/*
* Function does the actual depth first search to detect if the thing has an acyclic ordering.
*/
static int64_t dfs(stHash *adjacencyList, stIntTuple *seqPos,
stSortedSet *started, stSortedSet *done) {
if(stSortedSet_search(started, seqPos) != NULL) {
if(stSortedSet_search(done, seqPos) == NULL) {
//We have detected a cycle
//st_logInfo("I have cycle %" PRIi64 " %" PRIi64 "\n", stIntTuple_getPosition(seqPos, 0), stIntTuple_getPosition(seqPos, 1));
return 1;
}
//We have already explored this area, but no cycle.
return 0;
}
stSortedSet_insert(started, seqPos);
int64_t cycle =0;
stIntTuple *nextSeqPos = stIntTuple_construct2( stIntTuple_get(seqPos, 0), stIntTuple_get(seqPos, 1) + 1);
stSortedSet *column = stHash_search(adjacencyList, nextSeqPos);
if(column != NULL) { //It is in the adjacency list, so we can do the recursion
assert(stSortedSet_search(column, nextSeqPos) != NULL);
stSortedSetIterator *it = stSortedSet_getIterator(column);
stIntTuple *seqPos2;
while((seqPos2 = stSortedSet_getNext(it)) != NULL) {
cycle = cycle || dfs(adjacencyList, seqPos2, started, done);
}
stSortedSet_destructIterator(it);
}
stIntTuple_destruct(nextSeqPos);
stSortedSet_insert(done, seqPos);
return cycle;
}
static void getMaximalHaplotypePathsCheck(Flower *flower,
stSortedSet *segmentSet, const char *eventString, stList *eventStrings) {
/*
* Do debug checks that the haplotypes paths are well formed.
*/
Flower_SegmentIterator *segmentIt = flower_getSegmentIterator(flower);
Segment *segment;
while ((segment = flower_getNextSegment(segmentIt)) != NULL) {
if (strcmp(event_getHeader(segment_getEvent(segment)), eventString) == 0) {
if (hasCapInEvents(cap_getEnd(segment_get5Cap(segment)), eventStrings)) { //isHaplotypeEnd(cap_getEnd(segment_get5Cap(segment)))) {
assert(stSortedSet_search(segmentSet, segment) != NULL
|| stSortedSet_search(segmentSet, segment_getReverse(
segment)) != NULL);
}
}
}
flower_destructSegmentIterator(segmentIt);
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (group_getNestedFlower(group) != NULL) {
getMaximalHaplotypePathsCheck(group_getNestedFlower(group),
segmentSet, eventString, eventStrings);
}
}
flower_destructGroupIterator(groupIt);
}
static void test_stSortedSet_searchGreaterThan(CuTest* testCase) {
sonLibSortedSetTestSetup();
for(int32_t i=0; i<size; i++) {
stSortedSet_insert(sortedSet, stIntTuple_construct(1, input[i]));
}
//static int32_t sortedInput[] = { -10, -1, 1, 3, 5, 10, 12 };
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, -11)) == stSortedSet_search(sortedSet, stIntTuple_construct(1, -10)));
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, -10)) == stSortedSet_search(sortedSet, stIntTuple_construct(1, -1)));
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, -5)) == stSortedSet_search(sortedSet, stIntTuple_construct(1, -1)));
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, 1)) == stSortedSet_search(sortedSet, stIntTuple_construct(1, 3)));
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, 13)) == NULL);
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet,
stIntTuple_construct(1, 12)) == NULL);
for(int32_t i=0; i<100; i++) {
stSortedSet_insert(sortedSet, stIntTuple_construct(1, st_randomInt(-1000, 1000)));
}
stList *list = stSortedSet_getList(sortedSet);
for(int32_t i=1; i<stList_length(list); i++) {
stIntTuple *p = stList_get(list, i-1);
stIntTuple *j = stList_get(list, i);
stIntTuple *k = stIntTuple_construct(1, st_randomInt(stIntTuple_getPosition(p, 0), stIntTuple_getPosition(j, 0)));
CuAssertTrue(testCase, stSortedSet_searchGreaterThan(sortedSet, k) == j);
stIntTuple_destruct(k);
}
stList_destruct(list);
sonLibSortedSetTestTeardown();
}
static void getMaximalHaplotypePathsP2(Segment *segment,
stList *maximalHaplotypePath, stSortedSet *segmentSet, stList *eventStrings) {
/*
* Iterate all the way to one end of the contig then start the traversal to define the maximal
* haplotype path.
*/
Cap *_5Cap = segment_get5Cap(segment);
assert(hasCapInEvents(cap_getEnd(segment_get3Cap(segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get3Cap(segment))));
if (trueAdjacency(_5Cap, eventStrings)) { //Check that the adjacency is supported by a haplotype path
Segment *otherSegment = getAdjacentCapsSegment(_5Cap);
assert(segment != otherSegment);
assert(segment_getReverse(segment) != otherSegment);
if (otherSegment != NULL) {
assert(stSortedSet_search(segmentSet, otherSegment) == NULL);
assert(stSortedSet_search(segmentSet, segment_getReverse(
otherSegment)) == NULL);
assert(hasCapInEvents(cap_getEnd(segment_get3Cap(otherSegment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get3Cap(otherSegment))));
getMaximalHaplotypePathsP2(otherSegment, maximalHaplotypePath,
segmentSet, eventStrings);
} else { //We need to start the maximal haplotype recursion
getMaximalHaplotypePathsP3(segment, maximalHaplotypePath,
segmentSet, eventStrings);
}
} else {
getMaximalHaplotypePathsP3(segment, maximalHaplotypePath, segmentSet, eventStrings);
}
}
static void getMaximalHaplotypePathsP3(Segment *segment,
stList *maximalHaplotypePath, stSortedSet *segmentSet, stList *eventStrings) {
stList_append(maximalHaplotypePath, segment);
assert(stSortedSet_search(segmentSet, segment) == NULL);
assert(stSortedSet_search(segmentSet, segment_getReverse(segment)) == NULL);
stSortedSet_insert(segmentSet, segment);
Cap *_3Cap = segment_get3Cap(segment);
if (trueAdjacency(_3Cap, eventStrings)) { //Continue on..
Segment *otherSegment = getAdjacentCapsSegment(_3Cap);
if (otherSegment != NULL) {
getMaximalHaplotypePathsP3(otherSegment, maximalHaplotypePath,
segmentSet, eventStrings);
}
}
}
static void makeMatchingPerfect(stList *chosenEdges, stList *adjacencyEdges,
stSortedSet *nodes) {
/*
* While the the number of edges is less than a perfect matching add random edges.
*/
stSortedSet *attachedNodes = getNodeSetOfEdges(chosenEdges);
stHash *nodesToAdjacencyEdges = getNodesToEdgesHash(adjacencyEdges);
stIntTuple *pNode = NULL;
stSortedSetIterator *it = stSortedSet_getIterator(nodes);
stIntTuple *node;
while((node = stSortedSet_getNext(it)) != NULL) {
if (stSortedSet_search(attachedNodes, node) == NULL) {
if (pNode == NULL) {
pNode = node;
} else {
stList_append(chosenEdges,
getEdgeForNodes(stIntTuple_get(pNode, 0), stIntTuple_get(node, 0), nodesToAdjacencyEdges));
pNode = NULL;
}
}
}
stSortedSet_destructIterator(it);
assert(pNode == NULL);
stSortedSet_destruct(attachedNodes);
assert(stList_length(chosenEdges) * 2 == stSortedSet_size(nodes));
stHash_destruct(nodesToAdjacencyEdges);
}
Block *flower_getBlock(Flower *flower, Name name) {
Block block;
BlockContents blockContents;
block.blockContents = &blockContents;
blockContents.name = name;
return stSortedSet_search(flower->blocks, &block);
}
/*
* Uses the functions above to build an adjacency list, then by DFS attempts to create
* a valid topological sort, returning non-zero if the graph contains a cycle.
*/
static int64_t containsACycle(stList *pairs, int64_t sequenceNumber) {
//Build an adjacency list structure..
stHash *adjacencyList = buildAdjacencyList(pairs, sequenceNumber);
//Do a topological sort of the adjacency list
stSortedSet *started = stSortedSet_construct3((int (*)(const void *, const void *))stIntTuple_cmpFn, NULL);
stSortedSet *done = stSortedSet_construct3((int (*)(const void *, const void *))stIntTuple_cmpFn, NULL);
int64_t cyclic = 0;
for(int64_t seq=0; seq<sequenceNumber; seq++) {
stIntTuple *seqPos = stIntTuple_construct2( seq, 0); //The following hacks avoid memory cleanup..
stSortedSet *column = stHash_search(adjacencyList, seqPos);
assert(column != NULL);
stIntTuple *seqPos2 = stSortedSet_search(column, seqPos);
assert(seqPos2 != NULL);
cyclic = cyclic || dfs(adjacencyList, seqPos2, started, done);
stIntTuple_destruct(seqPos);
}
//cleanup
stHashIterator *it = stHash_getIterator(adjacencyList);
stIntTuple *seqPos;
stSortedSet *columns = stSortedSet_construct2((void (*)(void *))stSortedSet_destruct);
while((seqPos = stHash_getNext(it)) != NULL) {
stSortedSet *column = stHash_search(adjacencyList, seqPos);
assert(column != NULL);
stSortedSet_insert(columns, column);
}
stHash_destructIterator(it);
stHash_destruct(adjacencyList);
stSortedSet_destruct(columns);
stSortedSet_destruct(started);
stSortedSet_destruct(done);
return cyclic;
}
static void checkComponents(CuTest *testCase, stList *filteredEdges) {
stHash *nodesToComponents = getComponents(filteredEdges);
//Check all components are smaller than threshold
stList *components = stHash_getValues(nodesToComponents);
for (int64_t i = 0; i < stList_length(components); i++) {
stSortedSet *component = stList_get(components, i);
CuAssertTrue(testCase, stSortedSet_size(component) <= maxComponentSize);
CuAssertTrue(testCase, stSortedSet_size(component) >= 1);
}
//Check no edges can be added from those filtered.
stSortedSet *filteredEdgesSet = stList_getSortedSet(filteredEdges, (int(*)(const void *, const void *)) stIntTuple_cmpFn);
for (int64_t i = 0; i < stList_length(edges); i++) {
stIntTuple *edge = stList_get(edges, i);
if (stSortedSet_search(filteredEdgesSet, edge) == NULL) {
stIntTuple *node1 = stIntTuple_construct1( stIntTuple_get(edge, 1));
stIntTuple *node2 = stIntTuple_construct1( stIntTuple_get(edge, 2));
stSortedSet *component1 = stHash_search(nodesToComponents, node1);
stSortedSet *component2 = stHash_search(nodesToComponents, node2);
CuAssertTrue(testCase, component1 != NULL && component2 != NULL);
CuAssertTrue(testCase, component1 != component2);
CuAssertTrue(testCase, stSortedSet_size(component1) + stSortedSet_size(component2) > maxComponentSize);
stIntTuple_destruct(node1);
stIntTuple_destruct(node2);
}
}
stSortedSet_destruct(filteredEdgesSet);
//Cleanup the components
stSortedSet *componentsSet = stList_getSortedSet(components, NULL);
stList_destruct(components);
stSortedSet_setDestructor(componentsSet, (void(*)(void *)) stSortedSet_destruct);
stSortedSet_destruct(componentsSet);
stHash_destruct(nodesToComponents);
}
static void getMetaSequencesForEventsP(stSortedSet *metaSequences,
Flower *flower, stList *eventStrings) {
//Iterate over the sequences in the flower.
Flower_SequenceIterator *seqIt = flower_getSequenceIterator(flower);
Sequence *sequence;
while ((sequence = flower_getNextSequence(seqIt)) != NULL) {
MetaSequence *metaSequence = sequence_getMetaSequence(sequence);
if (stringIsInList(event_getHeader(sequence_getEvent(sequence)),
eventStrings) == 0) {
if (stSortedSet_search(metaSequences, metaSequence) == NULL) {
stSortedSet_insert(metaSequences, metaSequence);
}
}
}
flower_destructSequenceIterator(seqIt);
//Recurse over the flowers
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (group_getNestedFlower(group) != NULL) {
getMetaSequencesForEventsP(metaSequences,
group_getNestedFlower(group), eventStrings);
}
}
flower_destructGroupIterator(groupIt);
}
Sequence *flower_getSequence(Flower *flower, Name name) {
Sequence sequence;
MetaSequence metaSequence;
sequence.metaSequence = &metaSequence;
metaSequence.name = name;
return stSortedSet_search(flower->sequences, &sequence);
}
Cap *flower_getCap(Flower *flower, Name name) {
Cap cap;
CapContents capContents;
cap.capContents = &capContents;
cap.capContents->instance = name;
return stSortedSet_search(flower->caps, &cap);
}
End *flower_getEnd(Flower *flower, Name name) {
End end;
EndContents endContents;
end.endContents = &endContents;
endContents.name = name;
end.orientation = 1;
return stSortedSet_search(flower->ends, &end);
}
static void test_stSortedSetIntersection(CuTest* testCase) {
sonLibSortedSetTestSetup();
//Check intersection of empty sets is okay..
stSortedSet *sortedSet3 = stSortedSet_getIntersection(sortedSet, sortedSet2);
CuAssertTrue(testCase, stSortedSet_size(sortedSet3) == 0);
stSortedSet_destruct(sortedSet3);
int32_t i;
for(i=0; i<size; i++) {
stSortedSet_insert(sortedSet, stIntTuple_construct(1, input[i]));
}
//Check intersection of empty and non-empty set is empty.
sortedSet3 = stSortedSet_getIntersection(sortedSet, sortedSet2);
CuAssertTrue(testCase, stSortedSet_size(sortedSet3) == 0);
stSortedSet_destruct(sortedSet3);
//Check intersection of two non-empty, overlapping sets in correct.
stSortedSet_insert(sortedSet2, stIntTuple_construct(1, 0));
stSortedSet_insert(sortedSet2, stIntTuple_construct(1, 1));
stSortedSet_insert(sortedSet2, stIntTuple_construct(1, 5));
sortedSet3 = stSortedSet_getIntersection(sortedSet, sortedSet2);
CuAssertTrue(testCase, stSortedSet_size(sortedSet3) == 2);
stIntTuple *intTuple = stIntTuple_construct(1, 1);
CuAssertTrue(testCase, stSortedSet_search(sortedSet3, intTuple) != NULL);
stIntTuple_destruct(intTuple);
intTuple = stIntTuple_construct(1, 5);
CuAssertTrue(testCase, stSortedSet_search(sortedSet3, intTuple) != NULL);
stIntTuple_destruct(intTuple);
stSortedSet_destruct(sortedSet3);
//Check we get an exception with sorted sets with different comparators.
stSortedSet *sortedSet4 = stSortedSet_construct();
stTry {
stSortedSet_getIntersection(sortedSet, sortedSet4);
} stCatch(except) {
CuAssertTrue(testCase, stExcept_getId(except) == SORTED_SET_EXCEPTION_ID);
}
stTryEnd
stSortedSet_destruct(sortedSet4);
sonLibSortedSetTestTeardown();
}
void stSortedSet_insert(stSortedSet *sortedSet, void *object) {
checkModifiable(sortedSet);
// FIXME: two passes, modify avl code.
if(stSortedSet_search(sortedSet, object) != NULL) {
avl_replace(sortedSet->sortedSet, object);
}
else {
avl_insert(sortedSet->sortedSet, object);
}
}
static stList *getEdgesThatBridgeComponents(stList *components,
stHash *nodesToNonZeroWeightedAdjacencyEdges) {
/*
* Get set of adjacency edges that bridge between (have a node in two) components.
*/
stList *bridgingAdjacencyEdges = stList_construct();
for (int64_t i = 0; i < stList_length(components); i++) {
stSortedSet *componentNodes = getNodeSetOfEdges(
stList_get(components, i));
stSortedSetIterator *it = stSortedSet_getIterator(componentNodes);
stIntTuple *node;
while ((node = stSortedSet_getNext(it)) != NULL) {
stList *edges = stHash_search(nodesToNonZeroWeightedAdjacencyEdges,
node);
if (edges != NULL) {
for (int64_t j = 0; j < stList_length(edges); j++) {
stIntTuple *edge = stList_get(edges, j);
stIntTuple *node1 = stIntTuple_construct1(
stIntTuple_get(edge, 0));
stIntTuple *node2 = stIntTuple_construct1(
stIntTuple_get(edge, 1));
assert(
stSortedSet_search(componentNodes, node1) != NULL
|| stSortedSet_search(componentNodes, node2)
!= NULL);
if (stSortedSet_search(componentNodes, node1) == NULL
|| stSortedSet_search(componentNodes, node2)
== NULL) {
stList_append(bridgingAdjacencyEdges, edge);
}
stIntTuple_destruct(node1);
stIntTuple_destruct(node2);
}
}
}
stSortedSet_destructIterator(it);
stSortedSet_destruct(componentNodes);
}
return bridgingAdjacencyEdges;
}
static void test_stSortedSet(CuTest* testCase) {
sonLibSortedSetTestSetup();
int32_t i;
CuAssertIntEquals(testCase, 0, stSortedSet_size(sortedSet));
for(i=0; i<size; i++) {
stSortedSet_insert(sortedSet, stIntTuple_construct(1, input[i]));
}
CuAssertIntEquals(testCase, sortedSize, stSortedSet_size(sortedSet));
CuAssertIntEquals(testCase, sortedInput[0], stIntTuple_getPosition(stSortedSet_getFirst(sortedSet), 0));
CuAssertIntEquals(testCase, sortedInput[sortedSize-1], stIntTuple_getPosition(stSortedSet_getLast(sortedSet), 0));
for(i=0; i<sortedSize; i++) {
CuAssertIntEquals(testCase, sortedSize-i, stSortedSet_size(sortedSet));
stIntTuple *tuple = stIntTuple_construct(1, sortedInput[i]);
CuAssertTrue(testCase, stIntTuple_getPosition(stSortedSet_search(sortedSet, tuple), 0) == sortedInput[i]);
stSortedSet_remove(sortedSet, tuple);
CuAssertTrue(testCase, stSortedSet_search(sortedSet, tuple) == NULL);
stIntTuple_destruct(tuple);
}
sonLibSortedSetTestTeardown();
}
static void getMaximalHaplotypePathsP(Flower *flower,
stList *maximalHaplotypePaths, stSortedSet *segmentSet,
const char *eventString,
stList *eventStrings) {
/*
* Iterate through the segments in this flower.
*/
Flower_SegmentIterator *segmentIt = flower_getSegmentIterator(flower);
Segment *segment;
while ((segment = flower_getNextSegment(segmentIt)) != NULL) {
if (stSortedSet_search(segmentSet, segment) == NULL
&& stSortedSet_search(segmentSet, segment_getReverse(segment))
== NULL) { //Check we haven't yet seen this segment
if (strcmp(event_getHeader(segment_getEvent(segment)), eventString)
== 0) { //Check if the segment is in the assembly
if (hasCapInEvents(cap_getEnd(segment_get5Cap(segment)), eventStrings)) { //Is a block in a haplotype segment
assert(hasCapInEvents(cap_getEnd(segment_get3Cap(segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get3Cap(segment))));
stList *maximalHaplotypePath = stList_construct();
stList_append(maximalHaplotypePaths, maximalHaplotypePath);
getMaximalHaplotypePathsP2(segment, maximalHaplotypePath,
segmentSet, eventStrings);
} else {
assert(!hasCapInEvents(cap_getEnd(segment_get3Cap(segment)), eventStrings));//assert(!isHaplotypeEnd(cap_getEnd(segment_get3Cap(segment))));
}
}
}
}
flower_destructSegmentIterator(segmentIt);
/*
* Now recurse on the contained flowers.
*/
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (group_getNestedFlower(group) != NULL) {
getMaximalHaplotypePathsP(group_getNestedFlower(group),
maximalHaplotypePaths, segmentSet, eventString, eventStrings);
}
}
flower_destructGroupIterator(groupIt);
}
static int64_t intersectionSize(stSortedSet *set, stList *list) {
/*
* Returns the intersection size of the list and the set.
*/
int64_t count = 0;
for (int64_t j = 0; j < stList_length(list); j++) {
if (stSortedSet_search(set, stList_get(list, j)) != NULL) {
count++;
}
}
return count;
}
void testCactusDisk_getUniqueID_Unique(CuTest* testCase) {
cactusDiskTestSetup();
stSortedSet *uniqueNames = stSortedSet_construct3(testCactusDisk_getUniqueID_UniqueP, free);
for (int64_t i = 0; i < 100000; i++) { //Gets a billion ids, checks we are good.
Name uniqueName = cactusDisk_getUniqueID(cactusDisk);
CuAssertTrue(testCase, uniqueName > 0);
CuAssertTrue(testCase, uniqueName < INT64_MAX);
CuAssertTrue(testCase, uniqueName != NULL_NAME);
char *cA = cactusMisc_nameToString(uniqueName);
CuAssertTrue(testCase, stSortedSet_search(uniqueNames, cA) == NULL);
CuAssertTrue(testCase, cactusMisc_stringToName(cA) == uniqueName);
stSortedSet_insert(uniqueNames, cA);
}
stSortedSet_destruct(uniqueNames);
cactusDiskTestTeardown();
}
stSortedSet *stSortedSet_getDifference(stSortedSet *sortedSet1, stSortedSet *sortedSet2) {
if(!stSortedSet_comparatorsEqual(sortedSet1, sortedSet2)) {
stThrowNew(SORTED_SET_EXCEPTION_ID, "Comparators are not equal for creating the sorted set difference");
}
stSortedSet *sortedSet3 = stSortedSet_construct3(stSortedSet_getComparator(sortedSet1)->compareFn, NULL);
//Add those from sortedSet1 only if they are not in sortedSet2
stSortedSetIterator *it= stSortedSet_getIterator(sortedSet1);
void *o;
while((o = stSortedSet_getNext(it)) != NULL) {
if(stSortedSet_search(sortedSet2, o) == NULL) {
stSortedSet_insert(sortedSet3, o);
}
}
stSortedSet_destructIterator(it);
return sortedSet3;
}
void testCactusDisk_getUniqueID_UniqueIntervals(CuTest* testCase) {
cactusDiskTestSetup();
stSortedSet *uniqueNames = stSortedSet_construct3(testCactusDisk_getUniqueID_UniqueP, free);
for (int64_t i = 0; i < 10; i++) { //Gets a billion ids, checks we are good.
int64_t intervalSize = st_randomInt(0, 100000);
Name uniqueName = cactusDisk_getUniqueIDInterval(cactusDisk, intervalSize);
for(int64_t j=0; j<intervalSize; j++) {
CuAssertTrue(testCase, uniqueName > 0);
CuAssertTrue(testCase, uniqueName < INT64_MAX);
CuAssertTrue(testCase, uniqueName != NULL_NAME);
char *cA = cactusMisc_nameToString(uniqueName);
CuAssertTrue(testCase, stSortedSet_search(uniqueNames, cA) == NULL);
CuAssertTrue(testCase, cactusMisc_stringToName(cA) == uniqueName);
stSortedSet_insert(uniqueNames, cA);
uniqueName++;
}
}
stSortedSet_destruct(uniqueNames);
cactusDiskTestTeardown();
}
/*
* This builds an adjacency list structure for the the sequences. Every sequence-position
* has a column in the hash with which it can be aligned with.
*/
static stHash *buildAdjacencyList(stList *pairs, int64_t sequenceNumber) {
stHash *hash = stHash_construct3((uint64_t (*)(const void *))stIntTuple_hashKey,
(int (*)(const void *, const void *))stIntTuple_equalsFn,
(void (*)(void *))stIntTuple_destruct, NULL);
for(int64_t seq=0; seq<sequenceNumber; seq++) {
for(int64_t position=0; position<MAX_SEQUENCE_SIZE; position++) {
stIntTuple *seqPos = stIntTuple_construct2( seq, position);
stSortedSet *column = stSortedSet_construct3((int (*)(const void *, const void *))stIntTuple_cmpFn, NULL);
stSortedSet_insert(column, seqPos);
stHash_insert(hash, seqPos, column);
}
}
stListIterator *it = stList_getIterator(pairs);
stIntTuple *pair;
while((pair = stList_getNext(it)) != NULL) {
stIntTuple *seqPos1 = stIntTuple_construct2( stIntTuple_get(pair, 0), stIntTuple_get(pair, 1));
stIntTuple *seqPos2 = stIntTuple_construct2( stIntTuple_get(pair, 2), stIntTuple_get(pair, 3));
stSortedSet *column1 = stHash_search(hash, seqPos1);
assert(column1 != NULL);
stSortedSet *column2 = stHash_search(hash, seqPos2);
assert(column2 != NULL);
if(column1 != column2) { //Merge the columns
stSortedSetIterator *it2 = stSortedSet_getIterator(column2);
stIntTuple *seqPos3;
while((seqPos3 = stSortedSet_getNext(it2)) != NULL) {
assert(stSortedSet_search(column1, seqPos3) == NULL);
stSortedSet_insert(column1, seqPos3);
assert(stHash_search(hash, seqPos3) == column2);
stHash_insert(hash, seqPos3, column1);
assert(stHash_search(hash, seqPos3) == column1);
}
stSortedSet_destructIterator(it2);
stSortedSet_destruct(column2);
}
//Cleanup loop.
stIntTuple_destruct(seqPos1);
stIntTuple_destruct(seqPos2);
}
stList_destructIterator(it);
return hash;
}
static void getOrderedSegmentsP(Flower *flower,
stSortedSet *segments) {
Flower_SegmentIterator *segmentIt = flower_getSegmentIterator(flower);
Segment *segment;
while ((segment = flower_getNextSegment(segmentIt)) != NULL) {
if (!segment_getStrand(segment)) {
segment = segment_getReverse(segment);
}
assert(stSortedSet_search(segments, segment) == NULL);
stSortedSet_insert(segments, segment);
}
flower_destructSegmentIterator(segmentIt);
//Recurse over the flowers
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (group_getNestedFlower(group) != NULL) {
getOrderedSegmentsP(group_getNestedFlower(group),
segments);
}
}
flower_destructGroupIterator(groupIt);
}
static void getComponentsP(stHash *nodesToEdges, int64_t node,
stSortedSet *component) {
stIntTuple *key = stIntTuple_construct1( node);
stList *edges = stHash_search(nodesToEdges, key);
if (edges != NULL) {
stHash_remove(nodesToEdges, key);
for (int64_t i = 0; i < stList_length(edges); i++) {
stIntTuple *edge = stList_get(edges, i);
if (stSortedSet_search(component, edge) == NULL) {
stSortedSet_insert(component, edge);
}
/*
* Recursion on stack could equal the total number of nodes.
*/
getComponentsP(nodesToEdges, stIntTuple_get(edge, 0),
component);
getComponentsP(nodesToEdges, stIntTuple_get(edge, 1),
component);
}
stList_destruct(edges);
}
stIntTuple_destruct(key);
}
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 *block_getInstance(Block *block, Name name) {
Segment segment;
segment.name = name;
return block_getInstanceP(block, stSortedSet_search(block->blockContents->segments, &segment));
}
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;
}
Event *eventTree_getEvent(EventTree *eventTree, Name eventName) {
Event event;
event.name = eventName;
return stSortedSet_search(eventTree->events, &event);
}
void flower_removeFace(Flower *flower, Face *face) {
assert(stSortedSet_search(flower->faces, face) != NULL);
stSortedSet_remove(flower->faces, face);
}