static stHash *getComponents(stList *filteredEdges) {
/*
* A kind of stupid reimplementation of the greedy function, done just to trap typos.
*/
stHash *nodesToComponents = stHash_construct3((uint64_t(*)(const void *)) stIntTuple_hashKey,
(int(*)(const void *, const void *)) stIntTuple_equalsFn, NULL, NULL);
for (int64_t i = 0; i < stList_length(nodes); i++) {
stIntTuple *node = stList_get(nodes, i);
stSortedSet *component = stSortedSet_construct();
stSortedSet_insert(component, node);
stHash_insert(nodesToComponents, node, component);
}
for (int64_t i = 0; i < stList_length(filteredEdges); i++) {
stIntTuple *edge = stList_get(filteredEdges, i);
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);
assert(component1 != NULL && component2 != NULL);
if (component1 != component2) {
stSortedSet *component3 = stSortedSet_getUnion(component1, component2);
stSortedSetIterator *setIt = stSortedSet_getIterator(component3);
stIntTuple *node3;
while ((node3 = stSortedSet_getNext(setIt)) != NULL) {
stHash_insert(nodesToComponents, node3, component3);
}
stSortedSet_destructIterator(setIt);
stSortedSet_destruct(component1);
stSortedSet_destruct(component2);
}
stIntTuple_destruct(node1);
stIntTuple_destruct(node2);
}
return nodesToComponents;
}
static void test_stSet_search(CuTest* testCase) {
testSetup();
stIntTuple *i = stIntTuple_construct1( 0);
stIntTuple *j = stIntTuple_construct2(10, 0);
stIntTuple *k = stIntTuple_construct1( 5);
//Check search by memory address
CuAssertTrue(testCase, stSet_search(set0, one) == one);
CuAssertTrue(testCase, stSet_search(set0, two) == two);
CuAssertTrue(testCase, stSet_search(set0, three) == three);
CuAssertTrue(testCase, stSet_search(set0, four) == four);
CuAssertTrue(testCase, stSet_search(set0, five) == five);
CuAssertTrue(testCase, stSet_search(set0, six) == six);
//Check not present
CuAssertTrue(testCase, stSet_search(set0, i) == NULL);
CuAssertTrue(testCase, stSet_search(set0, j) == NULL);
CuAssertTrue(testCase, stSet_search(set0, k) == NULL);
//Check search by memory address
CuAssertTrue(testCase, stSet_search(set1, one) == one);
CuAssertTrue(testCase, stSet_search(set1, two) == two);
CuAssertTrue(testCase, stSet_search(set1, three) == three);
CuAssertTrue(testCase, stSet_search(set1, four) == four);
CuAssertTrue(testCase, stSet_search(set1, five) == five);
CuAssertTrue(testCase, stSet_search(set1, six) == six);
//Check not present
CuAssertTrue(testCase, stSet_search(set1, j) == NULL);
//Check is searching by memory
CuAssertTrue(testCase, stSet_search(set1, i) == one);
CuAssertTrue(testCase, stSet_search(set1, k) == six);
stIntTuple_destruct(i);
stIntTuple_destruct(j);
stIntTuple_destruct(k);
testTeardown();
}
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 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();
}
/*
* 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;
}
/*
* 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 stList *readMatching(FILE *fileHandle, stList *originalEdges) {
/*
* Reads the matching created by Blossum.
*/
stHash *originalEdgesHash = putEdgesInHash(originalEdges);
char *line = stFile_getLineFromFile(fileHandle);
assert(line != NULL);
int64_t nodeNumber, edgeNumber;
int64_t i = sscanf(line, "%" PRIi64 " %" PRIi64 "\n", &nodeNumber, &edgeNumber);
assert(i == 2);
free(line);
stList *chosenEdges = stList_construct();
for(int64_t j=0; j<edgeNumber; j++) {
line = stFile_getLineFromFile(fileHandle);
int64_t node1, node2;
i = sscanf(line, "%" PRIi64 " %" PRIi64 "", &node1, &node2);
assert(i == 2);
free(line);
assert(node1 >= 0);
assert(node1 < nodeNumber);
assert(node2 >= 0);
assert(node2 < nodeNumber);
stIntTuple *edge = constructEdge(node1, node2);
stIntTuple *originalEdge = stHash_search(originalEdgesHash, edge);
if(originalEdge != NULL) {
stList_append(chosenEdges, originalEdge);
}
stIntTuple_destruct(edge);
}
stHash_destruct(originalEdgesHash);
return chosenEdges;
}
/*
* Gets the position in sequence2 that the position in sequence1 must be greater than or equal to in the alignment
*/
static stIntTuple *getConstraint_greaterThan(stPosetAlignment *posetAlignment, int64_t sequence1, int64_t position1, int64_t sequence2) {
stIntTuple *pos = stIntTuple_construct2(INT64_MAX, position1);
//Get less than or equal
stIntTuple *constraint = stSortedSet_searchLessThanOrEqual(getConstraintList(posetAlignment, sequence2, sequence1), pos);
stIntTuple_destruct(pos);
assert(constraint == NULL || position1 >= stIntTuple_get(constraint, 1));
return constraint;
}
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();
}
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_stPosetAlignment_addAndIsPossible(CuTest *testCase) {
for(int64_t trial=0; trial<100; trial++) {
setup();
//Make random number of sequences.
stList *sequenceLengths = stList_construct3(0, (void (*)(void *))stIntTuple_destruct);
for(int64_t i=0; i<sequenceNumber; i++) {
stList_append(sequenceLengths, stIntTuple_construct1( st_randomInt(0, MAX_SEQUENCE_SIZE)));
}
//Propose random alignment pairs...
stList *pairs = stList_construct3(0, (void(*)(void *))stIntTuple_destruct);
int64_t maxAlignedPairs = st_randomInt(0, MAX_ALIGNMENTS);
if(sequenceNumber > 0) {
for(int64_t i=0; i<maxAlignedPairs; i++) {
int64_t seq1 = st_randomInt(0, sequenceNumber);
int64_t seqLength1 = stIntTuple_get(stList_get(sequenceLengths, seq1), 0);
if(seqLength1 == 0) {
continue;
}
int64_t position1 = st_randomInt(0, seqLength1);
int64_t seq2 = st_randomInt(0, sequenceNumber);
int64_t seqLength2 = stIntTuple_get(stList_get(sequenceLengths, seq1), 0);
if(seqLength2 == 0) {
continue;
}
int64_t position2 = st_randomInt(0, seqLength2);
if(seq1 != seq2) {
stList_append(pairs, stIntTuple_construct4( seq1, position1, seq2, position2));
if(stPosetAlignment_isPossible(posetAlignment, seq1, position1, seq2, position2)) {
st_logInfo("In %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " \n", seq1, position1, seq2, position2);
//For each accepted pair check it doesn't create a cycle.
CuAssertTrue(testCase, !containsACycle(pairs, sequenceNumber));
CuAssertTrue(testCase, stPosetAlignment_add(posetAlignment, seq1, position1, seq2, position2));
}
else {
st_logInfo("Out %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " \n", seq1, position1, seq2, position2);
//For each rejected pair check it creates a cycle..
CuAssertTrue(testCase, containsACycle(pairs, sequenceNumber));
CuAssertTrue(testCase, !stPosetAlignment_isPossible(posetAlignment, seq1, position1, seq2, position2));
stIntTuple_destruct(stList_pop(pairs)); //remove the pair which created the cycle.
CuAssertTrue(testCase, !containsACycle(pairs, sequenceNumber)); //Check we're back to being okay..
}
}
}
}
//Cleanup
stList_destruct(sequenceLengths);
stList_destruct(pairs);
teardown();
st_logInfo("Passed a random ordering test with %" PRIi64 " sequences and %" PRIi64 " aligned pairs\n", sequenceNumber, maxAlignedPairs);
}
}
/*
* 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 test_stSet_insert(CuTest* testCase) {
/*
* Tests inserting already present keys.
*/
testSetup();
CuAssertTrue(testCase, stSet_search(set0, one) == one);
stSet_insert(set0, one);
CuAssertTrue(testCase, stSet_search(set0, one) == one);
stSet_insert(set0, three);
CuAssertTrue(testCase, stSet_search(set0, three) == three);
stIntTuple *seven = stIntTuple_construct2(7, 7);
CuAssertTrue(testCase, stSet_search(set0, seven) == NULL);
stSet_insert(set0, seven);
CuAssertTrue(testCase, stSet_search(set0, seven) == seven);
stIntTuple_destruct(seven);
testTeardown();
}
// copied from cPecanRealign
struct PairwiseAlignment *convertAlignedPairsToPairwiseAlignment(char *seqName1, char *seqName2, double score,
int64_t length1, int64_t length2, stList *alignedPairs) {
//Make pairwise alignment
int64_t pX = -1, pY = -1, mL = 0;
//Create an end matched pair, which is used to ensure the alignment has the correct end indels.
struct List *opList = constructEmptyList(0, (void (*)(void *)) destructAlignmentOperation);
stList_append(alignedPairs, stIntTuple_construct2(length1, length2));
for (int64_t i = 0; i < stList_length(alignedPairs); i++) {
stIntTuple *alignedPair = stList_get(alignedPairs, i);
int64_t x = stIntTuple_get(alignedPair, 0);
int64_t y = stIntTuple_get(alignedPair, 1);
assert(x - pX > 0);
assert(y - pY > 0);
if (x - pX > 0 && y - pY > 0) { //This is a hack for filtering
if (x - pX > 1) { //There is an indel.
if (mL > 0) {
listAppend(opList, constructAlignmentOperation(PAIRWISE_MATCH, mL, 0));
mL = 0;
}
listAppend(opList, constructAlignmentOperation(PAIRWISE_INDEL_X, x - pX - 1, 0));
}
if (y - pY > 1) {
if (mL > 0) {
listAppend(opList, constructAlignmentOperation(PAIRWISE_MATCH, mL, 0));
mL = 0;
}
listAppend(opList, constructAlignmentOperation(PAIRWISE_INDEL_Y, y - pY - 1, 0));
}
mL++;
pX = x;
pY = y;
}
}
//Deal with a trailing match, but exclude the final match
if (mL > 1) {
listAppend(opList, constructAlignmentOperation(PAIRWISE_MATCH, mL - 1, 0));
}
stIntTuple_destruct(stList_pop(alignedPairs));
//Construct the alignment
struct PairwiseAlignment *pA = constructPairwiseAlignment(seqName1, 0, length1, 1, seqName2, 0, length2, 1, score,
opList);
return pA;
}
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();
}
stList *getComponents(stList *edges) {
/*
* Gets a list of connected components, each connected component
* being represented as a list of the edges, such that each edge is in exactly one
* connected component. Allows for multi-graphs (multiple edges connecting two nodes).
*/
stHash *nodesToEdges = getNodesToEdgesHash(edges);
/*
* Traverse the edges greedily
*/
stList *components =
stList_construct3(0, (void(*)(void *)) stList_destruct);
stList *nodes = stHash_getKeys(nodesToEdges);
while (stList_length(nodes) > 0) {
stIntTuple *node = stList_pop(nodes);
stList *edges = stHash_search(nodesToEdges, node);
if (edges != NULL) { //We have a component to build
stSortedSet *component = stSortedSet_construct();
stHash_remove(nodesToEdges, node);
for (int64_t i = 0; i < stList_length(edges); i++) {
stIntTuple *edge = stList_get(edges, i);
getComponentsP(nodesToEdges, stIntTuple_get(edge, 0),
component);
getComponentsP(nodesToEdges, stIntTuple_get(edge, 1),
component);
}
stList_append(components, stSortedSet_getList(component));
//Cleanup
stSortedSet_destruct(component);
stList_destruct(edges);
}
stIntTuple_destruct(node);
}
assert(stHash_size(nodesToEdges) == 0);
stHash_destruct(nodesToEdges);
stList_destruct(nodes);
return components;
}
/*
* Adds a prime less than (or equals) constraint to the list of prime constraints, removing any redundant constraints in the process.
* The or equals is specified by making the lessThanOrEquals argument non-zero.
*/
void addConstraint_lessThan(stPosetAlignment *posetAlignment, int64_t sequence1, int64_t position1, int64_t sequence2, int64_t position2, int64_t lessThanOrEquals) {
stSortedSet *constraintList = getConstraintList(posetAlignment, sequence1, sequence2);
assert(position1 != INT64_MAX);
assert(position2 != INT64_MAX);
stIntTuple *constraint1 = stIntTuple_construct3( position1, position2, lessThanOrEquals);
stIntTuple *constraint2;
while((constraint2 = stSortedSet_searchLessThanOrEqual(constraintList, constraint1)) != NULL) {
assert(stIntTuple_get(constraint2, 0) <= position1);
if(stIntTuple_get(constraint2, 1) >= position2) {
if(stIntTuple_get(constraint2, 1) == position2) { //Check we are not removing an equivalent or more severe constraint.
assert((!lessThanOrEquals && stIntTuple_get(constraint2, 2)) || stIntTuple_get(constraint2, 0) < position1);
}
stSortedSet_remove(constraintList, constraint2);
stIntTuple_destruct(constraint2);
}
else {
assert(stIntTuple_get(constraint2, 0) < position1); //Check the constraint does not overshadow our proposed constraint.
break;
}
}
stSortedSet_insert(constraintList, constraint1);
}
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;
}
static stHash *getScaffoldPathsP(stList *haplotypePaths, stHash *haplotypePathToScaffoldPathHash,
stList *haplotypeEventStrings, stList *contaminationEventStrings, CapCodeParameters *capCodeParameters) {
stHash *haplotypeToMaximalHaplotypeLengthHash = buildContigPathToContigPathLengthHash(haplotypePaths);
stHash *segmentToMaximalHaplotypePathHash = buildSegmentToContigPathHash(haplotypePaths);
for (int64_t i = 0; i < stList_length(haplotypePaths); i++) {
stSortedSet *bucket = stSortedSet_construct();
stHash_insert(haplotypePathToScaffoldPathHash, stList_get(haplotypePaths, i), bucket);
stSortedSet_insert(bucket, stList_get(haplotypePaths, i));
}
for (int64_t i = 0; i < stList_length(haplotypePaths); i++) {
stList *haplotypePath = stList_get(haplotypePaths, i);
assert(stList_length(haplotypePath) > 0);
Segment *_5Segment = stList_get(haplotypePath, 0);
if (!segment_getStrand(_5Segment)) {
_5Segment = segment_getReverse(stList_get(haplotypePath, stList_length(haplotypePath) - 1));
}
assert(segment_getStrand(_5Segment));
if (getAdjacentCapsSegment(segment_get5Cap(_5Segment)) != NULL) {
assert(!trueAdjacency(segment_get5Cap(_5Segment), haplotypeEventStrings));
}
int64_t insertLength;
int64_t deleteLength;
Cap *otherCap;
enum CapCode _5CapCode = getCapCode(segment_get5Cap(_5Segment), &otherCap, haplotypeEventStrings, contaminationEventStrings, &insertLength, &deleteLength, capCodeParameters);
if (_5CapCode == SCAFFOLD_GAP || _5CapCode == AMBIGUITY_GAP) {
assert(stHash_search(haplotypeToMaximalHaplotypeLengthHash, haplotypePath) != NULL);
int64_t j = stIntTuple_get(stHash_search(haplotypeToMaximalHaplotypeLengthHash, haplotypePath), 0);
Segment *adjacentSegment = getAdjacentCapsSegment(segment_get5Cap(_5Segment));
assert(adjacentSegment != NULL);
while (!hasCapInEvents(cap_getEnd(segment_get5Cap(adjacentSegment)), haplotypeEventStrings)) { //is not a haplotype end
adjacentSegment = getAdjacentCapsSegment(segment_get5Cap(adjacentSegment));
assert(adjacentSegment != NULL);
}
assert(adjacentSegment != NULL);
assert(hasCapInEvents(cap_getEnd(segment_get5Cap(adjacentSegment)), haplotypeEventStrings)); //is a haplotype end
stList *adjacentHaplotypePath = stHash_search(segmentToMaximalHaplotypePathHash, adjacentSegment);
if (adjacentHaplotypePath == NULL) {
adjacentHaplotypePath = stHash_search(segmentToMaximalHaplotypePathHash, segment_getReverse(
adjacentSegment));
}
assert(adjacentHaplotypePath != NULL);
assert(adjacentHaplotypePath != haplotypePath);
assert(stHash_search(haplotypeToMaximalHaplotypeLengthHash, adjacentHaplotypePath) != NULL);
int64_t k = stIntTuple_get(stHash_search(haplotypeToMaximalHaplotypeLengthHash, adjacentHaplotypePath), 0);
//Now merge the buckets and make new int tuples..
stSortedSet *bucket1 = stHash_search(haplotypePathToScaffoldPathHash, haplotypePath);
stSortedSet *bucket2 = stHash_search(haplotypePathToScaffoldPathHash, adjacentHaplotypePath);
assert(bucket1 != NULL);
assert(bucket2 != NULL);
assert(bucket1 != bucket2);
stSortedSet *bucket3 = stSortedSet_getUnion(bucket1, bucket2);
stSortedSetIterator *bucketIt = stSortedSet_getIterator(bucket3);
stList *l;
while ((l = stSortedSet_getNext(bucketIt)) != NULL) {
//Do the bucket first
assert(stHash_search(haplotypePathToScaffoldPathHash, l) == bucket1 || stHash_search(haplotypePathToScaffoldPathHash, l) == bucket2);
stHash_remove(haplotypePathToScaffoldPathHash, l);
stHash_insert(haplotypePathToScaffoldPathHash, l, bucket3);
//Now the length
stIntTuple *m = stHash_remove(haplotypeToMaximalHaplotypeLengthHash, l);
assert(m != NULL);
assert(stIntTuple_get(m, 0) == j || stIntTuple_get(m, 0) == k);
stHash_insert(haplotypeToMaximalHaplotypeLengthHash, l, stIntTuple_construct1( j + k));
stIntTuple_destruct(m);
}
assert(stHash_search(haplotypePathToScaffoldPathHash, haplotypePath) == bucket3);
assert(stHash_search(haplotypePathToScaffoldPathHash, adjacentHaplotypePath) == bucket3);
stSortedSet_destructIterator(bucketIt);
}
}
stHash_destruct(segmentToMaximalHaplotypePathHash);
return haplotypeToMaximalHaplotypeLengthHash;
}
static void readWriteAndRemoveRecordsLotsIteration(CuTest *testCase, int numRecords, bool reopenDatabase) {
//Make a big old list of records..
stSortedSet *set = stSortedSet_construct3((int(*)(const void *, const void *)) stIntTuple_cmpFn,
(void(*)(void *)) stIntTuple_destruct);
while (stSortedSet_size(set) < numRecords) {
int32_t key = st_randomInt(0, 100 * numRecords);
stIntTuple *tuple = stIntTuple_construct(1, key);
if (stSortedSet_search(set, tuple) == NULL) {
CuAssertTrue(testCase, !stKVDatabase_containsRecord(database, key));
stSortedSet_insert(set, tuple);
stKVDatabase_insertRecord(database, key, &key, sizeof(int32_t));
CuAssertTrue(testCase, stKVDatabase_containsRecord(database, key));
} else {
CuAssertTrue(testCase, stKVDatabase_containsRecord(database, key));
stIntTuple_destruct(tuple); // already in db
}
}
readWriteAndRemoveRecordsLotsCheck(testCase, set, 1);
//Update all records to negate values
stSortedSetIterator *it = stSortedSet_getIterator(set);
stIntTuple *tuple;
while ((tuple = stSortedSet_getNext(it)) != NULL) {
int32_t *value = (int32_t *) stKVDatabase_getRecord(database, stIntTuple_getPosition(tuple, 0));
*value *= -1;
stKVDatabase_updateRecord(database, stIntTuple_getPosition(tuple, 0), value, sizeof(int32_t));
CuAssertTrue(testCase, stKVDatabase_containsRecord(database, stIntTuple_getPosition(tuple, 0)));
free(value);
}
stSortedSet_destructIterator(it);
readWriteAndRemoveRecordsLotsCheck(testCase, set, -1);
//Try optionally committing the transaction and reloading the database..
if (reopenDatabase) {
//stKVDatabase_commitTransaction(database);
stKVDatabase_destruct(database);
database = stKVDatabase_construct(conf, false);
//stKVDatabase_startTransaction(database);
}
//Now remove each one..
it = stSortedSet_getIterator(set);
while ((tuple = stSortedSet_getNext(it)) != NULL) {
CuAssertTrue(testCase, stKVDatabase_containsRecord(database, stIntTuple_getPosition(tuple, 0)));
stKVDatabase_removeRecord(database, stIntTuple_getPosition(tuple, 0));
CuAssertTrue(testCase, !stKVDatabase_containsRecord(database, stIntTuple_getPosition(tuple, 0)));
//Test we get exception if we remove twice.
stTry {
stKVDatabase_removeRecord(database, stIntTuple_getPosition(tuple, 0));
CuAssertTrue(testCase, 0);
}
stCatch(except)
{
CuAssertTrue(testCase, stExcept_getId(except) == ST_KV_DATABASE_EXCEPTION_ID);
}stTryEnd;
}
stSortedSet_destructIterator(it);
CuAssertIntEquals(testCase, 0, stKVDatabase_getNumberOfRecords(database));
stSortedSet_destruct(set);
}
// copied from cPecanRealign, which is sloppy.
void *convertToAnchorPair(void *aPair, void *extraArg) {
stIntTuple *i = stIntTuple_construct2(stIntTuple_get(aPair, 1), stIntTuple_get(aPair, 2));
stIntTuple_destruct(aPair);
return i;
}
