void flower_destruct(Flower *flower, int64_t recursive) {
Flower_GroupIterator *iterator;
Sequence *sequence;
End *end;
Block *block;
Group *group;
Chain *chain;
Flower *nestedFlower;
if (recursive) {
iterator = flower_getGroupIterator(flower);
while ((group = flower_getNextGroup(iterator)) != NULL) {
nestedFlower = group_getNestedFlower(group);
if (nestedFlower != NULL) {
flower_destruct(nestedFlower, recursive);
}
}
flower_destructGroupIterator(iterator);
}
cactusDisk_removeFlower(flower->cactusDisk, flower);
flower_destructFaces(flower);
stSortedSet_destruct(flower->faces);
assert(flower_getEventTree(flower) != NULL);
eventTree_destruct(flower_getEventTree(flower));
while ((sequence = flower_getFirstSequence(flower)) != NULL) {
sequence_destruct(sequence);
}
stSortedSet_destruct(flower->sequences);
while ((chain = flower_getFirstChain(flower)) != NULL) {
chain_destruct(chain);
}
stSortedSet_destruct(flower->chains);
while ((end = flower_getFirstEnd(flower)) != NULL) {
end_destruct(end);
}
stSortedSet_destruct(flower->caps);
stSortedSet_destruct(flower->ends);
while ((block = flower_getFirstBlock(flower)) != NULL) {
block_destruct(block);
}
stSortedSet_destruct(flower->segments);
stSortedSet_destruct(flower->blocks);
while ((group = flower_getFirstGroup(flower)) != NULL) {
group_destruct(group);
}
stSortedSet_destruct(flower->groups);
free(flower);
}
void flower_writeBinaryRepresentation(Flower *flower, void(*writeFn)(const void * ptr, size_t size, size_t count)) {
Flower_SequenceIterator *sequenceIterator;
Flower_EndIterator *endIterator;
Flower_BlockIterator *blockIterator;
Flower_GroupIterator *groupIterator;
Flower_ChainIterator *chainIterator;
Sequence *sequence;
End *end;
Block *block;
Group *group;
Chain *chain;
binaryRepresentation_writeElementType(CODE_FLOWER, writeFn);
binaryRepresentation_writeName(flower_getName(flower), writeFn);
binaryRepresentation_writeBool(flower_builtBlocks(flower), writeFn);
binaryRepresentation_writeBool(flower_builtTrees(flower), writeFn);
binaryRepresentation_writeBool(flower_builtFaces(flower), writeFn);
binaryRepresentation_writeName(flower->parentFlowerName, writeFn);
if (flower_getEventTree(flower) != NULL) {
eventTree_writeBinaryRepresentation(flower_getEventTree(flower), writeFn);
}
sequenceIterator = flower_getSequenceIterator(flower);
while ((sequence = flower_getNextSequence(sequenceIterator)) != NULL) {
sequence_writeBinaryRepresentation(sequence, writeFn);
}
flower_destructSequenceIterator(sequenceIterator);
endIterator = flower_getEndIterator(flower);
while ((end = flower_getNextEnd(endIterator)) != NULL) {
end_writeBinaryRepresentation(end, writeFn);
}
flower_destructEndIterator(endIterator);
blockIterator = flower_getBlockIterator(flower);
while ((block = flower_getNextBlock(blockIterator)) != NULL) {
block_writeBinaryRepresentation(block, writeFn);
}
flower_destructBlockIterator(blockIterator);
groupIterator = flower_getGroupIterator(flower);
while ((group = flower_getNextGroup(groupIterator)) != NULL) {
group_writeBinaryRepresentation(group, writeFn);
}
flower_destructGroupIterator(groupIterator);
chainIterator = flower_getChainIterator(flower);
while ((chain = flower_getNextChain(chainIterator)) != NULL) {
chain_writeBinaryRepresentation(chain, writeFn);
}
flower_destructChainIterator(chainIterator);
binaryRepresentation_writeElementType(CODE_FLOWER, writeFn); //this avoids interpretting things wrong.
}
void flower_check(Flower *flower) {
eventTree_check(flower_getEventTree(flower));
Flower_GroupIterator *groupIterator = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIterator)) != NULL) {
group_check(group);
}
flower_destructGroupIterator(groupIterator);
Flower_ChainIterator *chainIterator = flower_getChainIterator(flower);
Chain *chain;
while ((chain = flower_getNextChain(chainIterator)) != NULL) {
chain_check(chain);
}
flower_destructCapIterator(chainIterator);
//We check built trees in here.
Flower_EndIterator *endIterator = flower_getEndIterator(flower);
End *end;
while ((end = flower_getNextEnd(endIterator)) != NULL) {
end_check(end);
end_check(end_getReverse(end)); //We will test everything backwards also.
}
flower_destructEndIterator(endIterator);
if (flower_builtFaces(flower)) {
Flower_FaceIterator *faceIterator = flower_getFaceIterator(flower);
Face *face;
while ((face = flower_getNextFace(faceIterator)) != NULL) {
face_check(face);
}
flower_destructFaceIterator(faceIterator);
face_checkFaces(flower);
} else {
cactusCheck(flower_getFaceNumber(flower) == 0);
}
if (flower_builtBlocks(flower)) { //Note that a flower for which the blocks are not yet built must be a leaf.
Flower_BlockIterator *blockIterator = flower_getBlockIterator(flower);
Block *block;
while ((block = flower_getNextBlock(blockIterator)) != NULL) {
block_check(block);
block_check(block_getReverse(block)); //We will test everything backwards also.
}
flower_destructBlockIterator(blockIterator);
} else {
cactusCheck(flower_isLeaf(flower)); //Defensive
cactusCheck(flower_isTerminal(flower)); //Checks that a flower without built blocks is a leaf and does not
//contain any blocks.
}
Flower_SequenceIterator *sequenceIterator = flower_getSequenceIterator(flower);
Sequence *sequence;
while ((sequence = flower_getNextSequence(sequenceIterator)) != NULL) {
sequence_check(sequence);
}
flower_destructSequenceIterator(sequenceIterator);
}
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);
}
static void cactusEventTestSetup() {
if(!nestedTest) {
cactusEventTestTeardown();
cactusDisk = testCommon_getTemporaryCactusDisk();
flower = flower_construct(cactusDisk);
eventTree = flower_getEventTree(flower);
rootEvent = eventTree_getRootEvent(eventTree);
internalEvent = event_construct3("INTERNAL", 0.5, rootEvent, eventTree);
leafEvent1 = event_construct3("LEAF1", 0.2, internalEvent, eventTree);
leafEvent2 = event_construct3(NULL, 1.3, internalEvent, eventTree);
event_setOutgroupStatus(leafEvent1, 1);
event_setOutgroupStatus(leafEvent2, 0);
}
}
void event_check(Event *event) {
EventTree *eventTree = event_getEventTree(event);
Event *ancestorEvent = event_getParent(event);
//Check event and eventree properly linked
cactusCheck(eventTree_getEvent(event_getEventTree(event), event_getName(event)) == event);
//Event has parent, unless it is root.
if (eventTree_getRootEvent(eventTree) == event) {
cactusCheck(ancestorEvent == NULL);
} else { //not root, so must have ancestor.
cactusCheck(ancestorEvent != NULL);
}
//Each child event has event as parent.
int64_t i = 0;
for (i = 0; i < event_getChildNumber(event); i++) {
Event *childEvent = event_getChild(event, i);
cactusCheck(event_getParent(childEvent) == event);
}
//Ancestor-event --> event edge is consistent with any event tree that is in the parent of the containing flower.
Group *parentGroup = flower_getParentGroup(eventTree_getFlower(
event_getEventTree(event)));
if (parentGroup != NULL) {
EventTree *parentEventTree = flower_getEventTree(group_getFlower(
parentGroup));
Event *parentEvent = eventTree_getEvent(parentEventTree, event_getName(
event));
if (parentEvent != NULL) {
if (ancestorEvent == NULL) { //the case where they are both root.
cactusCheck(eventTree_getRootEvent(parentEventTree) == parentEvent);
} else {
//Check edge ancestorEvent --> event is in parent event tree.
while (1) {
Event *parentAncestorEvent = eventTree_getEvent(
parentEventTree, event_getName(ancestorEvent));
if (parentAncestorEvent != NULL) {
cactusCheck(event_isAncestor(parentEvent, parentAncestorEvent));
break;
}
ancestorEvent = event_getParent(ancestorEvent);
cactusCheck(ancestorEvent != NULL);
}
}
}
}
}
void testBlock_getChain(CuTest* testCase) {
cactusBlockTestSetup();
Block *block2 = block_construct(2, flower);
Flower *flower2 = flower_construct(cactusDisk);
eventTree_copyConstruct(flower_getEventTree(flower), flower2, NULL);
end_copyConstruct(block_get3End(block), flower2);
end_copyConstruct(block_get5End(block2), flower2);
Group *group = group_construct(flower, flower2);
Chain *chain = chain_construct(flower);
link_construct(block_get3End(block), block_get5End(block2), group, chain);
CuAssertTrue(testCase, block_getChain(block_construct(2, flower)) == NULL);
CuAssertTrue(testCase, block_getChain(block) == chain);
CuAssertTrue(testCase, block_getChain(block2) == chain);
cactusBlockTestTeardown();
}
int main(int argc, char *argv[]) {
//////////////////////////////////////////////
//Parse the inputs
//////////////////////////////////////////////
parseBasicArguments(argc, argv, "coverageStats");
assert(referenceEventString != NULL);
assert(otherReferenceEventString != NULL);
assert(outgroupEventString != NULL);
///////////////////////////////////////////////////////////////////////////
// Calculate and print to file a crap load of numbers.
///////////////////////////////////////////////////////////////////////////
Sequence *referenceSequence = NULL;
Sequence *otherReferenceSequence = NULL;
Flower_SequenceIterator *sequenceIt = flower_getSequenceIterator(flower);
Sequence *sequence;
while ((sequence = flower_getNextSequence(sequenceIt)) != NULL) {
const char *eventHeader = event_getHeader(sequence_getEvent(sequence));
if (eventHeader != NULL && strcmp(eventHeader, referenceEventString)
== 0) {
if (referenceSequence == NULL || sequence_getLength(sequence)
>= sequence_getLength(referenceSequence)) {
referenceSequence = sequence;
}
}
if (eventHeader != NULL && strcmp(eventHeader,
otherReferenceEventString) == 0) {
if (otherReferenceSequence == NULL || sequence_getLength(sequence)
>= sequence_getLength(otherReferenceSequence)) {
otherReferenceSequence = sequence;
}
}
}
flower_destructSequenceIterator(sequenceIt);
assert(referenceSequence != NULL);
assert(otherReferenceSequence != NULL);
FILE *fileHandle = fopen(outputFile, "w");
fprintf(
fileHandle,
"<coverageStats referenceSequenceLength=\"%i\" otherReferenceSequenceLength=\"%i\">\n",
sequence_getLength(referenceSequence),
sequence_getLength(otherReferenceSequence));
EventTree_Iterator *eventIt = eventTree_getIterator(
flower_getEventTree(flower));
eventNumber = eventTree_getEventNumber(flower_getEventTree(flower));
Event * event;
totalBaseCoverages = st_calloc(sizeof(int32_t), eventNumber + 1);
totalReferenceBases = 0;
totalOtherReferenceBases = 0;
int32_t totalSamples = 0;
ignoreOtherReferenceBlocks = 0;
while ((event = eventTree_getNext(eventIt)) != NULL) {
sampleEventString = event_getHeader(event);
if (sampleEventString != NULL && strcmp(sampleEventString, "ROOT")
!= 0 && strcmp(sampleEventString, "") != 0) {
baseCoverages = st_calloc(sizeof(int32_t), eventNumber + 1);
baseCoverages[0] = strcmp(sampleEventString, referenceEventString) != 0 ? getTotalLengthOfAdjacencies(flower,
sampleEventString) : 0;
referenceBases = 0;
otherReferenceBases = 0;
getMAFs(flower, fileHandle, getMAFBlock2);
if(strcmp(sampleEventString, referenceEventString) == 0) {
for(int32_t i=2; i<eventNumber + 1; i++) {
baseCoverages[i-1] = baseCoverages[i];
}
baseCoverages[eventNumber] = 0;
}
printStatsForSample(
strcmp(sampleEventString, referenceEventString) != 0 && strcmp(sampleEventString, outgroupEventString) != 0,
fileHandle, 1);
free(baseCoverages);
totalSamples += (strcmp(sampleEventString, referenceEventString)
!= 0 && strcmp(sampleEventString, outgroupEventString) != 0) ? 1 : 0;
}
}
eventTree_destructIterator(eventIt);
//Do average base coverages..
sampleEventString = "average";
baseCoverages = totalBaseCoverages;
referenceBases = totalReferenceBases;
otherReferenceBases = totalOtherReferenceBases;
printStatsForSample(0, fileHandle, totalSamples);
//Do all..
sampleEventString = referenceEventString;
baseCoverages = st_calloc(sizeof(int32_t), eventNumber + 1);
baseCoverages[0] = totalBaseCoverages[0];
referenceBases = 0;
getMAFs(flower, fileHandle, getMAFBlock2);
for(int32_t i=2; i<eventNumber + 1; i++) {
baseCoverages[i-1] = baseCoverages[i];
}
baseCoverages[eventNumber] = 0;
otherReferenceBases = sequence_getLength(otherReferenceSequence);
sampleEventString = "all";
printStatsForSample(0, fileHandle, 1);
free(baseCoverages);
//Do blocks without other reference
ignoreOtherReferenceBlocks = 1;
sampleEventString = referenceEventString;
baseCoverages = st_calloc(sizeof(int32_t), eventNumber + 1);
baseCoverages[0] = totalBaseCoverages[0] - getTotalLengthOfAdjacencies(flower,
otherReferenceEventString);
referenceBases = 0;
otherReferenceBases = 0;
getMAFs(flower, fileHandle, getMAFBlock2);
for(int32_t i=2; i<eventNumber + 1; i++) {
baseCoverages[i-1] = baseCoverages[i];
}
baseCoverages[eventNumber] = 0;
sampleEventString = "minusOtherReference";
printStatsForSample(0, fileHandle, 1);
free(baseCoverages);
fprintf(fileHandle, "</coverageStats>\n");
st_logInfo("Finished writing out the stats.\n");
fclose(fileHandle);
return 0;
}
void flower_removeEventTree(Flower *flower, EventTree *eventTree) {
assert(flower_getEventTree(flower) == eventTree);
flower->eventTree = NULL;
}
void flower_setEventTree(Flower *flower, EventTree *eventTree) {
if (flower_getEventTree(flower) != NULL) {
eventTree_destruct(flower_getEventTree(flower));
}
flower->eventTree = eventTree;
}
void testEventTree_addSiblingUnaryEvent(CuTest *testCase) {
cactusEventTreeTestSetup();
//Create two sibling flowers with the basic event tree..
//then try adding events from on into the other.
Group *group1 = group_construct2(flower);
Group *group2 = group_construct2(flower);
Flower *flower2 = flower_construct(cactusDisk);
Flower *flower3 = flower_construct(cactusDisk);
flower_setParentGroup(flower2, group1);
flower_setParentGroup(flower3, group2);
EventTree *eventTree2 = eventTree_copyConstruct(flower_getEventTree(flower), flower2, NULL);
Event *parentUnaryEvent1 = event_construct4("UNARY1", 0.1, internalEvent, leafEvent1, eventTree);
Event *parentUnaryEvent2 = event_construct4("UNARY2", 0.1, parentUnaryEvent1, leafEvent1, eventTree);
Event *parentUnaryEvent3 = event_construct4("UNARY3", 0.1, internalEvent, leafEvent2, eventTree);
//now event tree contains the added unary events.
EventTree *eventTree3 = eventTree_copyConstruct(flower_getEventTree(flower), flower3, NULL);
//add a couple of denovo events into the new event tree
Event *internalEventChild = eventTree_getEvent(eventTree3, event_getName(internalEvent));
Event *unaryEvent1 = eventTree_getEvent(eventTree3, event_getName(parentUnaryEvent1));
Event *unaryEvent2 = eventTree_getEvent(eventTree3, event_getName(parentUnaryEvent2));
Event *unaryEvent3 = eventTree_getEvent(eventTree3, event_getName(parentUnaryEvent3));
Event *unaryEvent4 = event_construct4("UNARY4", 0.1,
internalEventChild, unaryEvent3, eventTree3);
Event *unaryEvent5 = event_construct4("UNARY5", 0.1,
internalEventChild, unaryEvent4, eventTree3);
//Now event-tree 2 does not contain the unary events but event-tree 3 does..
CuAssertTrue(testCase, eventTree_getEvent(eventTree2, event_getName(unaryEvent1)) == NULL);
CuAssertTrue(testCase, eventTree_getEvent(eventTree2, event_getName(unaryEvent2)) == NULL);
CuAssertTrue(testCase, eventTree_getEvent(eventTree2, event_getName(unaryEvent3)) == NULL);
CuAssertTrue(testCase, eventTree_getEvent(eventTree2, event_getName(unaryEvent4)) == NULL);
CuAssertTrue(testCase, eventTree_getEvent(eventTree2, event_getName(unaryEvent5)) == NULL);
eventTree_addSiblingUnaryEvent(eventTree2, unaryEvent1);
Event *unaryEvent12 = eventTree_getEvent(eventTree2, event_getName(unaryEvent1));
CuAssertTrue(testCase, unaryEvent12 != NULL);
CuAssertTrue(testCase, event_getName(event_getParent(unaryEvent12)) == event_getName(internalEvent));
CuAssertTrue(testCase, event_getChildNumber(unaryEvent12) == 1);
CuAssertTrue(testCase, event_getName(event_getChild(unaryEvent12, 0)) == event_getName(leafEvent1));
eventTree_addSiblingUnaryEvent(eventTree2, unaryEvent2);
Event *unaryEvent22 = eventTree_getEvent(eventTree2, event_getName(unaryEvent2));
CuAssertTrue(testCase, unaryEvent22 != NULL);
CuAssertTrue(testCase, event_getName(event_getParent(unaryEvent22)) == event_getName(unaryEvent1));
CuAssertTrue(testCase, event_getChildNumber(unaryEvent22) == 1);
CuAssertTrue(testCase, event_getName(event_getChild(unaryEvent22, 0)) == event_getName(leafEvent1));
eventTree_addSiblingUnaryEvent(eventTree2, unaryEvent3);
Event *unaryEvent32 = eventTree_getEvent(eventTree2, event_getName(unaryEvent3));
CuAssertTrue(testCase, unaryEvent32 != NULL);
CuAssertTrue(testCase, event_getName(event_getParent(unaryEvent32)) == event_getName(internalEvent));
CuAssertTrue(testCase, event_getChildNumber(unaryEvent32) == 1);
CuAssertTrue(testCase, event_getName(event_getChild(unaryEvent32, 0)) == event_getName(leafEvent2));
eventTree_addSiblingUnaryEvent(eventTree2, unaryEvent4);
Event *unaryEvent42 = eventTree_getEvent(eventTree2, event_getName(unaryEvent4));
CuAssertTrue(testCase, unaryEvent42 != NULL);
CuAssertTrue(testCase, event_getName(event_getParent(unaryEvent42)) == event_getName(internalEvent));
CuAssertTrue(testCase, event_getChildNumber(unaryEvent42) == 1);
CuAssertTrue(testCase, event_getName(event_getChild(unaryEvent42, 0)) == event_getName(unaryEvent3));
eventTree_addSiblingUnaryEvent(eventTree2, unaryEvent5);
Event *unaryEvent52 = eventTree_getEvent(eventTree2, event_getName(unaryEvent5));
CuAssertTrue(testCase, unaryEvent52 != NULL);
CuAssertTrue(testCase, event_getName(event_getParent(unaryEvent52)) == event_getName(internalEvent));
CuAssertTrue(testCase, event_getChildNumber(unaryEvent52) == 1);
CuAssertTrue(testCase, event_getName(event_getChild(unaryEvent52, 0)) == event_getName(unaryEvent4));
//uglyf("Event-tree-1 %s\n", eventTree_makeNewickString(eventTree));
//uglyf("Event-tree-2 %s\n", eventTree_makeNewickString(eventTree3));
//uglyf("Event-tree-3 %s\n", eventTree_makeNewickString(eventTree2));
cactusEventTreeTestTeardown();
}
int main(int argc, char *argv[]) {
/*
* Script for adding a reference genome to a flower.
*/
/*
* Arguments/options
*/
char * logLevelString = NULL;
char * cactusDiskDatabaseString = NULL;
char *referenceEventString =
(char *) cactusMisc_getDefaultReferenceEventHeader();
char *outputFile = NULL;
Name flowerName = NULL_NAME;
///////////////////////////////////////////////////////////////////////////
// (0) Parse the inputs handed by genomeCactus.py / setup stuff.
///////////////////////////////////////////////////////////////////////////
while (1) {
static struct option long_options[] = { { "logLevel",
required_argument, 0, 'a' }, { "cactusDisk", required_argument,
0, 'c' }, { "flowerName", required_argument,
0, 'd' },
{ "referenceEventString", required_argument, 0, 'g' }, {
"help", no_argument, 0, 'h' }, { "outputFile",
required_argument, 0, 'k' },
{ 0, 0, 0, 0 } };
int option_index = 0;
int key = getopt_long(argc, argv, "a:c:d:g:hk:", long_options,
&option_index);
if (key == -1) {
break;
}
switch (key) {
case 'a':
logLevelString = stString_copy(optarg);
break;
case 'c':
cactusDiskDatabaseString = stString_copy(optarg);
break;
case 'd':
flowerName = cactusMisc_stringToName(optarg);
break;
case 'g':
referenceEventString = stString_copy(optarg);
break;
case 'h':
usage();
return 0;
case 'k':
outputFile = stString_copy(optarg);
break;
default:
usage();
return 1;
}
}
///////////////////////////////////////////////////////////////////////////
// (0) Check the inputs.
///////////////////////////////////////////////////////////////////////////
assert(cactusDiskDatabaseString != NULL);
//////////////////////////////////////////////
//Set up logging
//////////////////////////////////////////////
st_setLogLevelFromString(logLevelString);
//////////////////////////////////////////////
//Load the database
//////////////////////////////////////////////
stKVDatabaseConf *kvDatabaseConf = stKVDatabaseConf_constructFromString(
cactusDiskDatabaseString);
CactusDisk *cactusDisk = cactusDisk_construct(kvDatabaseConf, 0);
stKVDatabaseConf_destruct(kvDatabaseConf);
st_logInfo("Set up the flower disk\n");
///////////////////////////////////////////////////////////////////////////
// Get the set of flowers to manipulate
///////////////////////////////////////////////////////////////////////////
Flower *flower = cactusDisk_getFlower(cactusDisk, flowerName);
///////////////////////////////////////////////////////////////////////////
// Get the reference event name
///////////////////////////////////////////////////////////////////////////
Event *referenceEvent = eventTree_getEventByHeader(
flower_getEventTree(flower), referenceEventString);
assert(referenceEvent != NULL);
Name referenceEventName = event_getName(referenceEvent);
///////////////////////////////////////////////////////////////////////////
// Now process each flower in turn.
///////////////////////////////////////////////////////////////////////////
if(outputFile == NULL) {
st_errAbort("No output file specified\n");
}
FILE *fileHandle = fopen(outputFile, "w");
printFastaSequences(flower, fileHandle, referenceEventName);
if(fileHandle != NULL) {
fclose(fileHandle);
}
///////////////////////////////////////////////////////////////////////////
//Clean up memory
///////////////////////////////////////////////////////////////////////////
cactusDisk_destruct(cactusDisk);
//return 0; //Exit without clean up is quicker, enable cleanup when doing memory leak detection.
free(cactusDiskDatabaseString);
free(referenceEventString);
free(logLevelString);
st_logInfo("Cleaned stuff up and am finished\n");
//while(1);
return 0;
}
void testFlower_getEventTree(CuTest* testCase) {
cactusFlowerTestSetup();
CuAssertTrue(testCase, flower_getEventTree(flower) == eventTree);
cactusFlowerTestTeardown();
}
Event *sequence_getEvent(Sequence *sequence) {
return eventTree_getEvent(flower_getEventTree(sequence_getFlower(sequence)), metaSequence_getEventName(sequence->metaSequence));
}
int main(int argc, char *argv[]) {
/*
* Script for adding a reference genome to a flower.
*/
/*
* Arguments/options
*/
char * logLevelString = NULL;
char * cactusDiskDatabaseString = NULL;
char * secondaryDatabaseString = NULL;
char *referenceEventString = (char *) cactusMisc_getDefaultReferenceEventHeader();
bool bottomUpPhase = 0;
///////////////////////////////////////////////////////////////////////////
// (0) Parse the inputs handed by genomeCactus.py / setup stuff.
///////////////////////////////////////////////////////////////////////////
while (1) {
static struct option long_options[] = { { "logLevel", required_argument, 0, 'a' }, { "cactusDisk", required_argument, 0, 'b' }, { "secondaryDisk", required_argument, 0, 'd' }, { "referenceEventString", required_argument, 0, 'g' }, { "help", no_argument,
0, 'h' }, { "bottomUpPhase", no_argument, 0, 'j' }, { 0, 0, 0, 0 } };
int option_index = 0;
int key = getopt_long(argc, argv, "a:b:c:d:e:g:hi:j", long_options, &option_index);
if (key == -1) {
break;
}
switch (key) {
case 'a':
logLevelString = stString_copy(optarg);
break;
case 'b':
cactusDiskDatabaseString = stString_copy(optarg);
break;
case 'd':
secondaryDatabaseString = stString_copy(optarg);
break;
case 'g':
referenceEventString = stString_copy(optarg);
break;
case 'h':
usage();
return 0;
case 'j':
bottomUpPhase = 1;
break;
default:
usage();
return 1;
}
}
///////////////////////////////////////////////////////////////////////////
// (0) Check the inputs.
///////////////////////////////////////////////////////////////////////////
assert(cactusDiskDatabaseString != NULL);
//////////////////////////////////////////////
//Set up logging
//////////////////////////////////////////////
st_setLogLevelFromString(logLevelString);
//////////////////////////////////////////////
//Load the database
//////////////////////////////////////////////
st_logInfo("referenceEventString = %s\n", referenceEventString);
st_logInfo("bottomUpPhase = %i\n", bottomUpPhase);
stKVDatabaseConf *kvDatabaseConf = stKVDatabaseConf_constructFromString(cactusDiskDatabaseString);
CactusDisk *cactusDisk = cactusDisk_construct(kvDatabaseConf, false, true);
stKVDatabaseConf_destruct(kvDatabaseConf);
st_logInfo("Set up the flower disk\n");
stKVDatabase *sequenceDatabase = NULL;
if (secondaryDatabaseString != NULL) {
kvDatabaseConf = stKVDatabaseConf_constructFromString(secondaryDatabaseString);
sequenceDatabase = stKVDatabase_construct(kvDatabaseConf, 0);
stKVDatabaseConf_destruct(kvDatabaseConf);
}
FlowerStream *flowerStream = flowerWriter_getFlowerStream(cactusDisk, stdin);
Flower *flower;
while ((flower = flowerStream_getNext(flowerStream)) != NULL) {
st_logDebug("Processing flower %" PRIi64 "\n", flower_getName(flower));
///////////////////////////////////////////////////////////////////////////
// Get the appropriate event names
///////////////////////////////////////////////////////////////////////////
st_logInfo("%s\n", eventTree_makeNewickString(flower_getEventTree(flower)));
Event *referenceEvent = eventTree_getEventByHeader(flower_getEventTree(flower), referenceEventString);
if (referenceEvent == NULL) {
st_errAbort("Reference event %s not found in tree. Check your "
"--referenceEventString option", referenceEventString);
}
Name referenceEventName = event_getName(referenceEvent);
///////////////////////////////////////////////////////////////////////////
// Now do bottom up or top down, depending
///////////////////////////////////////////////////////////////////////////
stList *flowers = stList_construct();
stList_append(flowers, flower);
preCacheNestedFlowers(cactusDisk, flowers);
if (bottomUpPhase) {
assert(sequenceDatabase != NULL);
cactusDisk_preCacheSegmentStrings(cactusDisk, flowers);
bottomUp(flowers, sequenceDatabase, referenceEventName, !flower_hasParentGroup(flower), generateJukesCantorMatrix);
// Unload the nested flowers to save memory. They haven't
// been changed, so we don't write them to the cactus
// disk.
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (!group_isLeaf(group)) {
flower_unload(group_getNestedFlower(group));
}
}
flower_destructGroupIterator(groupIt);
assert(!flower_isParentLoaded(flower));
// Write this flower to disk.
cactusDisk_addUpdateRequest(cactusDisk, flower);
} else {
topDown(flower, referenceEventName);
// We've changed the nested flowers, but not this
// flower. We write the nested flowers to disk, then
// unload them to save memory. This flower will be
// unloaded by the flower-stream code.
Flower_GroupIterator *groupIt = flower_getGroupIterator(flower);
Group *group;
while ((group = flower_getNextGroup(groupIt)) != NULL) {
if (!group_isLeaf(group)) {
cactusDisk_addUpdateRequest(cactusDisk, group_getNestedFlower(group));
flower_unload(group_getNestedFlower(group));
}
}
flower_destructGroupIterator(groupIt);
}
stList_destruct(flowers);
}
///////////////////////////////////////////////////////////////////////////
// Write the flower(s) back to disk.
///////////////////////////////////////////////////////////////////////////
cactusDisk_write(cactusDisk);
st_logInfo("Updated the flower on disk\n");
///////////////////////////////////////////////////////////////////////////
//Clean up.
///////////////////////////////////////////////////////////////////////////
if (sequenceDatabase != NULL) {
stKVDatabase_destruct(sequenceDatabase);
}
cactusDisk_destruct(cactusDisk);
return 0; //Exit without clean up is quicker, enable cleanup when doing memory leak detection.
free(cactusDiskDatabaseString);
free(referenceEventString);
free(logLevelString);
st_logInfo("Cleaned stuff up and am finished\n");
return 0;
}