int main(int argc, char *argv[]) {
//////////////////////////////////////////////
//Parse the inputs
//////////////////////////////////////////////
parseBasicArguments(argc, argv, "linkageStats");
///////////////////////////////////////////////////////////////////////////
// Get the intervals
///////////////////////////////////////////////////////////////////////////
stList *haplotypeEventStrings = getEventStrings(
treatHaplotype1AsContamination ? NULL : hap1EventString,
treatHaplotype2AsContamination ? NULL : hap2EventString);
stList *assemblyEventStringInList = stList_construct();
stList_append(assemblyEventStringInList, assemblyEventString);
stList *intervals = stList_construct3(0, (void (*)(void *))sequenceInterval_destruct);
for(int64_t i=0; i<stList_length(haplotypeEventStrings); i++) {
const char *hapEventString = stList_get(haplotypeEventStrings, i);
st_logInfo("Getting contig paths for haplotype: %s", hapEventString);
stList *contigPaths = getContigPaths(flower, hapEventString, assemblyEventStringInList);
stList *hapIntervals = getSplitContigPathIntervals(flower, contigPaths, hapEventString,
assemblyEventStringInList);
stList_destruct(contigPaths);
st_logInfo("Getting contig paths\n");
stList_appendAll(intervals, hapIntervals);
stList_setDestructor(hapIntervals, NULL);
stList_destruct(hapIntervals);
}
st_logDebug("Got a total of %" PRIi64 " intervals\n", stList_length(intervals));
///////////////////////////////////////////////////////////////////////////
// Write it out.
///////////////////////////////////////////////////////////////////////////
FILE *fileHandle = fopen(outputFile, "w");
for (int64_t i = 0; i < stList_length(intervals); i++) {
SequenceInterval *sequenceInterval = stList_get(intervals, i);
st_logDebug("We have a path interval %s %" PRIi64 " %" PRIi64 "\n", sequenceInterval->sequenceName,
sequenceInterval->start, sequenceInterval->end);
fprintf(fileHandle, "%s %" PRIi64 " %" PRIi64 "\n", sequenceInterval->sequenceName,
sequenceInterval->start, sequenceInterval->end);
}
st_logInfo("Finished writing out the stats.\n");
fclose(fileHandle);
return 0;
}
stList *getMatchingWithCyclicConstraints(stSortedSet *nodes,
stList *adjacencyEdges, stList *stubEdges, stList *chainEdges,
bool makeStubCyclesDisjoint,
stList *(*matchingAlgorithm)(stList *edges, int64_t nodeNumber)) {
/*
* Check the inputs.
*/
checkInputs(nodes, adjacencyEdges, stubEdges, chainEdges);
st_logDebug("Checked the inputs\n");
if (stSortedSet_size(nodes) == 0) { //Some of the following functions assume there are at least 2 nodes.
return stList_construct();
}
stList *chosenEdges = getPerfectMatching(nodes, adjacencyEdges, matchingAlgorithm);
stSortedSet *allAdjacencyEdges = stList_getSortedSet(adjacencyEdges,
(int(*)(const void *, const void *)) stIntTuple_cmpFn);
stList *nonZeroWeightAdjacencyEdges = getEdgesWithGreaterThanZeroWeight(
adjacencyEdges);
stList *updatedChosenEdges = makeMatchingObeyCyclicConstraints(nodes, chosenEdges, allAdjacencyEdges, nonZeroWeightAdjacencyEdges, stubEdges, chainEdges, makeStubCyclesDisjoint);
stList_destruct(chosenEdges);
chosenEdges = updatedChosenEdges;
stList_destruct(nonZeroWeightAdjacencyEdges);
stSortedSet_destruct(allAdjacencyEdges);
return chosenEdges;
}
static void setup() {
teardown();
assert(nodeNumber == -1);
while(nodeNumber % 2 != 0) {
nodeNumber = st_randomInt(0, 100);
}
assert(nodeNumber >= 0);
assert(nodeNumber % 2 == 0);
stubs = stList_construct3(0, (void (*)(void *))stIntTuple_destruct);
chains = stList_construct3(0, (void (*)(void *))stIntTuple_destruct);
for(int64_t i=0; i<nodeNumber/2; i++) {
assert(nodeNumber/2 > 0);
stIntTuple *edge = stIntTuple_construct2(i, nodeNumber/2 + i);
if(stList_length(stubs) == 0 || st_random() > 0.9) {
stList_append(stubs, edge);
}
else {
stList_append(chains, edge);
}
}
zMatrix = st_calloc(nodeNumber*nodeNumber, sizeof(float));
for(int64_t i=0; i<nodeNumber; i++) {
for(int64_t j=i+1; j<nodeNumber; j++) {
double score = st_random();
zMatrix[i * nodeNumber + j] = score;
zMatrix[j * nodeNumber + i] = score;
}
}
st_logDebug("To test the adjacency problem we've created a problem with %" PRIi64 " nodes %" PRIi64 " stubs and %" PRIi64 " chains\n", nodeNumber, stList_length(stubs), stList_length(chains));
}
void testBlock_makeNewickString(CuTest *testCase) {
assert(testCase != NULL);
cactusBlockTestSetup();
char *cA1 = block_makeNewickString(block, 1, 0);
char *cA2 = block_makeNewickString(block, 1, 1);
char *cA3 = block_makeNewickString(block, 0, 0);
char *cA4 = block_makeNewickString(block, 0, 1);
st_logDebug("I got the block tree string 1 0: %s\n", cA1);
st_logDebug("I got the block tree string 1 1: %s\n", cA2);
st_logDebug("I got the block tree string 0 0: %s\n", cA3);
st_logDebug("I got the block tree string 0 1: %s\n", cA4);
//CuAssertStrEquals(testCase, "(B,E)8;", cA);
free(cA1);
free(cA2);
free(cA3);
free(cA4);
cactusBlockTestTeardown();
}
stList *getContigPaths(Flower *flower, const char *eventString, stList *eventStrings) {
stList *maximalHaplotypePaths = stList_construct3(0,
(void(*)(void *)) stList_destruct);
stSortedSet *segmentSet = stSortedSet_construct();
getMaximalHaplotypePathsP(flower, maximalHaplotypePaths, segmentSet, eventString, eventStrings);
//Do some debug checks..
st_logDebug("We have %" PRIi64 " maximal haplotype paths\n", stList_length(
maximalHaplotypePaths));
getMaximalHaplotypePathsCheck(flower, segmentSet, eventString, eventStrings);
for (int64_t i = 0; i < stList_length(maximalHaplotypePaths); i++) {
stList *maximalHaplotypePath = stList_get(maximalHaplotypePaths, i);
st_logDebug("We have a maximal haplotype path with length %" PRIi64 "\n",
stList_length(maximalHaplotypePath));
assert(stList_length(maximalHaplotypePath) > 0);
Segment *_5Segment = stList_get(maximalHaplotypePath, 0);
Segment *_3Segment = stList_get(maximalHaplotypePath, stList_length(
maximalHaplotypePath) - 1);
if (getAdjacentCapsSegment(segment_get5Cap(_5Segment)) != NULL) {
assert(!trueAdjacency(segment_get5Cap(_5Segment), eventStrings));
}
if (getAdjacentCapsSegment(segment_get3Cap(_3Segment)) != NULL) {
assert(!trueAdjacency(segment_get3Cap(_3Segment), eventStrings));
}
for (int64_t j = 0; j < stList_length(maximalHaplotypePath) - 1; j++) {
_5Segment = stList_get(maximalHaplotypePath, j);
_3Segment = stList_get(maximalHaplotypePath, j + 1);
assert(trueAdjacency(segment_get3Cap(_5Segment), eventStrings));
assert(trueAdjacency(segment_get5Cap(_3Segment), eventStrings));
assert(cap_getAdjacency(getTerminalCap(segment_get3Cap(_5Segment)))
== getTerminalCap(segment_get5Cap(_3Segment)));
assert(strcmp(event_getHeader(segment_getEvent(_5Segment)),
eventString) == 0);
assert(strcmp(event_getHeader(segment_getEvent(_3Segment)),
eventString) == 0);
assert(hasCapInEvents(cap_getEnd(segment_get5Cap(_5Segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get5Cap(_5Segment))));
assert(hasCapInEvents(cap_getEnd(segment_get5Cap(_3Segment)), eventStrings)); //isHaplotypeEnd(cap_getEnd(segment_get5Cap(_3Segment))));
}
}
stSortedSet_destruct(segmentSet);
return maximalHaplotypePaths;
}
static void testCalculateZScore(CuTest *testCase) {
for(int64_t i=0; i<100; i++) {
int64_t n = st_randomInt(0, 100);
int64_t m = st_randomInt(0, 100);
int64_t k = st_randomInt(1, 10000);
double theta = st_random() > 0.05 ? st_random() : 0.0;
double zScore = calculateZScore(n, m, k, theta);
double zScoreSlow = calculateZScoreSlow(n, m, k, theta);
st_logDebug("The slow computed score: %f the fast computed score: %f, n: %" PRIi64 " m: %" PRIi64 " k: %" PRIi64 ", theta: %lf\n", zScoreSlow, zScore, n, m, k, theta);
CuAssertDblEquals(testCase, zScoreSlow, zScore, 0.000001);
}
}
static stList *chooseAdjacencyPairing_externalProgram(stList *edges, int64_t nodeNumber, const char *programName) {
/*
* We create temp files to hold stuff.
*/
if(nodeNumber <= 1) {
assert(stList_length(edges) == 0);
return stList_construct();
}
char *tempInputFile = getTempFile(), *tempOutputFile = getTempFile();
/*
* We write the graph to a temp file.
*/
FILE *fileHandle = fopen(tempInputFile, "w");
if(strcmp(programName, "blossom5") == 0) { //Must be all connected as
//generates perfect matchings.
writeCliqueGraph(fileHandle, edges, nodeNumber, 1);
}
else {
writeGraph(fileHandle, edges, nodeNumber);
}
fclose(fileHandle);
/*
* We run the external program.
*/
char *command = stString_print("%s -e %s -w %s >& /dev/null", programName, tempInputFile, tempOutputFile);
int64_t i = st_system(command);
if(i != 0) {
st_errAbort("Something went wrong with the command: %s", command);
//For some reason this causes a seg fault
//stThrowNew(MATCHING_EXCEPTION, "Something went wrong with the command: %s", command);
}
free(command);
/*
* We get back the matching.
*/
fileHandle = fopen(tempOutputFile, "r");
stList *matching = readMatching(fileHandle, edges);
fclose(fileHandle);
st_logDebug("The adjacency matching for %" PRIi64 " nodes with %" PRIi64 " initial edges contains %" PRIi64 " edges\n", nodeNumber, stList_length(edges), stList_length(matching));
/*
* Get rid of the temp files..
*/
st_system("rm -rf %s %s", tempInputFile, tempOutputFile);
free(tempInputFile);
free(tempOutputFile);
return matching;
}
stList *makeMatchingObeyCyclicConstraints(stSortedSet *nodes,
stList *chosenEdges,
stSortedSet *allAdjacencyEdges, stList *nonZeroWeightAdjacencyEdges,
stList *stubEdges, stList *chainEdges,
bool makeStubCyclesDisjoint) {
if (stSortedSet_size(nodes) == 0) { //Some of the following functions assume there are at least 2 nodes.
return stList_construct();
}
/*
* Merge in the stub free components.
*/
chosenEdges = mergeSimpleCycles2(chosenEdges,
nonZeroWeightAdjacencyEdges, allAdjacencyEdges, stubEdges,
chainEdges);
st_logDebug(
"After merging in chain only cycles the matching has %" PRIi64 " edges, %" PRIi64 " cardinality and %" PRIi64 " weight\n",
stList_length(chosenEdges), matchingCardinality(chosenEdges),
matchingWeight(chosenEdges));
/*
* Split stub components.
*/
if (makeStubCyclesDisjoint) {
stList *updatedChosenEdges = splitMultipleStubCycles(chosenEdges,
nonZeroWeightAdjacencyEdges, allAdjacencyEdges, stubEdges,
chainEdges);
stList_destruct(chosenEdges);
chosenEdges = updatedChosenEdges;
st_logDebug(
"After making stub cycles disjoint the matching has %" PRIi64 " edges, %" PRIi64 " cardinality and %" PRIi64 " weight\n",
stList_length(chosenEdges), matchingCardinality(chosenEdges),
matchingWeight(chosenEdges));
} else {
st_logDebug("Not making stub cycles disjoint\n");
}
return chosenEdges;
}
stList *getPerfectMatching(stSortedSet *nodes,
stList *adjacencyEdges,
stList *(*matchingAlgorithm)(stList *edges, int64_t nodeNumber)) {
checkEdges(adjacencyEdges, nodes, 1, 1); //Checks edges are clique
if (stSortedSet_size(nodes) == 0) { //Some of the following functions assume there are at least 2 nodes.
return stList_construct();
}
stList *nonZeroWeightAdjacencyEdges = getEdgesWithGreaterThanZeroWeight(
adjacencyEdges);
stList *chosenEdges = getSparseMatching(nodes, nonZeroWeightAdjacencyEdges, matchingAlgorithm);
stList_destruct(nonZeroWeightAdjacencyEdges);
makeMatchingPerfect(chosenEdges, adjacencyEdges, nodes);
st_logDebug(
"Chosen a perfect matching with %" PRIi64 " edges, %" PRIi64 " cardinality and %" PRIi64 " weight\n",
stList_length(chosenEdges), matchingCardinality(chosenEdges),
matchingWeight(chosenEdges));
return chosenEdges;
}
static CactusDisk *cactusDisk_constructPrivate(stKVDatabaseConf *conf, bool create, const char *sequencesFileName) {
//sequencesFileName = NULL; //Disable the ability to store the sequences on disk.
CactusDisk *cactusDisk = st_calloc(1, sizeof(CactusDisk));
//construct lists of in memory objects
cactusDisk->metaSequences = stSortedSet_construct3(cactusDisk_constructMetaSequencesP, NULL);
cactusDisk->flowers = stSortedSet_construct3(cactusDisk_constructFlowersP, NULL);
cactusDisk->flowerNamesMarkedForDeletion = stSortedSet_construct3((int (*)(const void *, const void *)) strcmp,
free);
cactusDisk->updateRequests = stList_construct3(0, (void (*)(void *)) stKVDatabaseBulkRequest_destruct);
//Now open the database
cactusDisk->database = stKVDatabase_construct(conf, create);
cactusDisk->cache = stCache_construct();
cactusDisk->stringCache = stCache_construct();
//initialise the unique ids.
int64_t seed = (clock() << 24) | (time(NULL) << 16) | (getpid() & 65535); //Likely to be unique
st_logDebug("The cactus disk is seeding the random number generator with the value %" PRIi64 "\n", seed);
st_randomSeed(seed);
cactusDisk->uniqueNumber = 0;
cactusDisk->maxUniqueNumber = 0;
//Now load any stuff..
if (containsRecord(cactusDisk, CACTUS_DISK_PARAMETER_KEY)) {
if (create) {
stThrowNew(CACTUS_DISK_EXCEPTION_ID, "Tried to create a cactus disk, but the cactus disk already exists");
}
if (sequencesFileName != NULL) {
stThrowNew(CACTUS_DISK_EXCEPTION_ID,
"A sequences file name is specified, but the cactus disk is not being created");
}
void *record = getRecord(cactusDisk, CACTUS_DISK_PARAMETER_KEY, "cactus_disk parameters");
void *record2 = record;
cactusDisk_loadFromBinaryRepresentation(&record, cactusDisk, conf);
free(record2);
} else {
assert(create);
if (sequencesFileName == NULL) {
cactusDisk->storeSequencesInAFile = 0;
cactusDisk->sequencesFileName = NULL;
cactusDisk->sequencesReadFileHandle = NULL;
cactusDisk->sequencesWriteFileHandle = NULL;
cactusDisk->absSequencesFileName = NULL;
} else {
if (stKVDatabaseConf_getDir(conf) == NULL) {
stThrowNew(CACTUS_DISK_EXCEPTION_ID,
"The database conf does not contain a directory in which the sequence file is to be found!\n");
}
cactusDisk->storeSequencesInAFile = 1;
cactusDisk->sequencesFileName = stString_copy(sequencesFileName);
cactusDisk->absSequencesFileName = stString_print("%s/%s", stKVDatabaseConf_getDir(conf),
cactusDisk->sequencesFileName);
//Make sure the file exists
cactusDisk->sequencesReadFileHandle = fopen(cactusDisk->absSequencesFileName, "w");
assert(cactusDisk->sequencesReadFileHandle != NULL);
fclose(cactusDisk->sequencesReadFileHandle); //Flush it first time.
cactusDisk->sequencesReadFileHandle = NULL;
cactusDisk->sequencesWriteFileHandle = NULL;
}
}
return cactusDisk;
}
static void testGibbsSamplingWithSimulatedAnnealing_WithCooling(
CuTest *testCase) {
st_logDebug("Running adjacency problem tests using gibbs sampling, with exponentially decreasing temperature function\n");
testGibbsSamplingWithSimulatedAnnealing(testCase,
exponentiallyDecreasingTemperatureFn, 0, INT64_MAX);
}
static void testGibbsSamplingWithSimulatedAnnealing_ConstantTemperature(
CuTest *testCase) {
st_logDebug("Running adjacency problem tests using gibbs sampling, but with constant temperature\n");
testGibbsSamplingWithSimulatedAnnealing(testCase, constantTemperatureFn, 0, INT64_MAX);
}
static void testGibbsSamplingWithSimulatedAnnealing_NoExponentiation(
CuTest *testCase) {
st_logDebug("Running adjacency problem tests using gibbs sampling, but no exponentiation\n");
testGibbsSamplingWithSimulatedAnnealing(testCase,
NULL, 0, INT64_MAX);
}
static void testGibbsSamplingWithSimulatedAnnealing_NoExponentiation_Greedy_Fast(
CuTest *testCase) {
st_logDebug("Running adjacency problem tests using gibbs sampling, but greedy sampling\n");
testGibbsSamplingWithSimulatedAnnealing(testCase,
NULL, 1, 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 cactusDisk_write(CactusDisk *cactusDisk) {
Flower *flower;
int64_t recordSize;
stList *removeRequests = stList_construct3(0, (void (*)(void *)) stIntTuple_destruct);
st_logDebug("Starting to write the cactus to disk\n");
stSortedSetIterator *it = stSortedSet_getIterator(cactusDisk->flowers);
//Sort flowers to update.
while ((flower = stSortedSet_getNext(it)) != NULL) {
cactusDisk_addUpdateRequest(cactusDisk, flower);
}
stSortedSet_destructIterator(it);
st_logDebug("Got the flowers to update\n");
//Remove nets that are marked for deletion..
it = stSortedSet_getIterator(cactusDisk->flowerNamesMarkedForDeletion);
char *nameString;
while ((nameString = stSortedSet_getNext(it)) != NULL) {
Name name = cactusMisc_stringToName(nameString);
if (containsRecord(cactusDisk, name)) {
stList_append(cactusDisk->updateRequests, stKVDatabaseBulkRequest_constructUpdateRequest(name, &name, 0)); //We set it to null in the first atomic operation.
stList_append(removeRequests, stIntTuple_construct1(name));
}
}
stSortedSet_destructIterator(it);
st_logDebug("Avoided updating nets marked for deletion\n");
// Insert and/or update meta-sequences.
it = stSortedSet_getIterator(cactusDisk->metaSequences);
MetaSequence *metaSequence;
while ((metaSequence = stSortedSet_getNext(it)) != NULL) {
void *vA =
binaryRepresentation_makeBinaryRepresentation(metaSequence,
(void (*)(void *, void (*)(const void * ptr, size_t size, size_t count))) metaSequence_writeBinaryRepresentation,
&recordSize);
//Compression
vA = compress(vA, &recordSize);
if (!containsRecord(cactusDisk, metaSequence_getName(metaSequence))) {
stList_append(cactusDisk->updateRequests,
stKVDatabaseBulkRequest_constructInsertRequest(metaSequence_getName(metaSequence), vA, recordSize));
} else {
stList_append(cactusDisk->updateRequests,
stKVDatabaseBulkRequest_constructUpdateRequest(metaSequence_getName(metaSequence), vA, recordSize));
}
free(vA);
}
stSortedSet_destructIterator(it);
st_logDebug("Got the sequences we are going to add to the database.\n");
if (!containsRecord(cactusDisk, CACTUS_DISK_PARAMETER_KEY)) { //We only write the parameters once.
//Finally the database info.
void *cactusDiskParameters =
binaryRepresentation_makeBinaryRepresentation(cactusDisk,
(void (*)(void *, void (*)(const void * ptr, size_t size, size_t count))) cactusDisk_writeBinaryRepresentation,
&recordSize);
//Compression
cactusDiskParameters = compress(cactusDiskParameters, &recordSize);
stList_append(cactusDisk->updateRequests,
stKVDatabaseBulkRequest_constructInsertRequest(CACTUS_DISK_PARAMETER_KEY, cactusDiskParameters,
recordSize));
free(cactusDiskParameters);
}
st_logDebug("Checked if need to write the initial parameters\n");
if (stList_length(cactusDisk->updateRequests) > 0) {
st_logDebug("Going to write %" PRIi64 " updates\n", stList_length(cactusDisk->updateRequests));
stTry
{
st_logDebug("Writing %" PRIi64 " updates\n", stList_length(cactusDisk->updateRequests));
assert(stList_length(cactusDisk->updateRequests) > 0);
stKVDatabase_bulkSetRecords(cactusDisk->database, cactusDisk->updateRequests);
}
stCatch(except)
{
stThrowNewCause(except, ST_KV_DATABASE_EXCEPTION_ID,
"Failed when trying to set records in updating the cactus disk");
}stTryEnd
;
}
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[]) {
/*
* 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);
}
