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 void setAdjacencyLength(Cap *cap, Cap *adjacentCap, int64_t adjacencyLength) {
//Set the coordinates of the caps to the adjacency size
cap_setCoordinates(cap, adjacencyLength, cap_getStrand(cap), NULL);
cap_setCoordinates(adjacentCap, adjacencyLength, cap_getStrand(adjacentCap),
NULL);
assert(cap_getCoordinate(cap) == cap_getCoordinate(adjacentCap));
}
void testCap_segmentCoordinatesReverseStrand(CuTest* testCase) {
/*
* Tests the coordinates of an segment and its 5 and 3 prime caps.
*/
cactusCapTestSetup();
Block *block = block_construct(3, flower);
Segment *segment = segment_construct2(block, 2, 0, sequence);
Cap *_5Cap = segment_get5Cap(segment);
Cap *_3Cap = segment_get3Cap(segment);
CuAssertTrue(testCase, cap_getSide(_5Cap));
CuAssertTrue(testCase, !cap_getSide(_3Cap));
CuAssertTrue(testCase, !cap_getStrand(_5Cap));
CuAssertTrue(testCase, !cap_getStrand(_3Cap));
CuAssertIntEquals(testCase, 4, cap_getCoordinate(_5Cap));
CuAssertIntEquals(testCase, 2, cap_getCoordinate(_3Cap));
CuAssertTrue(testCase, !segment_getStrand(segment));
CuAssertIntEquals(testCase, 4, segment_getStart(segment));
CuAssertIntEquals(testCase, 3, segment_getLength(segment));
CuAssertTrue(testCase, !cap_getSide(cap_getReverse(_5Cap)));
CuAssertTrue(testCase, cap_getSide(cap_getReverse(_3Cap)));
CuAssertTrue(testCase, cap_getStrand(cap_getReverse(_5Cap)));
CuAssertTrue(testCase, cap_getStrand(cap_getReverse(_3Cap)));
CuAssertIntEquals(testCase, 4, cap_getCoordinate(cap_getReverse(_5Cap)));
CuAssertIntEquals(testCase, 2, cap_getCoordinate(cap_getReverse(_3Cap)));
CuAssertTrue(testCase, segment_getStrand(segment_getReverse(segment)));
CuAssertIntEquals(testCase, 2, segment_getStart(segment_getReverse(segment)));
CuAssertIntEquals(testCase, 3, segment_getLength(segment_getReverse(segment)));
cactusCapTestTeardown();
}
bool capsAreAdjacent(Cap *cap1, Cap *cap2, int64_t *separationDistance) {
if (cap_getName(cap2) != cap_getName(cap1) && cap_getCoordinate(cap1) != cap_getCoordinate(cap2)) { //This can happen if end1 == end2
if (sequence_getMetaSequence(cap_getSequence(cap1)) == sequence_getMetaSequence(cap_getSequence(cap2))) {
assert(strcmp(event_getHeader(cap_getEvent(cap1)), event_getHeader(
cap_getEvent(cap2))) == 0);
assert(cap_getPositiveOrientation(cap1)
!= cap_getPositiveOrientation(cap2));
assert(cap_getName(cap1) != cap_getName(cap2));
assert(sequence_getMetaSequence(cap_getSequence(cap1))
== sequence_getMetaSequence(cap_getSequence(cap2)));
if (!cap_getStrand(cap1)) {
cap1 = cap_getReverse(cap1);
}
if (!cap_getStrand(cap2)) {
cap2 = cap_getReverse(cap2);
}
assert(cap_getStrand(cap1));
assert(cap_getStrand(cap2));
if (cap_getCoordinate(cap1) < cap_getCoordinate(cap2)) {
if (!cap_getSide(cap1) && cap_getSide(cap2)) {
*separationDistance = cap_getCoordinate(cap2) - cap_getCoordinate(cap1) - 1; //The minus 1, to give the length of the sequence between the two caps.
return 1;
}
} else {
if (cap_getSide(cap1) && !cap_getSide(cap2)) {
*separationDistance = cap_getCoordinate(cap1) - cap_getCoordinate(cap2) - 1;
return 1;
}
}
}
}
return 0;
}
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;
}
}
}
void testCap_setCoordinate(CuTest* testCase) {
cactusCapTestSetup();
CuAssertTrue(testCase, cap_getCoordinate(rootCap) == INT64_MAX);
CuAssertTrue(testCase, cap_getStrand(rootCap));
CuAssertTrue(testCase, cap_getSequence(rootCap) == NULL);
cap_setCoordinates(rootCap, 5, 0, NULL);
CuAssertTrue(testCase, cap_getCoordinate(rootCap) == 5);
CuAssertTrue(testCase, !cap_getStrand(rootCap));
CuAssertTrue(testCase, cap_getSequence(rootCap) == NULL);
cap_setCoordinates(rootCap, INT64_MAX, 1, NULL);
CuAssertTrue(testCase, cap_getCoordinate(rootCap) == INT64_MAX);
CuAssertTrue(testCase, cap_getStrand(rootCap));
CuAssertTrue(testCase, cap_getSequence(rootCap) == NULL);
cactusCapTestTeardown();
}
static void recoverBrokenAdjacencies(Flower *flower, stList *recoveredCaps, Name referenceEventName) {
/*
* Find reference intervals that are book-ended by stubs created in a child flower.
*/
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while((group = flower_getNextGroup(groupIt)) != NULL) {
Flower *nestedFlower;
if((nestedFlower = group_getNestedFlower(group)) != NULL) {
Flower_EndIterator *endIt = flower_getEndIterator(nestedFlower);
End *childEnd;
while((childEnd = flower_getNextEnd(endIt)) != NULL) {
if(end_isStubEnd(childEnd) && flower_getEnd(flower, end_getName(childEnd)) == NULL) { //We have a thread we need to promote
Cap *childCap = getCapForReferenceEvent(childEnd, referenceEventName); //The cap in the reference
assert(childCap != NULL);
assert(!end_isAttached(childEnd));
childCap = cap_getStrand(childCap) ? childCap : cap_getReverse(childCap);
if (!cap_getSide(childCap)) {
Cap *adjacentChildCap = NULL;
int64_t adjacencyLength = traceThreadLength(childCap, &adjacentChildCap);
Cap *cap = copyCapToParent(childCap, recoveredCaps);
assert(adjacentChildCap != NULL);
assert(!end_isAttached(cap_getEnd(adjacentChildCap)));
assert(!cap_getSide(cap));
Cap *adjacentCap = copyCapToParent(adjacentChildCap, recoveredCaps);
cap_makeAdjacent(cap, adjacentCap);
setAdjacencyLength(cap, adjacentCap, adjacencyLength);
}
}
}
flower_destructEndIterator(endIt);
}
}
flower_destructGroupIterator(groupIt);
}
void stCaf_addAdjacencies(Flower *flower) {
//Build a list of caps.
stList *list = stList_construct();
Flower_EndIterator *endIterator = flower_getEndIterator(flower);
End *end;
while ((end = flower_getNextEnd(endIterator)) != NULL) {
End_InstanceIterator *instanceIterator = end_getInstanceIterator(end);
Cap *cap;
while ((cap = end_getNext(instanceIterator)) != NULL) {
if (!cap_getStrand(cap)) {
cap = cap_getReverse(cap);
}
stList_append(list, cap);
}
end_destructInstanceIterator(instanceIterator);
}
flower_destructEndIterator(endIterator);
assert(stList_length(list) % 2 == 0);
//Sort the list of caps.
stList_sort(list, (int(*)(const void *, const void *)) addAdjacenciesPP);
//Now make the adjacencies.
for (int64_t i = 1; i < stList_length(list); i += 2) {
Cap *cap = stList_get(list, i - 1);
Cap *cap2 = stList_get(list, i);
cap_makeAdjacent(cap, cap2);
}
//Clean up.
stList_destruct(list);
}
static int addAdjacenciesPP(Cap *cap1, Cap *cap2) {
assert(cap_getStrand(cap1) && cap_getStrand(cap2));
Sequence *sequence1 = cap_getSequence(cap1);
Sequence *sequence2 = cap_getSequence(cap2);
int64_t i = cactusMisc_nameCompare(sequence_getName(sequence1), sequence_getName(sequence2));
if (i == 0) {
int64_t j = cap_getCoordinate(cap1);
int64_t k = cap_getCoordinate(cap2);
i = j > k ? 1 : (j < k ? -1 : 0);
if (i == 0) {
assert(cap_getSegment(cap1) == cap_getSegment(cap2));
j = cap_getSide(cap1);
k = cap_getSide(cap2);
assert((j && !k) || (!j && k));
i = j ? -1 : 1;
}
}
return i;
}
void bottomUp(stList *flowers, stKVDatabase *sequenceDatabase, Name referenceEventName,
bool isTop, stMatrix *(*generateSubstitutionMatrix)(double)) {
/*
* 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.
*/
stList *caps = getCaps(flowers, referenceEventName);
for (int64_t i = stList_length(caps) - 1; i >= 0; i--) { //Start from end, as we add to this list.
setAdjacencyLengthsAndRecoverNewCapsAndBrokenAdjacencies(stList_get(caps, i), caps);
}
for(int64_t i=0; i<stList_length(flowers); i++) {
recoverBrokenAdjacencies(stList_get(flowers, i), caps, referenceEventName);
}
//Build the phylogenetic event trees for base calling.
segmentWriteFn_flowerToPhylogeneticTreeHash = stHash_construct2(NULL, (void (*)(void *))cleanupPhylogeneticTree);
for(int64_t i=0; i<stList_length(flowers); i++) {
Flower *flower = stList_get(flowers, i);
Event *refEvent = eventTree_getEvent(flower_getEventTree(flower), referenceEventName);
assert(refEvent != NULL);
stHash_insert(segmentWriteFn_flowerToPhylogeneticTreeHash, flower, getPhylogeneticTreeRootedAtGivenEvent(refEvent, generateSubstitutionMatrix));
}
if (isTop) {
stList *threadStrings = buildRecursiveThreadsInList(sequenceDatabase, caps, segmentWriteFn,
terminalAdjacencyWriteFn);
assert(stList_length(threadStrings) == stList_length(caps));
int64_t nonTrivialSeqIndex = 0, trivialSeqIndex = stList_length(threadStrings); //These are used as indices for the names of trivial and non-trivial sequences.
for (int64_t i = 0; i < stList_length(threadStrings); i++) {
Cap *cap = stList_get(caps, i);
assert(cap_getStrand(cap));
assert(!cap_getSide(cap));
Flower *flower = end_getFlower(cap_getEnd(cap));
char *threadString = stList_get(threadStrings, i);
bool trivialString = isTrivialString(&threadString); //This alters the original string
MetaSequence *metaSequence = addMetaSequence(flower, cap, trivialString ? trivialSeqIndex++ : nonTrivialSeqIndex++,
threadString, trivialString);
free(threadString);
int64_t endCoordinate = setCoordinates(flower, metaSequence, cap, metaSequence_getStart(metaSequence) - 1);
(void) endCoordinate;
assert(endCoordinate == metaSequence_getLength(metaSequence) + metaSequence_getStart(metaSequence));
}
stList_setDestructor(threadStrings, NULL); //The strings are already cleaned up by the above loop
stList_destruct(threadStrings);
} else {
buildRecursiveThreads(sequenceDatabase, caps, segmentWriteFn, terminalAdjacencyWriteFn);
}
stHash_destruct(segmentWriteFn_flowerToPhylogeneticTreeHash);
stList_destruct(caps);
}
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;
}
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_getStrand(CuTest* testCase) {
cactusCapTestSetup();
CuAssertTrue(testCase, cap_getStrand(rootCap));
CuAssertTrue(testCase, !cap_getStrand(cap_getReverse(rootCap)));
CuAssertTrue(testCase, cap_getStrand(leaf1Cap));
CuAssertTrue(testCase, !cap_getStrand(cap_getReverse(leaf1Cap)));
CuAssertTrue(testCase, !cap_getStrand(leaf2Cap));
CuAssertTrue(testCase, cap_getStrand(cap_getReverse(leaf2Cap)));
cactusCapTestTeardown();
}
static Cap *copyCapToParent(Cap *cap, stList *recoveredCaps) {
/*
* Get the adjacent stub end by looking at the reference adjacency in the parent.
*/
End *end = cap_getEnd(cap);
assert(end != NULL);
Group *parentGroup = flower_getParentGroup(end_getFlower(end));
assert(parentGroup != NULL);
End *copiedEnd = end_copyConstruct(end, group_getFlower(parentGroup));
end_setGroup(copiedEnd, parentGroup); //Set group
Cap *copiedCap = end_getInstance(copiedEnd, cap_getName(cap));
assert(copiedCap != NULL);
copiedCap = cap_getStrand(copiedCap) ? copiedCap : cap_getReverse(copiedCap);
if (!cap_getSide(copiedCap)) {
stList_append(recoveredCaps, copiedCap);
}
return copiedCap;
}
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 stList *getCaps(stList *flowers, Name referenceEventName) {
stList *caps = stList_construct();
for (int64_t i = 0; i < stList_length(flowers); i++) {
Flower *flower = stList_get(flowers, i);
//Get list of caps
Flower_EndIterator *endIt = flower_getEndIterator(flower);
End *end;
while ((end = flower_getNextEnd(endIt)) != NULL) {
if (end_isStubEnd(end)) {
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)) {
stList_append(caps, cap);
}
}
}
}
flower_destructEndIterator(endIt);
}
return caps;
}
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;
}
