static void checkChromNameAndSize(struct lineFile *lf, char *s, unsigned chromEnd)
/* Check that the name is non-empty and exists, and that chromEnd <= chromSize. Abort on error. */
{
unsigned *chromSize;
if (strlen(s) > 0)
{
if (chrHash)
{
if ( (chromSize = hashFindVal(chrHash, s)) != NULL)
{
if (chromEnd > *chromSize)
lineFileAbort(lf, "chromEnd (%d) > chromEnd (%d)\n", chromEnd, *chromSize);
return;
}
else
{
lineFileAbort(lf, "chrom %s not found", s);
}
}
else
{
return; // chrom name not blank, and not validating against chromInfo
}
}
lineFileAbort(lf, "chrom column empty");
}
INLINE void noTabixSupport(struct lineFile *lf, char *where)
{
#ifdef USE_TABIX
if (lf->tabix != NULL)
lineFileAbort(lf, "%s: not implemented for lineFile opened with lineFileTabixMayOpen.", where);
#endif // USE_TABIX
}
void bedIntersect(char *aFile, char *bFile, char *outFile)
/* bedIntersect - Intersect two bed files. */
{
struct lineFile *lf = lineFileOpen(aFile, TRUE);
struct hash *bHash = readBed(bFile);
FILE *f = mustOpen(outFile, "w");
char *row[40];
int wordCount;
while ((wordCount = (strictTab ? lineFileChopTab(lf, row) : lineFileChop(lf, row))) != 0)
{
char *chrom = row[0];
int start = lineFileNeedNum(lf, row, 1);
int end = lineFileNeedNum(lf, row, 2);
if (start > end)
errAbort("start after end line %d of %s", lf->lineIx, lf->fileName);
if (start == end && !allowStartEqualEnd)
lineFileAbort(lf, "start==end (if this is legit, use -allowStartEqualEnd)");
struct binKeeper *bk = hashFindVal(bHash, chrom);
if (bk != NULL)
{
struct binElement *hitList = NULL, *hit;
if (allowStartEqualEnd && start == end)
hitList = binKeeperFind(bk, start-1, end+1);
else
hitList = binKeeperFind(bk, start, end);
if (aHitAny)
{
for (hit = hitList; hit != NULL; hit = hit->next)
{
float cov = getCov(start, end, hit->val);
if (cov >= minCoverage)
{
outputBed(f, row, wordCount, start, end, hit->val);
break;
}
else
{
struct bed5 *b = hit->val;
verbose(1, "filter out %s %d %d %d %d overlap %d %d %d %.3f\n",
chrom, start, end, b->start, b->end,
positiveRangeIntersection(start, end, b->start, b->end),
end-start, b->end-b->start, cov);
}
}
}
else
{
for (hit = hitList; hit != NULL; hit = hit->next)
{
if (getCov(start, end, hit->val) >= minCoverage)
outputBed(f, row, wordCount, start, end, hit->val);
}
}
slFreeList(&hitList);
}
}
}
struct hash *readBed(char *fileName)
/* Read bed and return it as a hash keyed by chromName
* with binKeeper values. */
{
char *row[5];
struct lineFile *lf = lineFileOpen(fileName, TRUE);
struct hash *hash = newHash(0);
int expectedCols = bScore ? 5 : 3;
while (lineFileNextRow(lf, row, expectedCols))
{
struct binKeeper *bk;
struct bed5 *bed;
struct hashEl *hel = hashLookup(hash, row[0]);
if (hel == NULL)
{
bk = binKeeperNew(0, 1024*1024*1024);
hel = hashAdd(hash, row[0], bk);
}
bk = hel->val;
AllocVar(bed);
bed->chrom = hel->name;
bed->start = lineFileNeedNum(lf, row, 1);
bed->end = lineFileNeedNum(lf, row, 2);
if (bScore)
bed->score = lineFileNeedNum(lf, row, 4);
if (bed->start > bed->end)
errAbort("start after end line %d of %s", lf->lineIx, lf->fileName);
if (bed->start == bed->end)
{
if (allowStartEqualEnd)
// Note we are tweaking binKeeper coords here, so use bed->start and bed->end.
binKeeperAdd(bk, max(0, bed->start-1), bed->end+1, bed);
else
lineFileAbort(lf, "start==end (if this is legit, use -allowStartEqualEnd)");
}
else
binKeeperAdd(bk, bed->start, bed->end, bed);
}
lineFileClose(&lf);
return hash;
}
struct encodePeak *encodePeakLineFileLoad(char **row, enum encodePeakType pt, struct lineFile *lf)
/* From a linefile line, load an encodePeak row. Errors outputted */
/* have line numbers, etc. Does more error checking as well. */
{
struct encodePeak *peak;
if (!pt)
errAbort("Unknown peak type set for track");
AllocVar(peak);
peak->chrom = cloneString(row[0]);
peak->chromStart = lineFileNeedNum(lf, row, 1);
peak->chromEnd = lineFileNeedNum(lf, row, 2);
peak->peak = -1;
if (peak->chromEnd < 1)
lineFileAbort(lf, "chromEnd less than 1 (%d)", peak->chromEnd);
if (peak->chromEnd < peak->chromStart)
lineFileAbort(lf, "chromStart after chromEnd (%d > %d)",
peak->chromStart, peak->chromEnd);
peak->name = cloneString(row[3]);
peak->score = lineFileNeedNum(lf, row, 4);
safecpy(peak->strand, sizeof(peak->strand), row[5]);
if (peak->strand[0] != '+' && peak->strand[0] != '-' && peak->strand[0] != '.')
lineFileAbort(lf, "Expecting +, -, or . in strand");
if (pt != gappedPeak)
/* deal with signalValue, pValue, qValue, and peak */
{
peak->signalValue = (float)lineFileNeedDouble(lf, row, 6);
peak->pValue = (float)lineFileNeedDouble(lf, row, 7);
peak->qValue = (float)lineFileNeedDouble(lf, row, 8);
if ((pt == narrowPeak) || (pt == encodePeak))
{
peak->peak = lineFileNeedNum(lf, row, 9);
if (peak->peak >= (int)peak->chromEnd)
lineFileAbort(lf, "peak site past chromEnd (%d > %d)", peak->peak, peak->chromEnd);
}
}
else /* must be gappedPeak */
/* deal with thickStart, thickEnd, itemRgb even though they're not used */
{
int thickStart = lineFileNeedNum(lf, row, 6);
int thickEnd = lineFileNeedNum(lf, row, 7);
int itemRgb = 0;
char *comma;
/* Allow comma separated list of rgb values here */
comma = strchr(row[8], ',');
if (comma)
itemRgb = bedParseRgb(row[8]);
else
itemRgb = lineFileNeedNum(lf, row, 8);
if ((thickStart != 0) || (thickEnd != 0) || (itemRgb != 0))
lineFileAbort(lf, "thickStart, thickEnd, and itemRgb columns not used in gappedPeak type, set all to 0");
}
/* Deal with blocks */
if ((pt == gappedPeak) || (pt == encodePeak))
{
int i, count;
int lastEnd, lastStart;
int blockCountIx, blockSizesIx, blockStartsIx;
if (pt == gappedPeak)
{
blockCountIx = 9;
blockSizesIx = 10;
blockStartsIx = 11;
}
else
{
blockCountIx = 10;
blockSizesIx = 11;
blockStartsIx = 12;
}
peak->blockCount = lineFileNeedNum(lf, row, blockCountIx);
sqlUnsignedDynamicArray(row[blockSizesIx], &peak->blockSizes, &count);
if (count != peak->blockCount)
lineFileAbort(lf, "expecting %d elements in array", peak->blockCount);
sqlUnsignedDynamicArray(row[blockStartsIx], &peak->blockStarts, &count);
if (count != peak->blockCount)
lineFileAbort(lf, "expecting %d elements in array", peak->blockCount);
// tell the user if they appear to be using absolute starts rather than
// relative... easy to forget! Also check block order, coord ranges...
lastStart = -1;
lastEnd = 0;
for (i=0; i < peak->blockCount; i++)
{
if (peak->blockStarts[i]+peak->chromStart >= peak->chromEnd)
{
if (peak->blockStarts[i] >= peak->chromStart)
lineFileAbort(lf,
"BED blockStarts offsets must be relative to chromStart, "
"not absolute. Try subtracting chromStart from each offset "
"in blockStarts.");
else
lineFileAbort(lf,
"BED blockStarts[i]+chromStart must be less than chromEnd.");
}
lastStart = peak->blockStarts[i];
lastEnd = peak->chromStart + peak->blockStarts[i] + peak->blockSizes[i];
}
if (peak->blockStarts[0] != 0)
lineFileAbort(lf,
"BED blocks must span chromStart to chromEnd. "
"BED blockStarts[0] must be 0 (==%d) so that (chromStart + "
"blockStarts[0]) equals chromStart.", peak->blockStarts[0]);
i = peak->blockCount-1;
if ((peak->chromStart + peak->blockStarts[i] + peak->blockSizes[i]) !=
peak->chromEnd)
{
lineFileAbort(lf,
"BED blocks must span chromStart to chromEnd. (chromStart + "
"blockStarts[last] + blockSizes[last]) must equal chromEnd.");
}
}
if (pt == gappedPeak)
/* deal with final three columns of a gappedPeak */
{
peak->signalValue = (float)lineFileNeedDouble(lf, row, 12);
peak->pValue = (float)lineFileNeedDouble(lf, row, 13);
peak->qValue = (float)lineFileNeedDouble(lf, row, 14);
}
return peak;
}
