int main(int argc, char *argv[])
/* The program */
{
struct bed *bedList = NULL;
struct dnaSeq *targets = NULL, *target;
struct dnaSeq *queries = NULL, *query;
if (argc != 4)
usage();
targets = dnaLoadAll(argv[2]);
queries = dnaLoadAll(argv[1]);
for (target = targets; target != NULL; target = target->next)
for (query = queries; query != NULL; query = query->next)
{
struct bed *oneList = oligoMatch(target, query);
bedList = slCat(bedList, oneList);
}
outputBed6(bedList, argv[3]);
bedFreeList(&bedList);
dnaSeqFreeList(&targets);
dnaSeqFreeList(&queries);
return 0;
}
void itsaMake(int inCount, char *inputs[], char *output)
/* itsaMake - Make a suffix array file out of input DNA sequences.. */
{
verboseTimeInit();
bits64 maxGenomeSize = 1024LL*1024*1024*4;
itsaBaseToValInit();
/* Load all DNA, make sure names are unique, and alphabetize by name. */
struct dnaSeq *seqList = NULL, *seq;
struct hash *uniqSeqHash = hashNew(0);
bits64 totalDnaSize = 1; /* FOr space between. */
int inputIx;
for (inputIx=0; inputIx<inCount; ++inputIx)
{
char * input = inputs[inputIx];
struct dnaLoad *dl = dnaLoadOpen(input);
while ((seq = dnaLoadNext(dl)) != NULL)
{
verbose(1, "read %s with %d bases\n", seq->name, seq->size);
if (hashLookup(uniqSeqHash, seq->name))
errAbort("Input sequence name %s repeated, all must be unique.", seq->name);
totalDnaSize += seq->size + 1;
if (totalDnaSize > maxGenomeSize)
errAbort("Too much DNA. Can only handle up to %lld bases", maxGenomeSize);
slAddHead(&seqList, seq);
}
dnaLoadClose(&dl);
}
slSort(&seqList, dnaSeqCmpName);
verboseTime(1, "Loaded %lld bases in %d sequences", totalDnaSize, slCount(seqList));
/* Allocate big buffer for all DNA. */
DNA *allDna = globalAllDna = needHugeMem(totalDnaSize);
allDna[0] = 0;
bits64 chromOffset = 1; /* Have zeroes between each chrom, and before and after. */
/* Copy DNA to a single big buffer, and create chromInfo on each sequence. */
struct chromInfo *chrom, *chromList = NULL;
for (seq = seqList; seq != NULL; seq = seq->next)
{
AllocVar(chrom);
chrom->name = cloneString(seq->name);
chrom->size = seq->size;
chrom->offset = chromOffset;
slAddHead(&chromList, chrom);
toUpperN(seq->dna, seq->size);
memcpy(allDna + chromOffset, seq->dna, seq->size + 1);
chromOffset += seq->size + 1;
}
slReverse(&chromList);
/* Free up separate dna sequences because we're going to need a lot of RAM soon. */
/* Allocate index array, and offset and list arrays. */
dnaSeqFreeList(&seqList);
bits32 *index13;
AllocArray(index13, itsaSlotCount);
bits32 *offsetArray = needHugeMem(totalDnaSize * sizeof(bits32));
bits32 *listArray = needHugeZeroedMem(totalDnaSize * sizeof(bits32));
verboseTime(1, "Allocated buffers %lld bytes total",
(long long)(9LL*totalDnaSize + itsaSlotCount*sizeof(bits32)));
/* Where normally we'd keep some sort of structure with a next element to form a list
* of matching positions in each slot of our index, to conserve memory we'll do this
* with two parallel arrays. Because we're such cheapskates in terms of memory we'll
* (and still using 9*genomeSize bytes of RAM) we'll use these arrays for two different
* purposes.
* In the first phase they will together be used to form linked lists of
* offsets, and the 13mer index will point to the first item in each list. In this
* phase the offsetArray contains offsets into the allDna structure, and the listArray
* contains the next pointers for the list. After the first phase we write out the
* suffix array to disk.
* In the second phase we read the suffix array back into the offsetArray, and
* use the listArray for the traverseArray. We write out the traverse array to finish
* things up. */
/* Load up all DNA buffer. */
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
verbose(2, " About to do first pass index\n");
indexChromPass1(chrom, allDna, offsetArray, listArray, index13);
verbose(2, " Done first pass index\n");
}
verboseTime(1, "Done big bucket sort");
slReverse(&chromList);
itsaWriteMerged(chromList, allDna, offsetArray, listArray, index13, output);
}
