3Segment)),
eventString) == 0);
assert(hasCapInEvents(cap_getEnd(segment_get5Cap(_5Segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get5Cap(_5Segment))));
assert(hasCapInEvents(cap_getEnd(segment_get5Cap(_3Segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get5Cap(_3Segment))));
}
}
stSortedSet_destruct(segmentSet);
return maximalHaplotypePaths;
}
char *getTerminalAdjacencySubString(Cap *cap) {
if(getTerminalAdjacencyLength_ignoreAdjacencies) {
return stString_copy("");
}
cap = getTerminalCap(cap);
cap = cap_getStrand(cap) ? cap : cap_getReverse(cap); //This ensures the asserts are as expected.
Cap *adjacentCap = cap_getAdjacency(cap);
int64_t i = cap_getCoordinate(cap) - cap_getCoordinate(adjacentCap);
assert(i != 0);
if (i > 0) {
assert(cap_getSide(cap));
assert(!cap_getSide(adjacentCap));
return sequence_getString(cap_getSequence(cap),
cap_getCoordinate(adjacentCap) + 1, i - 1, 1);
} else {
assert(cap_getSide(adjacentCap));
assert(!cap_getSide(cap));
return sequence_getString(cap_getSequence(cap), cap_getCoordinate(cap) + 1, -i - 1, 1);
}
}
static void block_splitP2(Segment *segment,
Segment *parentLeftSegment,
Segment *parentRightSegment,
Block *leftBlock, Block *rightBlock) {
Segment *leftSegment = block_splitP(segment, leftBlock, rightBlock);
Segment *rightSegment = cap_getSegment(cap_getAdjacency(segment_get3Cap(leftSegment)));
if(parentLeftSegment != NULL) {
assert(parentRightSegment != NULL);
segment_makeParentAndChild(parentLeftSegment, leftSegment);
segment_makeParentAndChild(parentRightSegment, rightSegment);
}
else {
assert(parentRightSegment == NULL);
block_setRootInstance(leftBlock, leftSegment);
block_setRootInstance(rightBlock, rightSegment);
}
int64_t i;
for(i=0; i<segment_getChildNumber(segment); i++) {
block_splitP2(segment_getChild(segment, i), leftSegment, rightSegment, leftBlock, rightBlock);
}
}
static stList *getSubstringsForFlowers(stList *flowers) {
/*
* Get the set of substrings for sequence intervals in the given set of flowers.
*/
stList *substrings = stList_construct3(0, (void (*)(void *)) substring_destruct);
for (int64_t i = 0; i < stList_length(flowers); i++) {
Flower *flower = stList_get(flowers, i);
Flower_EndIterator *endIt = flower_getEndIterator(flower);
End *end;
while ((end = flower_getNextEnd(endIt)) != NULL) {
if (end_isStubEnd(end)) {
End_InstanceIterator *instanceIt = end_getInstanceIterator(end);
Cap *cap;
while ((cap = end_getNext(instanceIt)) != NULL) {
Sequence *sequence;
if ((sequence = cap_getSequence(cap)) != NULL) {
cap = cap_getStrand(cap) ? cap : cap_getReverse(cap);
if (!cap_getSide(cap)) { //We have a sequence interval of interest
Cap *adjacentCap = cap_getAdjacency(cap);
assert(adjacentCap != NULL);
int64_t length = cap_getCoordinate(adjacentCap) - cap_getCoordinate(cap) - 1;
assert(length >= 0);
if (length > 0) {
stList_append(substrings,
substring_construct(sequence_getMetaSequence(sequence)->stringName,
cap_getCoordinate(cap) + 1 - sequence_getStart(sequence), length));
}
}
}
}
end_destructInstanceIterator(instanceIt);
}
}
flower_destructEndIterator(endIt);
}
return substrings;
}
stSortedSet *makeEndAlignment(StateMachine *sM, End *end, int64_t spanningTrees, int64_t maxSequenceLength,
bool useProgressiveMerging, float gapGamma,
PairwiseAlignmentParameters *pairwiseAlignmentBandingParameters) {
//Make an alignment of the sequences in the ends
//Get the adjacency sequences to be aligned.
Cap *cap;
End_InstanceIterator *it = end_getInstanceIterator(end);
stList *sequences = stList_construct3(0, (void (*)(void *))adjacencySequence_destruct);
stList *seqFrags = stList_construct3(0, (void (*)(void *))seqFrag_destruct);
stHash *endInstanceNumbers = stHash_construct2(NULL, free);
while((cap = end_getNext(it)) != NULL) {
if(cap_getSide(cap)) {
cap = cap_getReverse(cap);
}
AdjacencySequence *adjacencySequence = adjacencySequence_construct(cap, maxSequenceLength);
stList_append(sequences, adjacencySequence);
assert(cap_getAdjacency(cap) != NULL);
End *otherEnd = end_getPositiveOrientation(cap_getEnd(cap_getAdjacency(cap)));
stList_append(seqFrags, seqFrag_construct(adjacencySequence->string, 0, end_getName(otherEnd)));
//Increase count of seqfrags with a given end.
int64_t *c = stHash_search(endInstanceNumbers, otherEnd);
if(c == NULL) {
c = st_calloc(1, sizeof(int64_t));
assert(*c == 0);
stHash_insert(endInstanceNumbers, otherEnd, c);
}
(*c)++;
}
end_destructInstanceIterator(it);
//Get the alignment.
MultipleAlignment *mA = makeAlignment(sM, seqFrags, spanningTrees, 100000000, useProgressiveMerging, gapGamma, pairwiseAlignmentBandingParameters);
//Build an array of weights to reweight pairs in the alignment.
int64_t *pairwiseAlignmentsPerSequenceNonCommonEnds = st_calloc(stList_length(seqFrags), sizeof(int64_t));
int64_t *pairwiseAlignmentsPerSequenceCommonEnds = st_calloc(stList_length(seqFrags), sizeof(int64_t));
//First build array on number of pairwise alignments to each sequence, distinguishing alignments between sequences sharing
//common ends.
for(int64_t i=0; i<stList_length(mA->chosenPairwiseAlignments); i++) {
stIntTuple *pairwiseAlignment = stList_get(mA->chosenPairwiseAlignments, i);
int64_t seq1 = stIntTuple_get(pairwiseAlignment, 1);
int64_t seq2 = stIntTuple_get(pairwiseAlignment, 2);
assert(seq1 != seq2);
SeqFrag *seqFrag1 = stList_get(seqFrags, seq1);
SeqFrag *seqFrag2 = stList_get(seqFrags, seq2);
int64_t *pairwiseAlignmentsPerSequence = seqFrag1->rightEndId == seqFrag2->rightEndId
? pairwiseAlignmentsPerSequenceCommonEnds : pairwiseAlignmentsPerSequenceNonCommonEnds;
pairwiseAlignmentsPerSequence[seq1]++;
pairwiseAlignmentsPerSequence[seq2]++;
}
//Now calculate score adjustments.
double *scoreAdjustmentsNonCommonEnds = st_malloc(stList_length(seqFrags) * sizeof(double));
double *scoreAdjustmentsCommonEnds = st_malloc(stList_length(seqFrags) * sizeof(double));
for(int64_t i=0; i<stList_length(seqFrags); i++) {
SeqFrag *seqFrag = stList_get(seqFrags, i);
End *otherEnd = flower_getEnd(end_getFlower(end), seqFrag->rightEndId);
assert(otherEnd != NULL);
assert(stHash_search(endInstanceNumbers, otherEnd) != NULL);
int64_t commonInstanceNumber = *(int64_t *)stHash_search(endInstanceNumbers, otherEnd);
int64_t nonCommonInstanceNumber = stList_length(seqFrags) - commonInstanceNumber;
assert(commonInstanceNumber > 0 && nonCommonInstanceNumber >= 0);
assert(pairwiseAlignmentsPerSequenceNonCommonEnds[i] <= nonCommonInstanceNumber);
assert(pairwiseAlignmentsPerSequenceNonCommonEnds[i] >= 0);
assert(pairwiseAlignmentsPerSequenceCommonEnds[i] < commonInstanceNumber);
assert(pairwiseAlignmentsPerSequenceCommonEnds[i] >= 0);
//scoreAdjustmentsNonCommonEnds[i] = ((double)nonCommonInstanceNumber + commonInstanceNumber - 1)/(pairwiseAlignmentsPerSequenceNonCommonEnds[i] + pairwiseAlignmentsPerSequenceCommonEnds[i]);
//scoreAdjustmentsCommonEnds[i] = scoreAdjustmentsNonCommonEnds[i];
if(pairwiseAlignmentsPerSequenceNonCommonEnds[i] > 0) {
scoreAdjustmentsNonCommonEnds[i] = ((double)nonCommonInstanceNumber)/pairwiseAlignmentsPerSequenceNonCommonEnds[i];
assert(scoreAdjustmentsNonCommonEnds[i] >= 1.0);
assert(scoreAdjustmentsNonCommonEnds[i] <= nonCommonInstanceNumber);
}
else {
scoreAdjustmentsNonCommonEnds[i] = INT64_MIN;
}
if(pairwiseAlignmentsPerSequenceCommonEnds[i] > 0) {
scoreAdjustmentsCommonEnds[i] = ((double)commonInstanceNumber-1)/pairwiseAlignmentsPerSequenceCommonEnds[i];
assert(scoreAdjustmentsCommonEnds[i] >= 1.0);
assert(scoreAdjustmentsCommonEnds[i] <= commonInstanceNumber-1);
}
else {
scoreAdjustmentsCommonEnds[i] = INT64_MIN;
}
}
//Convert the alignment pairs to an alignment of the caps..
stSortedSet *sortedAlignment =
stSortedSet_construct3((int (*)(const void *, const void *))alignedPair_cmpFn,
(void (*)(void *))alignedPair_destruct);
while(stList_length(mA->alignedPairs) > 0) {
stIntTuple *alignedPair = stList_pop(mA->alignedPairs);
assert(stIntTuple_length(alignedPair) == 5);
int64_t seqIndex1 = stIntTuple_get(alignedPair, 1);
int64_t seqIndex2 = stIntTuple_get(alignedPair, 3);
AdjacencySequence *i = stList_get(sequences, seqIndex1);
AdjacencySequence *j = stList_get(sequences, seqIndex2);
assert(i != j);
int64_t offset1 = stIntTuple_get(alignedPair, 2);
int64_t offset2 = stIntTuple_get(alignedPair, 4);
int64_t score = stIntTuple_get(alignedPair, 0);
if(score <= 0) { //Happens when indel probs are included
score = 1; //This is the minimum
}
assert(score > 0 && score <= PAIR_ALIGNMENT_PROB_1);
SeqFrag *seqFrag1 = stList_get(seqFrags, seqIndex1);
SeqFrag *seqFrag2 = stList_get(seqFrags, seqIndex2);
assert(seqFrag1 != seqFrag2);
double *scoreAdjustments = seqFrag1->rightEndId == seqFrag2->rightEndId ? scoreAdjustmentsCommonEnds : scoreAdjustmentsNonCommonEnds;
assert(scoreAdjustments[seqIndex1] != INT64_MIN);
assert(scoreAdjustments[seqIndex2] != INT64_MIN);
AlignedPair *alignedPair2 = alignedPair_construct(
i->subsequenceIdentifier, i->start + (i->strand ? offset1 : -offset1), i->strand,
j->subsequenceIdentifier, j->start + (j->strand ? offset2 : -offset2), j->strand,
score*scoreAdjustments[seqIndex1], score*scoreAdjustments[seqIndex2]); //Do the reweighting here.
assert(stSortedSet_search(sortedAlignment, alignedPair2) == NULL);
assert(stSortedSet_search(sortedAlignment, alignedPair2->reverse) == NULL);
stSortedSet_insert(sortedAlignment, alignedPair2);
stSortedSet_insert(sortedAlignment, alignedPair2->reverse);
stIntTuple_destruct(alignedPair);
}
//Cleanup
stList_destruct(seqFrags);
stList_destruct(sequences);
free(pairwiseAlignmentsPerSequenceNonCommonEnds);
free(pairwiseAlignmentsPerSequenceCommonEnds);
free(scoreAdjustmentsNonCommonEnds);
free(scoreAdjustmentsCommonEnds);
multipleAlignment_destruct(mA);
stHash_destruct(endInstanceNumbers);
return sortedAlignment;
}
Segment *getAdjacentCapsSegment(Cap *cap) {
cap = getTerminalCap(cap);
cap = cap_getAdjacency(cap);
assert(cap != NULL);
return getCapsSegment(cap);
}
int mapGene(Cap *cap, int level, int exon, struct bed *gene, FILE *fileHandle){
/*
*Following cactus adjacencies, starting from 'cap', find regions that overlap with
*exons of input gene. Report chain relations of these regions with the exons.
*cap: current cap. Level = chain level. exon = exon number. gene = bed record of gene
*/
int64_t exonStart, exonEnd;
if(isStubCap(cap)){
Group *group = end_getGroup(cap_getEnd(cap));
Flower *nestedFlower = group_getNestedFlower(group);
if(nestedFlower != NULL){//recursive call
Cap *childCap = flower_getCap(nestedFlower, cap_getName(cap));
assert(childCap != NULL);
exon = mapGene(childCap, level + 1, exon, gene, fileHandle);
exonStart = gene->chromStarts->list[exon] + gene->chromStart;
exonEnd = exonStart + gene->blockSizes->list[exon];
}
}
cap = cap_getAdjacency(cap);
Cap *nextcap;
int64_t capCoor;
exonStart = gene->chromStarts->list[exon] + gene->chromStart;
exonEnd = exonStart + gene->blockSizes->list[exon];
Block *block = end_getBlock(cap_getEnd(cap));
if(block == NULL){
moveCapToNextBlock(&cap);
}
while(!isStubCap(cap) && exon < gene->blockCount){
End *cend = cap_getEnd(cap);
capCoor = cap_getCoordinate(cap);//Cap coordinate is always the coordinate on + strand
nextcap = cap_getAdjacency(cap_getOtherSegmentCap(cap));
st_logInfo("capCoor: %d, nextCap: %d, eStart: %d, eEnd: %d. Exon: %d\n",
capCoor, cap_getCoordinate(nextcap), exonStart, exonEnd, exon);
//keep moving if nextBlock Start is still upstream of current exon
if(cap_getCoordinate(nextcap) <= exonStart){
moveCapToNextBlock(&cap);
st_logInfo("Still upstream, nextcap <= exonStart. Move to next chainBlock\n");
}else if(capCoor >= exonEnd){//Done with current exon, move to next
st_logInfo("Done with current exon, move to next one\n\n");
fprintf(fileHandle, "\t\t</exon>\n");//end previous exon
exon++;
if(exon < gene->blockCount){
exonStart = gene->chromStarts->list[exon] + gene->chromStart;
exonEnd = exonStart + gene->blockSizes->list[exon];
fprintf(fileHandle, "\t\t<exon id=\"%d\" start=\"%" PRIi64 "\" end=\"%" PRIi64 "\">\n", exon, exonStart, exonEnd);
}
}else{//current exon overlaps with current block Or with lower level flower
Cap *oppcap = cap_getOtherSegmentCap(cap);
st_logInfo("Current exon overlaps with current block or with lower flower\n");
if(cap_getCoordinate(oppcap) >= exonStart && exonEnd > capCoor){
mapBlockToExon(cap, level, fileHandle);
if(exonEnd <= cap_getCoordinate(oppcap) + 1){
st_logInfo("Done with current exon, move to next one\n\n");
fprintf(fileHandle, "\t\t</exon>\n");//end previous exon
exon++;
if(exon < gene->blockCount){
exonStart = gene->chromStarts->list[exon] + gene->chromStart;
exonEnd = exonStart + gene->blockSizes->list[exon];
fprintf(fileHandle, "\t\t<exon id=\"%d\" start=\"%" PRIi64 "\" end=\"%" PRIi64 "\">\n", exon, exonStart, exonEnd);
}
continue;
}
}
//Traverse lower level flowers if exists
Group *group = end_getGroup(end_getOtherBlockEnd(cend));
Flower *nestedFlower = group_getNestedFlower(group);
if(nestedFlower != NULL){//recursive call
Cap *childCap = flower_getCap(nestedFlower, cap_getName(cap_getOtherSegmentCap(cap)));
assert(childCap != NULL);
exon = mapGene(childCap, level + 1, exon, gene, fileHandle);
exonStart = gene->chromStarts->list[exon] + gene->chromStart;
exonEnd = exonStart + gene->blockSizes->list[exon];
}
moveCapToNextBlock(&cap);
}
}
return exon;
}
int main(int argc, char *argv[]) {
st_setLogLevelFromString(argv[1]);
st_logDebug("Set up logging\n");
stKVDatabaseConf *kvDatabaseConf = stKVDatabaseConf_constructFromString(argv[2]);
CactusDisk *cactusDisk = cactusDisk_construct(kvDatabaseConf, 0);
stKVDatabaseConf_destruct(kvDatabaseConf);
st_logDebug("Set up the flower disk\n");
Name flowerName = cactusMisc_stringToName(argv[3]);
Flower *flower = cactusDisk_getFlower(cactusDisk, flowerName);
int64_t totalBases = flower_getTotalBaseLength(flower);
int64_t totalEnds = flower_getEndNumber(flower);
int64_t totalFreeEnds = flower_getFreeStubEndNumber(flower);
int64_t totalAttachedEnds = flower_getAttachedStubEndNumber(flower);
int64_t totalCaps = flower_getCapNumber(flower);
int64_t totalBlocks = flower_getBlockNumber(flower);
int64_t totalGroups = flower_getGroupNumber(flower);
int64_t totalChains = flower_getChainNumber(flower);
int64_t totalLinkGroups = 0;
int64_t maxEndDegree = 0;
int64_t maxAdjacencyLength = 0;
int64_t totalEdges = 0;
Flower_EndIterator *endIt = flower_getEndIterator(flower);
End *end;
while((end = flower_getNextEnd(endIt)) != NULL) {
assert(end_getOrientation(end));
if(end_getInstanceNumber(end) > maxEndDegree) {
maxEndDegree = end_getInstanceNumber(end);
}
stSortedSet *ends = stSortedSet_construct();
End_InstanceIterator *capIt = end_getInstanceIterator(end);
Cap *cap;
while((cap = end_getNext(capIt)) != NULL) {
if(cap_getSequence(cap) != NULL) {
Cap *adjacentCap = cap_getAdjacency(cap);
assert(adjacentCap != NULL);
End *adjacentEnd = end_getPositiveOrientation(cap_getEnd(adjacentCap));
stSortedSet_insert(ends, adjacentEnd);
int64_t adjacencyLength = cap_getCoordinate(cap) - cap_getCoordinate(adjacentCap);
if(adjacencyLength < 0) {
adjacencyLength *= -1;
}
assert(adjacencyLength >= 1);
if(adjacencyLength >= maxAdjacencyLength) {
maxAdjacencyLength = adjacencyLength;
}
}
}
end_destructInstanceIterator(capIt);
totalEdges += stSortedSet_size(ends);
if(stSortedSet_search(ends, end) != NULL) { //This ensures we count self edges twice, so that the division works.
totalEdges += 1;
}
stSortedSet_destruct(ends);
}
assert(totalEdges % 2 == 0);
flower_destructEndIterator(endIt);
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while((group = flower_getNextGroup(groupIt)) != NULL) {
if(group_getLink(group) != NULL) {
totalLinkGroups++;
}
}
flower_destructGroupIterator(groupIt);
printf("flower name: %" PRIi64 " total bases: %" PRIi64 " total-ends: %" PRIi64 " total-caps: %" PRIi64 " max-end-degree: %" PRIi64 " max-adjacency-length: %" PRIi64 " total-blocks: %" PRIi64 " total-groups: %" PRIi64 " total-edges: %" PRIi64 " total-free-ends: %" PRIi64 " total-attached-ends: %" PRIi64 " total-chains: %" PRIi64 " total-link groups: %" PRIi64 "\n",
flower_getName(flower), totalBases, totalEnds, totalCaps, maxEndDegree, maxAdjacencyLength, totalBlocks, totalGroups, totalEdges/2, totalFreeEnds, totalAttachedEnds, totalChains, totalLinkGroups);
return 0;
}
