void mapGenes(Flower *flower, FILE *fileHandle, struct bed *gene, char *species){
st_logInfo("Flower %s\n", cactusMisc_nameToString(flower_getName(flower)));
printOpeningTag("geneMap", fileHandle);
fprintf(fileHandle, "\n");
int level = 0;//Flower level
while(gene != NULL){
//Get the start of the target sequence:
st_logInfo("Gene %s:\n", gene->name);
Cap *startCap;
struct List *capList = flower_getThreadStarts(flower, species);
for(int i=0; i < capList->length; i++){
startCap = capList->list[i];
st_logInfo("Cap %d, %s\n", i, cactusMisc_nameToString(cap_getName(startCap)));
//Traverse cactus and get regions that overlap with exons of the gene, report the involved chains relations
fprintf(fileHandle, "\t<gene name=\"%s\" target=\"%s\" start=\"%" PRIi64 "\" end=\"%" PRIi64 "\" exonCount=\"%" PRIi64 "\" strand=\"%c\">\n",
gene->name, species, gene->chromStart, gene->chromEnd, gene->blockCount, gene->strand[0]);
fprintf(fileHandle, "\t\t<exon id=\"0\" start=\"%" PRIi64 "\" end=\"%" PRIi64 "\">\n",
gene->chromStart, gene->chromStart + gene->blockSizes->list[0]);
mapGene(startCap, level, 0, gene, fileHandle);
fprintf(fileHandle, "\t</gene>\n");
}
gene = gene->next;
}
printClosingTag("geneMap", fileHandle);
return;
}
Group *flower_getParentGroup(Flower *flower) {
if (flower->parentFlowerName == NULL_NAME) {
return NULL;
}
Flower *flower2 = cactusDisk_getFlower(flower_getCactusDisk(flower), flower->parentFlowerName);
assert(flower2 != NULL);
return flower_getGroup(flower2, flower_getName(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 cactusDisk_addUpdateRequest(CactusDisk *cactusDisk, Flower *flower) {
int64_t recordSize;
void *vA = binaryRepresentation_makeBinaryRepresentation(flower,
(void (*)(void *, void (*)(const void * ptr, size_t size, size_t count))) flower_writeBinaryRepresentation,
&recordSize);
//Compression
vA = compress(vA, &recordSize);
if (containsRecord(cactusDisk, flower_getName(flower))) {
int64_t recordSize2;
void *vA2 = stCache_getRecord(cactusDisk->cache, flower_getName(flower), 0, INT64_MAX, &recordSize2);
if (!stCache_recordsIdentical(vA, recordSize, vA2, recordSize2)) { //Only rewrite if we actually did something
stList_append(cactusDisk->updateRequests,
stKVDatabaseBulkRequest_constructUpdateRequest(flower_getName(flower), vA, recordSize));
}
free(vA2);
} else {
stList_append(cactusDisk->updateRequests,
stKVDatabaseBulkRequest_constructInsertRequest(flower_getName(flower), vA, recordSize));
}
free(vA);
}
void makeCactusTree_terminalNode(Flower *flower, FILE *fileHandle,
const char *parentNodeName, const char *parentEdgeColour) {
char *groupNameString = cactusMisc_nameToString(flower_getName(flower));
double scalingFactor = flower_getTotalBaseLength(flower) / totalProblemSize;
assert(scalingFactor <= 1.001);
assert(scalingFactor >= -0.001);
addNodeToGraph(groupNameString, fileHandle, scalingFactor, "triangle",
groupNameString);
//Write in the parent edge.
if (parentNodeName != NULL) {
graphViz_addEdgeToGraph(parentNodeName, groupNameString, fileHandle,
"", parentEdgeColour, 10, 1, "forward");
}
free(groupNameString);
}
void testGroup_makeNonLeaf(CuTest *testCase) {
cactusGroupTestSetup();
CuAssertTrue(testCase, group_isLeaf(group2));
end_setGroup(end4, group2);
group_makeNestedFlower(group2);
CuAssertTrue(testCase, !group_isLeaf(group2));
Flower *nestedFlower = group_getNestedFlower(group2);
CuAssertTrue(testCase, nestedFlower != NULL);
CuAssertTrue(testCase, !flower_builtBlocks(flower));
CuAssertTrue(testCase, !flower_builtTrees(flower));
CuAssertTrue(testCase, !flower_builtFaces(flower));
CuAssertTrue(testCase, flower_getName(nestedFlower) == group_getName(group2));
CuAssertTrue(testCase, flower_getParentGroup(nestedFlower) == group2);
CuAssertTrue(testCase, flower_getEndNumber(nestedFlower) == 1);
End *nestedEnd = flower_getFirstEnd(nestedFlower);
CuAssertTrue(testCase, end_getName(end4) == end_getName(nestedEnd));
CuAssertTrue(testCase, end_getGroup(nestedEnd) != NULL);
CuAssertTrue(testCase, flower_getGroupNumber(nestedFlower) == 1);
CuAssertTrue(testCase, flower_isTerminal(nestedFlower));
cactusGroupTestTeardown();
}
void testCactusDisk_getFlower(CuTest* testCase) {
cactusDiskTestSetup();
Flower *flower = flower_construct(cactusDisk);
Flower *flower2 = flower_construct(cactusDisk);
CuAssertTrue(testCase, cactusDisk_getFlower(cactusDisk, flower_getName(flower)) == flower);
CuAssertTrue(testCase, cactusDisk_getFlower(cactusDisk, flower_getName(flower2)) == flower2);
//now try closing the disk, then reloading it, to see if we get the same result.
Name name1 = flower_getName(flower);
Name name2 = flower_getName(flower2);
cactusDisk_write(cactusDisk);
cactusDisk_destruct(cactusDisk);
cactusDisk = cactusDisk_construct(conf, 0);
flower = cactusDisk_getFlower(cactusDisk, name1);
flower2 = cactusDisk_getFlower(cactusDisk, name2);
CuAssertTrue(testCase, flower != NULL);
CuAssertTrue(testCase, flower2 != NULL);
CuAssertTrue(testCase, flower_getName(flower) == name1);
CuAssertTrue(testCase, flower_getName(flower2) == name2);
cactusDiskTestTeardown();
}
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;
}
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;
}
void testGroup_getNestedFlowerName(CuTest* testCase) {
cactusGroupTestSetup();
CuAssertTrue(testCase, group_getName(group) == flower_getName(nestedFlower));
cactusGroupTestTeardown();
}
void flower_setParentGroup(Flower *flower, Group *group) {
//assert(flower->parentFlowerName == NULL_NAME); we can change this if merging the parent flowers, so this no longer applies.
flower->parentFlowerName = flower_getName(group_getFlower(group));
}
void testFlower_getName(CuTest* testCase) {
cactusFlowerTestSetup();
CuAssertTrue(testCase, flower_getName(flower) != NULL_NAME);
CuAssertTrue(testCase, cactusDisk_getFlower(cactusDisk, flower_getName(flower)) == flower);
cactusFlowerTestTeardown();
}
void testFlower_removeIfRedundant(CuTest *testCase) {
/*
* Do a simple test to see if function can remove a redundant flower.
*/
cactusFlowerTestSetup();
endsSetup();
//First construct a redundant flower from the root.
Flower *flower2 = flower_construct(cactusDisk);
Group *group = group_construct(flower, flower2);
end_setGroup(end, group);
end_setGroup(end2, group);
//Now hang another couple of flowers of that.
Flower *flower3 = flower_construct(cactusDisk);
group_construct(flower2, flower3);
//Now hang another flower of that.
Group *group3b = group_construct2(flower2);
//Finally hang one more flower on the end..
Flower *flower4 = flower_construct(cactusDisk);
group_construct(flower3, flower4);
//Copy the ends into the flowers.
end_copyConstruct(end, flower2);
end_copyConstruct(end2, flower2);
end_copyConstruct(end, flower3);
end_setGroup(flower_getEnd(flower2, end_getName(end2)), group3b);
end_copyConstruct(end, flower4);
//st_uglyf("I got %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 "\n", flower_getName(flower), flower_getName(flower2), flower_getName(flower3), flower_getName(flower4));
//Write the mess to disk.
cactusDisk_write(cactusDisk);
//Now test the removal function (check we get a negative on this leaf).
CuAssertTrue(testCase, !flower_removeIfRedundant(flower4));
//Check we can't remove the root..
CuAssertTrue(testCase, !flower_removeIfRedundant(flower));
//We will remove flower2
//Before
CuAssertTrue(testCase, flower_getGroupNumber(flower) == 1);
CuAssertTrue(testCase, group_getFlower(flower_getParentGroup(flower2)) == flower);
CuAssertTrue(testCase, flower_removeIfRedundant(flower2));
//After, check the flower/group connections
CuAssertTrue(testCase, flower_getGroupNumber(flower) == 2);
CuAssertTrue(testCase, !flower_isLeaf(flower));
CuAssertTrue(testCase, group_getFlower(flower_getParentGroup(flower3)) == flower);
group3b = end_getGroup(end2);
CuAssertTrue(testCase, group_getFlower(group3b) == flower);
CuAssertTrue(testCase, group_isLeaf(group3b));
CuAssertTrue(testCase, flower_getGroup(flower, flower_getName(flower3)) == flower_getParentGroup(flower3));
//Check the ends..
CuAssertTrue(testCase, flower_getEndNumber(flower) == 2);
CuAssertTrue(testCase, flower_getEndNumber(flower3) == 1);
CuAssertTrue(testCase, group_getEndNumber(group3b) == 1);
CuAssertTrue(testCase, end_getGroup(end) == flower_getParentGroup(flower3));
CuAssertTrue(testCase, end_getGroup(end2) == group3b);
CuAssertTrue(testCase, flower_getEnd(flower3, end_getName(end)) != NULL);
//Check the child of 3 is still okay..
CuAssertTrue(testCase, group_getFlower(flower_getParentGroup(flower4)) == flower3);
//Now do removal of flower3
CuAssertTrue(testCase, !flower_removeIfRedundant(flower));
CuAssertTrue(testCase, !flower_removeIfRedundant(flower4));
CuAssertTrue(testCase, flower_removeIfRedundant(flower3));
//Check groups again
CuAssertTrue(testCase, flower_getGroupNumber(flower) == 2);
CuAssertTrue(testCase, !flower_isLeaf(flower));
CuAssertTrue(testCase, group_getFlower(flower_getParentGroup(flower4)) == flower);
CuAssertTrue(testCase, group_getFlower(group3b) == flower);
CuAssertTrue(testCase, flower_getGroup(flower, flower_getName(flower4)) == flower_getParentGroup(flower4));
//Check the ends again..
CuAssertTrue(testCase, flower_getEndNumber(flower) == 2);
CuAssertTrue(testCase, flower_getEndNumber(flower4) == 1);
CuAssertTrue(testCase, group_getEndNumber(group3b) == 1);
CuAssertTrue(testCase, end_getGroup(end) == flower_getParentGroup(flower4));
CuAssertTrue(testCase, end_getGroup(end2) == group3b);
CuAssertTrue(testCase, flower_getEnd(flower4, end_getName(end)) != NULL);
cactusFlowerTestTeardown();
}
void makeCactusTree_flower(Flower *flower, FILE *fileHandle, const char *parentNodeName,
const char *parentEdgeColour) {
if(flower_isTerminal(flower)) {
makeCactusTree_terminalNode(flower, fileHandle, parentNodeName, parentEdgeColour);
}
else {
//Write the flower nodes.
char *flowerNameString = cactusMisc_nameToString(flower_getName(flower));
const char *edgeColour = graphViz_getColour();
addNodeToGraph(flowerNameString, fileHandle, flower_getTotalBaseLength(flower)
/ totalProblemSize, "ellipse", flowerNameString);
//Write in the parent edge.
if (parentNodeName != NULL) {
graphViz_addEdgeToGraph(parentNodeName, flowerNameString, fileHandle, "",
parentEdgeColour, 10, 1, "forward");
}
//Create the chains.
Flower_ChainIterator *chainIterator = flower_getChainIterator(flower);
Chain *chain;
while ((chain = flower_getNextChain(chainIterator)) != NULL) {
makeCactusTree_chain(chain, fileHandle, flowerNameString, edgeColour);
}
flower_destructChainIterator(chainIterator);
//Create the diamond node
char *diamondNodeNameString = st_malloc(sizeof(char) * (strlen(
flowerNameString) + 2));
sprintf(diamondNodeNameString, "z%s", flowerNameString);
const char *diamondEdgeColour = graphViz_getColour();
//Create all the groups linked to the diamond.
Flower_GroupIterator *groupIterator = flower_getGroupIterator(flower);
Group *group;
double size = 0.0; //get the size of the group organising node..
int64_t nonTrivialGroupCount = 0;
while ((group = flower_getNextGroup(groupIterator)) != NULL) {
assert(!group_isLeaf(group));
if (group_isTangle(group)) {
size += group_getTotalBaseLength(group);
nonTrivialGroupCount++;
}
}
flower_destructGroupIterator(groupIterator);
if(nonTrivialGroupCount == 0) {
assert(flower_getParentGroup(flower) == 0);
}
else {
//assert(nonTrivialGroupCount > 0);
addNodeToGraph(diamondNodeNameString, fileHandle, size
/ totalProblemSize, "diamond", "");
graphViz_addEdgeToGraph(flowerNameString, diamondNodeNameString,
fileHandle, "", edgeColour, 10, 1, "forward");
groupIterator = flower_getGroupIterator(flower);
while ((group = flower_getNextGroup(groupIterator)) != NULL) {
if (group_isTangle(group)) {
assert(!group_isLeaf(group));
makeCactusTree_flower(group_getNestedFlower(group), fileHandle,
diamondNodeNameString, diamondEdgeColour);
}
}
flower_destructGroupIterator(groupIterator);
}
free(flowerNameString);
free(diamondNodeNameString);
}
}
static int cactusDisk_constructFlowersP(const void *o1, const void *o2) {
return cactusMisc_nameCompare(flower_getName((Flower *) o1), flower_getName((Flower *) o2));
}
int main(int argc, char *argv[]) {
/*
* Script for adding alignments to cactus tree.
*/
int64_t startTime;
stKVDatabaseConf *kvDatabaseConf;
CactusDisk *cactusDisk;
int key, k;
bool (*filterFn)(stPinchSegment *, stPinchSegment *) = NULL;
stSet *outgroupThreads = NULL;
/*
* Arguments/options
*/
char * logLevelString = NULL;
char * alignmentsFile = NULL;
char * constraintsFile = NULL;
char * cactusDiskDatabaseString = NULL;
char * lastzArguments = "";
int64_t minimumSequenceLengthForBlast = 1;
//Parameters for annealing/melting rounds
int64_t *annealingRounds = NULL;
int64_t annealingRoundsLength = 0;
int64_t *meltingRounds = NULL;
int64_t meltingRoundsLength = 0;
//Parameters for melting
float maximumAdjacencyComponentSizeRatio = 10;
int64_t blockTrim = 0;
int64_t alignmentTrimLength = 0;
int64_t *alignmentTrims = NULL;
int64_t chainLengthForBigFlower = 1000000;
int64_t longChain = 2;
int64_t minLengthForChromosome = 1000000;
float proportionOfUnalignedBasesForNewChromosome = 0.8;
bool breakChainsAtReverseTandems = 1;
int64_t maximumMedianSequenceLengthBetweenLinkedEnds = INT64_MAX;
bool realign = 0;
char *realignArguments = "";
bool removeRecoverableChains = false;
bool (*recoverableChainsFilter)(stCactusEdgeEnd *, Flower *) = NULL;
int64_t maxRecoverableChainsIterations = 1;
int64_t maxRecoverableChainLength = INT64_MAX;
//Parameters for removing ancient homologies
bool doPhylogeny = false;
int64_t phylogenyNumTrees = 1;
enum stCaf_RootingMethod phylogenyRootingMethod = BEST_RECON;
enum stCaf_ScoringMethod phylogenyScoringMethod = COMBINED_LIKELIHOOD;
double breakpointScalingFactor = 1.0;
bool phylogenySkipSingleCopyBlocks = 0;
int64_t phylogenyMaxBaseDistance = 1000;
int64_t phylogenyMaxBlockDistance = 100;
bool phylogenyKeepSingleDegreeBlocks = 0;
stList *phylogenyTreeBuildingMethods = stList_construct();
enum stCaf_TreeBuildingMethod defaultMethod = GUIDED_NEIGHBOR_JOINING;
stList_append(phylogenyTreeBuildingMethods, &defaultMethod);
double phylogenyCostPerDupPerBase = 0.2;
double phylogenyCostPerLossPerBase = 0.2;
const char *debugFileName = NULL;
const char *referenceEventHeader = NULL;
double phylogenyDoSplitsWithSupportHigherThanThisAllAtOnce = 1.0;
int64_t numTreeBuildingThreads = 2;
int64_t minimumBlockDegreeToCheckSupport = 10;
double minimumBlockHomologySupport = 0.7;
double nucleotideScalingFactor = 1.0;
HomologyUnitType phylogenyHomologyUnitType = BLOCK;
enum stCaf_DistanceCorrectionMethod phylogenyDistanceCorrectionMethod = JUKES_CANTOR;
bool sortAlignments = false;
///////////////////////////////////////////////////////////////////////////
// (0) Parse the inputs handed by genomeCactus.py / setup stuff.
///////////////////////////////////////////////////////////////////////////
while (1) {
static struct option long_options[] = { { "logLevel", required_argument, 0, 'a' }, { "alignments", required_argument, 0, 'b' }, {
"cactusDisk", required_argument, 0, 'c' }, { "lastzArguments", required_argument, 0, 'd' },
{ "help", no_argument, 0, 'h' }, { "annealingRounds", required_argument, 0, 'i' }, { "trim", required_argument, 0, 'k' }, {
"trimChange", required_argument, 0, 'l', }, { "minimumTreeCoverage", required_argument, 0, 'm' }, { "blockTrim",
required_argument, 0, 'n' }, { "deannealingRounds", required_argument, 0, 'o' }, { "minimumDegree",
required_argument, 0, 'p' }, { "minimumIngroupDegree", required_argument, 0, 'q' }, {
"minimumOutgroupDegree", required_argument, 0, 'r' }, { "alignmentFilter", required_argument, 0, 't' }, {
"minimumSequenceLengthForBlast", required_argument, 0, 'v' }, { "maxAdjacencyComponentSizeRatio",
required_argument, 0, 'w' }, { "constraints", required_argument, 0, 'x' }, { "minLengthForChromosome",
required_argument, 0, 'y' }, { "proportionOfUnalignedBasesForNewChromosome", required_argument, 0, 'z' },
{ "maximumMedianSequenceLengthBetweenLinkedEnds", required_argument, 0, 'A' },
{ "realign", no_argument, 0, 'B' }, { "realignArguments", required_argument, 0, 'C' },
{ "phylogenyNumTrees", required_argument, 0, 'D' },
{ "phylogenyRootingMethod", required_argument, 0, 'E' },
{ "phylogenyScoringMethod", required_argument, 0, 'F' },
{ "phylogenyBreakpointScalingFactor", required_argument, 0, 'G' },
{ "phylogenySkipSingleCopyBlocks", no_argument, 0, 'H' },
{ "phylogenyMaxBaseDistance", required_argument, 0, 'I' },
{ "phylogenyMaxBlockDistance", required_argument, 0, 'J' },
{ "phylogenyDebugFile", required_argument, 0, 'K' },
{ "phylogenyKeepSingleDegreeBlocks", no_argument, 0, 'L' },
{ "phylogenyTreeBuildingMethod", required_argument, 0, 'M' },
{ "phylogenyCostPerDupPerBase", required_argument, 0, 'N' },
{ "phylogenyCostPerLossPerBase", required_argument, 0, 'O' },
{ "referenceEventHeader", required_argument, 0, 'P' },
{ "phylogenyDoSplitsWithSupportHigherThanThisAllAtOnce", required_argument, 0, 'Q' },
{ "numTreeBuildingThreads", required_argument, 0, 'R' },
{ "phylogeny", no_argument, 0, 'S' },
{ "minimumBlockHomologySupport", required_argument, 0, 'T' },
{ "phylogenyNucleotideScalingFactor", required_argument, 0, 'U' },
{ "minimumBlockDegreeToCheckSupport", required_argument, 0, 'V' },
{ "removeRecoverableChains", required_argument, 0, 'W' },
{ "minimumNumberOfSpecies", required_argument, 0, 'X' },
{ "phylogenyHomologyUnitType", required_argument, 0, 'Y' },
{ "phylogenyDistanceCorrectionMethod", required_argument, 0, 'Z' },
{ "maxRecoverableChainsIterations", required_argument, 0, '1' },
{ "maxRecoverableChainLength", required_argument, 0, '2' },
{ 0, 0, 0, 0 } };
int option_index = 0;
key = getopt_long(argc, argv, "a:b:c:hi:k:m:n:o:p:q:r:stv:w:x:y:z:A:BC:D:E:", long_options, &option_index);
if (key == -1) {
break;
}
switch (key) {
case 'a':
logLevelString = stString_copy(optarg);
st_setLogLevelFromString(logLevelString);
break;
case 'b':
alignmentsFile = stString_copy(optarg);
break;
case 'c':
cactusDiskDatabaseString = stString_copy(optarg);
break;
case 'd':
lastzArguments = stString_copy(optarg);
break;
case 'h':
usage();
return 0;
case 'i':
annealingRounds = getInts(optarg, &annealingRoundsLength);
break;
case 'o':
meltingRounds = getInts(optarg, &meltingRoundsLength);
break;
case 'k':
alignmentTrims = getInts(optarg, &alignmentTrimLength);
break;
case 'm':
k = sscanf(optarg, "%f", &minimumTreeCoverage);
assert(k == 1);
break;
case 'n':
k = sscanf(optarg, "%" PRIi64 "", &blockTrim);
assert(k == 1);
break;
case 'p':
k = sscanf(optarg, "%" PRIi64 "", &minimumDegree);
assert(k == 1);
break;
case 'q':
k = sscanf(optarg, "%" PRIi64 "", &minimumIngroupDegree);
assert(k == 1);
break;
case 'r':
k = sscanf(optarg, "%" PRIi64 "", &minimumOutgroupDegree);
assert(k == 1);
break;
case 't':
if (strcmp(optarg, "singleCopyOutgroup") == 0) {
sortAlignments = true;
filterFn = stCaf_filterByOutgroup;
} else if (strcmp(optarg, "relaxedSingleCopyOutgroup") == 0) {
sortAlignments = true;
filterFn = stCaf_relaxedFilterByOutgroup;
} else if (strcmp(optarg, "singleCopy") == 0) {
sortAlignments = true;
filterFn = stCaf_filterByRepeatSpecies;
} else if (strcmp(optarg, "relaxedSingleCopy") == 0) {
sortAlignments = true;
filterFn = stCaf_relaxedFilterByRepeatSpecies;
} else if (strcmp(optarg, "singleCopyChr") == 0) {
sortAlignments = true;
filterFn = stCaf_singleCopyChr;
} else if (strcmp(optarg, "singleCopyIngroup") == 0) {
sortAlignments = true;
filterFn = stCaf_singleCopyIngroup;
} else if (strcmp(optarg, "relaxedSingleCopyIngroup") == 0) {
sortAlignments = true;
filterFn = stCaf_relaxedSingleCopyIngroup;
} else if (strcmp(optarg, "none") == 0) {
sortAlignments = false;
filterFn = NULL;
} else {
st_errAbort("Could not recognize alignmentFilter option %s", optarg);
}
break;
case 'v':
k = sscanf(optarg, "%" PRIi64 "", &minimumSequenceLengthForBlast);
assert(k == 1);
break;
case 'w':
k = sscanf(optarg, "%f", &maximumAdjacencyComponentSizeRatio);
assert(k == 1);
break;
case 'x':
constraintsFile = stString_copy(optarg);
break;
case 'y':
k = sscanf(optarg, "%" PRIi64 "", &minLengthForChromosome);
assert(k == 1);
break;
case 'z':
k = sscanf(optarg, "%f", &proportionOfUnalignedBasesForNewChromosome);
assert(k == 1);
break;
case 'A':
k = sscanf(optarg, "%" PRIi64 "", &maximumMedianSequenceLengthBetweenLinkedEnds);
assert(k == 1);
break;
case 'B':
realign = 1;
break;
case 'C':
realignArguments = stString_copy(optarg);
break;
case 'D':
k = sscanf(optarg, "%" PRIi64, &phylogenyNumTrees);
assert(k == 1);
break;
case 'E':
if (!strcmp(optarg, "outgroupBranch")) {
phylogenyRootingMethod = OUTGROUP_BRANCH;
} else if (!strcmp(optarg, "longestBranch")) {
phylogenyRootingMethod = LONGEST_BRANCH;
} else if (!strcmp(optarg, "bestRecon")) {
phylogenyRootingMethod = BEST_RECON;
} else {
st_errAbort("Invalid tree rooting method: %s", optarg);
}
break;
case 'F':
if (!strcmp(optarg, "reconCost")) {
phylogenyScoringMethod = RECON_COST;
} else if (!strcmp(optarg, "nucLikelihood")) {
phylogenyScoringMethod = NUCLEOTIDE_LIKELIHOOD;
} else if (!strcmp(optarg, "reconLikelihood")) {
phylogenyScoringMethod = RECON_LIKELIHOOD;
} else if (!strcmp(optarg, "combinedLikelihood")) {
phylogenyScoringMethod = COMBINED_LIKELIHOOD;
} else {
st_errAbort("Invalid tree scoring method: %s", optarg);
}
break;
case 'G':
k = sscanf(optarg, "%lf", &breakpointScalingFactor);
assert(k == 1);
break;
case 'H':
phylogenySkipSingleCopyBlocks = true;
break;
case 'I':
k = sscanf(optarg, "%" PRIi64, &phylogenyMaxBaseDistance);
assert(k == 1);
break;
case 'J':
k = sscanf(optarg, "%" PRIi64, &phylogenyMaxBlockDistance);
assert(k == 1);
break;
case 'K':
debugFileName = stString_copy(optarg);
break;
case 'L':
phylogenyKeepSingleDegreeBlocks = true;
break;
case 'M':
// clear the default setting of the list
stList_destruct(phylogenyTreeBuildingMethods);
phylogenyTreeBuildingMethods = stList_construct();
stList *methodStrings = stString_splitByString(optarg, ",");
for (int64_t i = 0; i < stList_length(methodStrings); i++) {
char *methodString = stList_get(methodStrings, i);
enum stCaf_TreeBuildingMethod *method = st_malloc(sizeof(enum stCaf_TreeBuildingMethod));
if (strcmp(methodString, "neighborJoining") == 0) {
*method = NEIGHBOR_JOINING;
} else if (strcmp(methodString, "guidedNeighborJoining") == 0) {
*method = GUIDED_NEIGHBOR_JOINING;
} else if (strcmp(methodString, "splitDecomposition") == 0) {
*method = SPLIT_DECOMPOSITION;
} else if (strcmp(methodString, "strictSplitDecomposition") == 0) {
*method = STRICT_SPLIT_DECOMPOSITION;
} else if (strcmp(methodString, "removeBadChains") == 0) {
*method = REMOVE_BAD_CHAINS;
} else {
st_errAbort("Unknown tree building method: %s", methodString);
}
stList_append(phylogenyTreeBuildingMethods, method);
}
stList_destruct(methodStrings);
break;
case 'N':
k = sscanf(optarg, "%lf", &phylogenyCostPerDupPerBase);
assert(k == 1);
break;
case 'O':
k = sscanf(optarg, "%lf", &phylogenyCostPerLossPerBase);
assert(k == 1);
break;
case 'P':
referenceEventHeader = stString_copy(optarg);
break;
case 'Q':
k = sscanf(optarg, "%lf", &phylogenyDoSplitsWithSupportHigherThanThisAllAtOnce);
assert(k == 1);
break;
case 'R':
k = sscanf(optarg, "%" PRIi64, &numTreeBuildingThreads);
assert(k == 1);
break;
case 'S':
doPhylogeny = true;
break;
case 'T':
k = sscanf(optarg, "%lf", &minimumBlockHomologySupport);
assert(k == 1);
assert(minimumBlockHomologySupport <= 1.0);
assert(minimumBlockHomologySupport >= 0.0);
break;
case 'U':
k = sscanf(optarg, "%lf", &nucleotideScalingFactor);
assert(k == 1);
break;
case 'V':
k = sscanf(optarg, "%" PRIi64, &minimumBlockDegreeToCheckSupport);
assert(k == 1);
break;
case 'W':
if (strcmp(optarg, "1") == 0) {
removeRecoverableChains = true;
recoverableChainsFilter = NULL;
} else if (strcmp(optarg, "unequalNumberOfIngroupCopies") == 0) {
removeRecoverableChains = true;
recoverableChainsFilter = stCaf_chainHasUnequalNumberOfIngroupCopies;
} else if (strcmp(optarg, "unequalNumberOfIngroupCopiesOrNoOutgroup") == 0) {
removeRecoverableChains = true;
recoverableChainsFilter = stCaf_chainHasUnequalNumberOfIngroupCopiesOrNoOutgroup;
} else if (strcmp(optarg, "0") == 0) {
removeRecoverableChains = false;
} else {
st_errAbort("Could not parse removeRecoverableChains argument");
}
break;
case 'X':
k = sscanf(optarg, "%" PRIi64, &minimumNumberOfSpecies);
if (k != 1) {
st_errAbort("Error parsing the minimumNumberOfSpecies argument");
}
break;
case 'Y':
if (strcmp(optarg, "chain") == 0) {
phylogenyHomologyUnitType = CHAIN;
} else if (strcmp(optarg, "block") == 0) {
phylogenyHomologyUnitType = BLOCK;
} else {
st_errAbort("Could not parse the phylogenyHomologyUnitType argument");
}
break;
case 'Z':
if (strcmp(optarg, "jukesCantor") == 0) {
phylogenyDistanceCorrectionMethod = JUKES_CANTOR;
} else if (strcmp(optarg, "none") == 0 ) {
phylogenyDistanceCorrectionMethod = NONE;
} else {
st_errAbort("Could not parse the phylogenyDistanceCorrectionMethod argument");
}
break;
case '1':
k = sscanf(optarg, "%" PRIi64, &maxRecoverableChainsIterations);
if (k != 1) {
st_errAbort("Error parsing the maxRecoverableChainsIterations argument");
}
break;
case '2':
k = sscanf(optarg, "%" PRIi64, &maxRecoverableChainLength);
if (k != 1) {
st_errAbort("Error parsing the maxRecoverableChainLength argument");
}
break;
default:
usage();
return 1;
}
}
///////////////////////////////////////////////////////////////////////////
// (0) Check the inputs.
///////////////////////////////////////////////////////////////////////////
assert(cactusDiskDatabaseString != NULL);
assert(minimumTreeCoverage >= 0.0);
assert(minimumTreeCoverage <= 1.0);
assert(blockTrim >= 0);
assert(annealingRoundsLength >= 0);
for (int64_t i = 0; i < annealingRoundsLength; i++) {
assert(annealingRounds[i] >= 0);
}
assert(meltingRoundsLength >= 0);
for (int64_t i = 1; i < meltingRoundsLength; i++) {
assert(meltingRounds[i - 1] < meltingRounds[i]);
assert(meltingRounds[i - 1] >= 1);
}
assert(alignmentTrimLength >= 0);
for (int64_t i = 0; i < alignmentTrimLength; i++) {
assert(alignmentTrims[i] >= 0);
}
assert(minimumOutgroupDegree >= 0);
assert(minimumIngroupDegree >= 0);
//////////////////////////////////////////////
//Set up logging
//////////////////////////////////////////////
st_setLogLevelFromString(logLevelString);
//////////////////////////////////////////////
//Log (some of) the inputs
//////////////////////////////////////////////
st_logInfo("Flower disk name : %s\n", cactusDiskDatabaseString);
//////////////////////////////////////////////
//Load the database
//////////////////////////////////////////////
kvDatabaseConf = stKVDatabaseConf_constructFromString(cactusDiskDatabaseString);
cactusDisk = cactusDisk_construct(kvDatabaseConf, 0);
st_logInfo("Set up the flower disk\n");
///////////////////////////////////////////////////////////////////////////
// Sort the constraints
///////////////////////////////////////////////////////////////////////////
stPinchIterator *pinchIteratorForConstraints = NULL;
if (constraintsFile != NULL) {
pinchIteratorForConstraints = stPinchIterator_constructFromFile(constraintsFile);
st_logInfo("Created an iterator for the alignment constaints from file: %s\n", constraintsFile);
}
///////////////////////////////////////////////////////////////////////////
// Do the alignment
///////////////////////////////////////////////////////////////////////////
startTime = time(NULL);
stList *flowers = flowerWriter_parseFlowersFromStdin(cactusDisk);
if (alignmentsFile == NULL) {
cactusDisk_preCacheStrings(cactusDisk, flowers);
}
char *tempFile1 = NULL;
for (int64_t i = 0; i < stList_length(flowers); i++) {
flower = stList_get(flowers, i);
if (!flower_builtBlocks(flower)) { // Do nothing if the flower already has defined blocks
st_logDebug("Processing flower: %lli\n", flower_getName(flower));
stCaf_setFlowerForAlignmentFiltering(flower);
//Set up the graph and add the initial alignments
stPinchThreadSet *threadSet = stCaf_setup(flower);
//Build the set of outgroup threads
outgroupThreads = stCaf_getOutgroupThreads(flower, threadSet);
//Setup the alignments
stPinchIterator *pinchIterator;
stList *alignmentsList = NULL;
if (alignmentsFile != NULL) {
assert(i == 0);
assert(stList_length(flowers) == 1);
if (sortAlignments) {
tempFile1 = getTempFile();
stCaf_sortCigarsFileByScoreInDescendingOrder(alignmentsFile, tempFile1);
pinchIterator = stPinchIterator_constructFromFile(tempFile1);
} else {
pinchIterator = stPinchIterator_constructFromFile(alignmentsFile);
}
} else {
if (tempFile1 == NULL) {
tempFile1 = getTempFile();
}
alignmentsList = stCaf_selfAlignFlower(flower, minimumSequenceLengthForBlast, lastzArguments, realign, realignArguments, tempFile1);
if (sortAlignments) {
stCaf_sortCigarsByScoreInDescendingOrder(alignmentsList);
}
st_logDebug("Ran lastz and have %" PRIi64 " alignments\n", stList_length(alignmentsList));
pinchIterator = stPinchIterator_constructFromList(alignmentsList);
}
for (int64_t annealingRound = 0; annealingRound < annealingRoundsLength; annealingRound++) {
int64_t minimumChainLength = annealingRounds[annealingRound];
int64_t alignmentTrim = annealingRound < alignmentTrimLength ? alignmentTrims[annealingRound] : 0;
st_logDebug("Starting annealing round with a minimum chain length of %" PRIi64 " and an alignment trim of %" PRIi64 "\n", minimumChainLength, alignmentTrim);
stPinchIterator_setTrim(pinchIterator, alignmentTrim);
//Add back in the constraints
if (pinchIteratorForConstraints != NULL) {
stCaf_anneal(threadSet, pinchIteratorForConstraints, filterFn);
}
//Do the annealing
if (annealingRound == 0) {
stCaf_anneal(threadSet, pinchIterator, filterFn);
} else {
stCaf_annealBetweenAdjacencyComponents(threadSet, pinchIterator, filterFn);
}
// Dump the block degree and length distribution to a file
if (debugFileName != NULL) {
dumpBlockInfo(threadSet, stString_print("%s-blockStats-preMelting", debugFileName));
}
printf("Sequence graph statistics after annealing:\n");
printThreadSetStatistics(threadSet, flower, stdout);
// Check for poorly-supported blocks--those that have
// been transitively aligned together but with very
// few homologies supporting the transitive
// alignment. These "megablocks" can snarl up the
// graph so that a lot of extra gets thrown away in
// the first melting step.
stPinchThreadSetBlockIt blockIt = stPinchThreadSet_getBlockIt(threadSet);
stPinchBlock *block;
while ((block = stPinchThreadSetBlockIt_getNext(&blockIt)) != NULL) {
if (stPinchBlock_getDegree(block) > minimumBlockDegreeToCheckSupport) {
uint64_t supportingHomologies = stPinchBlock_getNumSupportingHomologies(block);
uint64_t possibleSupportingHomologies = numPossibleSupportingHomologies(block, flower);
double support = ((double) supportingHomologies) / possibleSupportingHomologies;
if (support < minimumBlockHomologySupport) {
fprintf(stdout, "Destroyed a megablock with degree %" PRIi64
" and %" PRIi64 " supporting homologies out of a maximum "
"of %" PRIi64 " (%lf%%).\n", stPinchBlock_getDegree(block),
supportingHomologies, possibleSupportingHomologies, support);
stPinchBlock_destruct(block);
}
}
}
//Do the melting rounds
for (int64_t meltingRound = 0; meltingRound < meltingRoundsLength; meltingRound++) {
int64_t minimumChainLengthForMeltingRound = meltingRounds[meltingRound];
st_logDebug("Starting melting round with a minimum chain length of %" PRIi64 " \n", minimumChainLengthForMeltingRound);
if (minimumChainLengthForMeltingRound >= minimumChainLength) {
break;
}
stCaf_melt(flower, threadSet, NULL, 0, minimumChainLengthForMeltingRound, 0, INT64_MAX);
} st_logDebug("Last melting round of cycle with a minimum chain length of %" PRIi64 " \n", minimumChainLength);
stCaf_melt(flower, threadSet, NULL, 0, minimumChainLength, breakChainsAtReverseTandems, maximumMedianSequenceLengthBetweenLinkedEnds);
//This does the filtering of blocks that do not have the required species/tree-coverage/degree.
stCaf_melt(flower, threadSet, blockFilterFn, blockTrim, 0, 0, INT64_MAX);
}
if (removeRecoverableChains) {
stCaf_meltRecoverableChains(flower, threadSet, breakChainsAtReverseTandems, maximumMedianSequenceLengthBetweenLinkedEnds, recoverableChainsFilter, maxRecoverableChainsIterations, maxRecoverableChainLength);
}
if (debugFileName != NULL) {
dumpBlockInfo(threadSet, stString_print("%s-blockStats-postMelting", debugFileName));
}
printf("Sequence graph statistics after melting:\n");
printThreadSetStatistics(threadSet, flower, stdout);
// Build a tree for each block, then use each tree to
// partition the homologies between the ingroups sequences
// into those that occur before the speciation with the
// outgroup and those which occur late.
if (stSet_size(outgroupThreads) > 0 && doPhylogeny) {
st_logDebug("Starting to build trees and partition ingroup homologies\n");
stHash *threadStrings = stCaf_getThreadStrings(flower, threadSet);
st_logDebug("Got sets of thread strings and set of threads that are outgroups\n");
stCaf_PhylogenyParameters params;
params.distanceCorrectionMethod = phylogenyDistanceCorrectionMethod;
params.treeBuildingMethods = phylogenyTreeBuildingMethods;
params.rootingMethod = phylogenyRootingMethod;
params.scoringMethod = phylogenyScoringMethod;
params.breakpointScalingFactor = breakpointScalingFactor;
params.nucleotideScalingFactor = nucleotideScalingFactor;
params.skipSingleCopyBlocks = phylogenySkipSingleCopyBlocks;
params.keepSingleDegreeBlocks = phylogenyKeepSingleDegreeBlocks;
params.costPerDupPerBase = phylogenyCostPerDupPerBase;
params.costPerLossPerBase = phylogenyCostPerLossPerBase;
params.maxBaseDistance = phylogenyMaxBaseDistance;
params.maxBlockDistance = phylogenyMaxBlockDistance;
params.numTrees = phylogenyNumTrees;
params.ignoreUnalignedBases = 1;
params.onlyIncludeCompleteFeatureBlocks = 0;
params.doSplitsWithSupportHigherThanThisAllAtOnce = phylogenyDoSplitsWithSupportHigherThanThisAllAtOnce;
params.numTreeBuildingThreads = numTreeBuildingThreads;
assert(params.numTreeBuildingThreads >= 1);
stCaf_buildTreesToRemoveAncientHomologies(
threadSet, phylogenyHomologyUnitType, threadStrings, outgroupThreads, flower, ¶ms,
debugFileName == NULL ? NULL : stString_print("%s-phylogeny", debugFileName), referenceEventHeader);
stHash_destruct(threadStrings);
st_logDebug("Finished building trees\n");
if (removeRecoverableChains) {
// We melt recoverable chains after splitting, as
// well as before, to alleviate coverage loss
// caused by bad splits.
stCaf_meltRecoverableChains(flower, threadSet, breakChainsAtReverseTandems, maximumMedianSequenceLengthBetweenLinkedEnds, recoverableChainsFilter, maxRecoverableChainsIterations, maxRecoverableChainLength);
}
// Enforce the block constraints on minimum degree,
// etc. after splitting.
stCaf_melt(flower, threadSet, blockFilterFn, 0, 0, 0, INT64_MAX);
}
//Sort out case when we allow blocks of degree 1
if (minimumDegree < 2) {
st_logDebug("Creating degree 1 blocks\n");
stCaf_makeDegreeOneBlocks(threadSet);
stCaf_melt(flower, threadSet, blockFilterFn, blockTrim, 0, 0, INT64_MAX);
} else if (maximumAdjacencyComponentSizeRatio < INT64_MAX) { //Deal with giant components
st_logDebug("Breaking up components greedily\n");
stCaf_breakupComponentsGreedily(threadSet, maximumAdjacencyComponentSizeRatio);
}
//Finish up
stCaf_finish(flower, threadSet, chainLengthForBigFlower, longChain, minLengthForChromosome,
proportionOfUnalignedBasesForNewChromosome); //Flower is then destroyed at this point.
st_logInfo("Ran the cactus core script\n");
//Cleanup
stPinchThreadSet_destruct(threadSet);
stPinchIterator_destruct(pinchIterator);
stSet_destruct(outgroupThreads);
if (alignmentsList != NULL) {
stList_destruct(alignmentsList);
}
st_logInfo("Cleaned up from main loop\n");
} else {
st_logInfo("We've already built blocks / alignments for this flower\n");
}
}
stList_destruct(flowers);
if (tempFile1 != NULL) {
st_system("rm %s", tempFile1);
}
if (constraintsFile != NULL) {
stPinchIterator_destruct(pinchIteratorForConstraints);
}
///////////////////////////////////////////////////////////////////////////
// Write the flower to disk.
///////////////////////////////////////////////////////////////////////////
st_logDebug("Writing the flowers to disk\n");
cactusDisk_write(cactusDisk);
st_logInfo("Updated the flower on disk and %" PRIi64 " seconds have elapsed\n", time(NULL) - startTime);
///////////////////////////////////////////////////////////////////////////
// Clean up.
///////////////////////////////////////////////////////////////////////////
cactusDisk_destruct(cactusDisk);
}
int main(int argc, char *argv[]) {
/*
* Open the database.
* Construct a flower.
* Construct an event tree representing the species tree.
* For each sequence contruct two ends each containing an cap.
* Make a file for the sequence.
* Link the two caps.
* Finish!
*/
int64_t key, j;
Group *group;
Flower_EndIterator *endIterator;
End *end;
bool makeEventHeadersAlphaNumeric = 0;
/*
* Arguments/options
*/
char * logLevelString = NULL;
char * speciesTree = NULL;
char * outgroupEvents = NULL;
///////////////////////////////////////////////////////////////////////////
// (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' }, {
"speciesTree", required_argument, 0, 'f' }, { "outgroupEvents", required_argument, 0, 'g' },
{ "help", no_argument, 0, 'h' }, { "makeEventHeadersAlphaNumeric", no_argument, 0, 'i' }, { 0, 0, 0, 0 } };
int option_index = 0;
key = getopt_long(argc, argv, "a:b:f:hg:i", long_options, &option_index);
if (key == -1) {
break;
}
switch (key) {
case 'a':
logLevelString = optarg;
break;
case 'b':
cactusDiskDatabaseString = optarg;
break;
case 'f':
speciesTree = optarg;
break;
case 'g':
outgroupEvents = optarg;
break;
case 'h':
usage();
return 0;
case 'i':
makeEventHeadersAlphaNumeric = 1;
break;
default:
usage();
return 1;
}
}
///////////////////////////////////////////////////////////////////////////
// (0) Check the inputs.
///////////////////////////////////////////////////////////////////////////
//assert(logLevelString == NULL || strcmp(logLevelString, "CRITICAL") == 0 || strcmp(logLevelString, "INFO") == 0 || strcmp(logLevelString, "DEBUG") == 0);
assert(cactusDiskDatabaseString != NULL);
assert(speciesTree != NULL);
//////////////////////////////////////////////
//Set up logging
//////////////////////////////////////////////
st_setLogLevelFromString(logLevelString);
//////////////////////////////////////////////
//Log (some of) the inputs
//////////////////////////////////////////////
st_logInfo("Flower disk name : %s\n", cactusDiskDatabaseString);
for (j = optind; j < argc; j++) {
st_logInfo("Sequence file/directory %s\n", argv[j]);
}
//////////////////////////////////////////////
//Load the database
//////////////////////////////////////////////
stKVDatabaseConf *kvDatabaseConf = kvDatabaseConf = stKVDatabaseConf_constructFromString(cactusDiskDatabaseString);
if (stKVDatabaseConf_getType(kvDatabaseConf) == stKVDatabaseTypeTokyoCabinet || stKVDatabaseConf_getType(kvDatabaseConf)
== stKVDatabaseTypeKyotoTycoon) {
assert(stKVDatabaseConf_getDir(kvDatabaseConf) != NULL);
cactusDisk = cactusDisk_construct2(kvDatabaseConf, "cactusSequences");
} else {
cactusDisk = cactusDisk_construct(kvDatabaseConf, 1);
}
st_logInfo("Set up the flower disk\n");
//////////////////////////////////////////////
//Construct the flower
//////////////////////////////////////////////
if (cactusDisk_getFlower(cactusDisk, 0) != NULL) {
cactusDisk_destruct(cactusDisk);
st_logInfo("The first flower already exists\n");
return 0;
}
flower = flower_construct2(0, cactusDisk);
assert(flower_getName(flower) == 0);
st_logInfo("Constructed the flower\n");
//////////////////////////////////////////////
//Construct the event tree
//////////////////////////////////////////////
st_logInfo("Going to build the event tree with newick string: %s\n", speciesTree);
stTree *tree = stTree_parseNewickString(speciesTree);
st_logInfo("Parsed the tree\n");
if (makeEventHeadersAlphaNumeric) {
makeEventHeadersAlphaNumericFn(tree);
}
stTree_setBranchLength(tree, INT64_MAX);
checkBranchLengthsAreDefined(tree);
eventTree = eventTree_construct2(flower); //creates the event tree and the root even
totalEventNumber = 1;
st_logInfo("Constructed the basic event tree\n");
// Construct a set of outgroup names so that ancestral outgroups
// get recognized.
stSet *outgroupNameSet = stSet_construct3(stHash_stringKey,
stHash_stringEqualKey,
free);
if(outgroupEvents != NULL) {
stList *outgroupNames = stString_split(outgroupEvents);
for(int64_t i = 0; i < stList_length(outgroupNames); i++) {
char *outgroupName = stList_get(outgroupNames, i);
stSet_insert(outgroupNameSet, stString_copy(outgroupName));
}
stList_destruct(outgroupNames);
}
//now traverse the tree
j = optind;
assignEventsAndSequences(eventTree_getRootEvent(eventTree), tree,
outgroupNameSet, argv, &j);
char *eventTreeString = eventTree_makeNewickString(eventTree);
st_logInfo(
"Constructed the initial flower with %" PRIi64 " sequences and %" PRIi64 " events with string: %s\n",
totalSequenceNumber, totalEventNumber, eventTreeString);
assert(event_getSubTreeBranchLength(eventTree_getRootEvent(eventTree)) >= 0.0);
free(eventTreeString);
//assert(0);
//////////////////////////////////////////////
//Label any outgroup events.
//////////////////////////////////////////////
if (outgroupEvents != NULL) {
stList *outgroupEventsList = stString_split(outgroupEvents);
for (int64_t i = 0; i < stList_length(outgroupEventsList); i++) {
char *outgroupEvent = makeEventHeadersAlphaNumeric ? makeAlphaNumeric(stList_get(outgroupEventsList, i)) : stString_copy(stList_get(outgroupEventsList, i));
Event *event = eventTree_getEventByHeader(eventTree, outgroupEvent);
if (event == NULL) {
st_errAbort("Got an outgroup string that does not match an event, outgroup string %s", outgroupEvent);
}
assert(!event_isOutgroup(event));
event_setOutgroupStatus(event, 1);
assert(event_isOutgroup(event));
free(outgroupEvent);
}
stList_destruct(outgroupEventsList);
}
//////////////////////////////////////////////
//Construct the terminal group.
//////////////////////////////////////////////
if (flower_getEndNumber(flower) > 0) {
group = group_construct2(flower);
endIterator = flower_getEndIterator(flower);
while ((end = flower_getNextEnd(endIterator)) != NULL) {
end_setGroup(end, group);
}
flower_destructEndIterator(endIterator);
assert(group_isLeaf(group));
// Create a one link chain if there is only one pair of attached ends..
group_constructChainForLink(group);
assert(!flower_builtBlocks(flower));
} else {
flower_setBuiltBlocks(flower, 1);
}
///////////////////////////////////////////////////////////////////////////
// Write the flower to disk.
///////////////////////////////////////////////////////////////////////////
//flower_check(flower);
cactusDisk_write(cactusDisk);
st_logInfo("Updated the flower on disk\n");
///////////////////////////////////////////////////////////////////////////
// Cleanup.
///////////////////////////////////////////////////////////////////////////
cactusDisk_destruct(cactusDisk);
return 0; //Exit without clean up is quicker, enable cleanup when doing memory leak detection.
stSet_destruct(outgroupNameSet);
stTree_destruct(tree);
stKVDatabaseConf_destruct(kvDatabaseConf);
return 0;
}
