void extractPslForLfs(char *pslFile, char *lfsFile, char *out)
/* Extract psl alignments for alignments in lfs bed file */
{
struct lineFile *pf, *bf;
FILE *pslOut;
char errorFile[256] = "error.log";
struct hash *pslHash = newHash(20);
struct bed *bedList = NULL;
struct psl *pslList = NULL;
fprintf(stdout,"NOTE: Program assumes that chrom, chromStart and chromEnd \n");
fprintf(stdout, "are unique for each item in the lfs file so only one PSL \n");
fprintf(stdout, "alignment will be picked for each specific item and \n");
fprintf(stdout, "genomic location.\n\n");
fprintf(stdout, "Opening files ...\n");
bf = lineFileOpen(lfsFile, TRUE);
pslOut = mustOpen(out, "w");
stderr = mustOpen(errorFile, "w");
/* Open psl alignments file */
pf = pslFileOpen(pslFile);
/* Read in psl file of alignments */
verbose(1, "Reading psl and storing alignments file ...\n");
readPslFile(pf, &pslHash);
/* Read in lfs BED file of alignments */
verbose(1, "Reading and storing lfs alignments file ...\n");
bedList = convertLfsToBed(bf);
verbose(1, "Extracting psl alignments ... \n");
pslList = getPslList(pslHash, bedList);
/* print out the psl file without the header */
pslWriteAll(pslList, out, FALSE);
/* close files */
lineFileClose(&bf);
lineFileClose(&pf);
carefulClose(&pslOut);
carefulClose(&stderr);
}
int main(int argc, char *argv[])
{
struct lineFile *pf, *prf;
char filename[64], *pslTable;
optionInit(&argc, argv, optionSpecs);
if (argc < 4)
usage();
VERBOSE = optionExists("verbose");
SLOP = optionExists("slop");
SHORT = optionExists("short");
LONG = optionExists("long");
MISMATCH = optionExists("mismatch");
ORPHAN = optionExists("orphan");
NORANDOM = optionExists("noRandom");
NOBIN = optionExists("noBin");
MIN = optionInt("min",MIN);
MAX = optionInt("max",MAX);
TINSERT = optionInt("tInsert",TINSERT);
HARDMAX = optionInt("hardMax",HARDMAX);
if (SLOP)
SLOPVAL = optionInt("slopval",SLOPVAL);
else
SLOPVAL = 0;
NEARTOP = optionFloat("nearTop",NEARTOP);
NEARTOP = 1 - NEARTOP;
MIN_ID = optionFloat("minId",MIN_ID);
MIN_ORPHAN_ID = optionFloat("minOrphanId",MIN_ORPHAN_ID);
pf = pslFileOpen(argv[1]);
prf = lineFileOpen(argv[2],TRUE);
pslTable = cloneString(argv[3]);
sprintf(filename, "%s.pairs", argv[4]);
of = mustOpen(filename, "w");
if (SLOP)
{
sprintf(filename, "%s.slop", argv[4]);
sf = mustOpen(filename, "w");
}
if (ORPHAN)
{
sprintf(filename, "%s.orphan", argv[4]);
orf = mustOpen(filename, "w");
}
if (MISMATCH)
{
sprintf(filename, "%s.mismatch", argv[4]);
mf = mustOpen(filename, "w");
}
if (SHORT)
{
sprintf(filename, "%s.short", argv[4]);
esf = mustOpen(filename, "w");
}
if (LONG)
{
sprintf(filename, "%s.long", argv[4]);
elf = mustOpen(filename, "w");
}
if (VERBOSE)
printf("Reading pair file\n");
clones = newHash(23);
leftNames = newHash(23);
rightNames = newHash(23);
readPairFile(prf);
if (VERBOSE)
printf("Reading psl file\n");
readPslFile(pf);
if (VERBOSE)
printf("Creating Pairs\n");
pslPairs();
if (VERBOSE)
printf("Writing to files\n");
printOut(pslTable);
lineFileClose(&pf);
lineFileClose(&prf);
fclose(of);
if (ORPHAN)
fclose(orf);
if (SLOP)
fclose(sf);
if (MISMATCH)
fclose(mf);
if (SHORT)
fclose(esf);
if (LONG)
fclose(elf);
return 0;
}
