void setPslBits(struct lineFile *lf,
Bits *bits, struct psl *psl, int winStart, int winEnd)
/* Set bits that are in psl. */
{
int i, s, e, w, blockCount = psl->blockCount;
boolean isRev = (psl->strand[1] == '-');
for (i=0; i<blockCount; ++i)
{
s = psl->tStarts[i];
e = s + psl->blockSizes[i];
if (isRev)
{
/* Use w as a temp variable to reverse coordinates s&e. */
w = psl->tSize - e;
e = psl->tSize - s;
s = w;
}
/* Clip, and if anything left set it. */
if (s < winStart) outOfRange(lf, psl->tName, psl->tSize);
if (e > winEnd) outOfRange(lf, psl->tName, psl->tSize);
w = e - s;
if (w > 0)
bitSetRange(bits, s, w);
}
}
void fbOrBed(Bits *acc, char *track, char *chrom, int chromSize)
/* Or in bits of psl file that correspond to chrom. */
{
struct lineFile *lf;
char fileName[512];
char *row[3];
int s, e, w;
chromFileName(track, chrom, fileName);
if (!fileExists(fileName))
return;
lf = lineFileOpen(fileName, TRUE);
while (lineFileRow(lf, row))
{
if (sameString(row[0], chrom))
{
s = lineFileNeedNum(lf, row, 1);
if (s < 0) outOfRange(lf, chrom, chromSize);
e = lineFileNeedNum(lf, row, 2);
if (e > chromSize) outOfRange(lf, chrom, chromSize);
w = e - s;
if (w > 0)
bitSetRange(acc, s, w);
}
}
lineFileClose(&lf);
}
struct twoBit *slurpInput(char *inName, struct hash *tbHash,
struct hash *bitmapHash)
/* Read .2bit file inName into memory and return list of twoBit items.
* Populate tbHash with twoBit items, and bitmapHash with bitmaps for
* easy masking. Both are hashed by twoBit sequence name. */
{
struct twoBit *twoBitList = NULL;
struct twoBit *twoBit = NULL;
twoBitList = twoBitFromFile(inName);
/* Free and clear the masking data (unless -add). Hash twoBits by name. */
for (twoBit = twoBitList; twoBit != NULL; twoBit = twoBit->next)
{
Bits *bits = bitAlloc(twoBit->size);
if (add)
{
/* Store the currently masked bits: */
int i;
for (i = 0; i < twoBit->maskBlockCount; i++)
{
bitSetRange(bits, twoBit->maskStarts[i], twoBit->maskSizes[i]);
}
}
/* Free the current representation of masking -- it will be replaced. */
twoBit->maskBlockCount = 0;
freez(&(twoBit->maskStarts));
freez(&(twoBit->maskSizes));
/* Hash twoBit and our new bitmap by sequence name. */
hashAddUnique(tbHash, twoBit->name, twoBit);
hashAddUnique(bitmapHash, twoBit->name, bits);
}
return twoBitList;
}
Bits *getMaskedBits(struct sqlConnection *conn, struct chromInfo *chrom)
/* Get bit array with parts that are masked by simple repeats etc. masked
* out. */
{
char query[512];
char **row;
struct sqlResult *sr;
char table[128];
struct wabaChromHit *wchList = NULL, *wch;
struct rmskOut ro;
Bits *b = bitAlloc(chrom->size);
int allCount = 0;
int simpCount = 0;
sqlSafef(query, sizeof query, "select * from %s_rmsk",
chrom->chrom);
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
++allCount;
rmskOutStaticLoad(row, &ro);
if (sameString(ro.repClass, "Simple_repeat") || sameString(ro.repClass, "Low_complexity"))
{
++simpCount;
assert(ro.genoEnd <= chrom->size);
bitSetRange(b, ro.genoStart, ro.genoEnd - ro.genoStart);
}
}
printf("Got %d repeats, %d simple/low complexity\n", allCount, simpCount);
sqlFreeResult(&sr);
return b;
}
void setOverlapBits(int s, int e, int overlapStart, int overlapEnd, Bits *bits)
/* Set part of bits corresponding to where s-e overlaps with overlapStart-End */
{
s = max(s, overlapStart);
e = min(e, overlapEnd);
if (s < e)
bitSetRange(bits, s - overlapStart, e - s);
}
void setWithPad(struct chrom *chrom, int s, int e)
/* Set bits from s to e with padding. */
{
s -= pad;
if (s < 0) s = 0;
e += pad;
if (e > chrom->size) e = chrom->size;
bitSetRange(chrom->bits, s, e-s);
}
void addMasking(struct hash *bitmapHash, char *seqName,
unsigned start, unsigned end)
/* Set bits in range. */
{
if (end > start)
{
Bits *bits = (Bits *)hashMustFindVal(bitmapHash, seqName);
bitSetRange(bits, start, (end - start));
}
}
static void bitSetClippedRange(Bits *bits, int bitSize, int s, int e)
/* Clip start and end to [0,bitSize) and set range. */
{
int w;
if (e > bitSize) e = bitSize;
if (s < 0) s = 0;
w = e - s;
if (w > 0)
bitSetRange(bits, s, w);
}
void addMasking(struct hash *twoBitHash, struct hash *bitmapHash, char *seqName,
unsigned start, unsigned end)
/* Set bits in range. */
{
if (end > start)
{
struct twoBit *tb = (struct twoBit *)hashMustFindVal(twoBitHash, seqName);
if ((end > tb->size) || (start >= tb->size))
errAbort("bed range (%d - %d) is off the end of chromosome %s size %d",
start, end, seqName, tb->size);
Bits *bits = (Bits *)hashMustFindVal(bitmapHash, seqName);
bitSetRange(bits, start, (end - start));
}
}
boolean bigWigValsOnChromFetchData(struct bigWigValsOnChrom *chromVals, char *chrom,
struct bbiFile *bigWig)
/* Fetch data for chromosome from bigWig. Returns FALSE if not data on that chrom. */
{
/* Fetch chromosome and size into self. */
freeMem(chromVals->chrom);
chromVals->chrom = cloneString(chrom);
long chromSize = chromVals->chromSize = bbiChromSize(bigWig, chrom);
if (chromSize <= 0)
return FALSE;
/* Make sure buffers are big enough. */
if (chromSize > chromVals->bufSize)
{
freeMem(chromVals->valBuf);
freeMem(chromVals->covBuf);
chromVals->valBuf = needHugeMem((sizeof(double))*chromSize);
chromVals->covBuf = bitAlloc(chromSize);
chromVals->bufSize = chromSize;
}
/* Zero out buffers */
bitClear(chromVals->covBuf, chromSize);
double *valBuf = chromVals->valBuf;
int i;
for (i=0; i<chromSize; ++i)
valBuf[i] = 0.0;
fetchIntoBuf(bigWig, chrom, 0, chromSize, chromVals);
#ifdef OLD
/* Fetch intervals for this chromosome and fold into buffers. */
struct lm *lm = lmInit(0);
struct bbiInterval *iv, *ivList = bigWigIntervalQuery(bigWig, chrom, 0, chromSize, lm);
for (iv = ivList; iv != NULL; iv = iv->next)
{
double val = iv->val;
int end = iv->end;
for (i=iv->start; i<end; ++i)
valBuf[i] = val;
bitSetRange(chromVals->covBuf, iv->start, iv->end - iv->start);
}
lmCleanup(&lm);
#endif /* OLD */
return TRUE;
}
void bitsForOut(char *fileName, int seqSize, Bits *bits)
/* Get bitmap that corresponds to outFile. */
{
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *line, *words[8];
int wordCount;
boolean firstTime = TRUE;
int start,end;
/* Check and skip over three line header */
if (!lineFileNext(lf, &line, NULL))
errAbort("%s is empty", fileName);
line = skipLeadingSpaces(line);
if (!startsWith("SW", line))
errAbort("%s is not a RepeatMasker .out file", fileName);
lineFileNext(lf, &line, NULL);
if (!startsWith("score", line))
errAbort("%s is not a RepeatMasker .out file", fileName);
lineFileNext(lf, &line, NULL);
for (;;)
{
if (!lineFileNext(lf, &line, NULL))
break;
wordCount = chopLine(line, words);
if (wordCount < 8)
errAbort("Short line %d of %s\n", lf->lineIx, lf->fileName);
start = lineFileNeedNum(lf, words, 5) - 1;
end = lineFileNeedNum(lf, words, 6);
if (start > end)
errAbort("Start after end line %d of %s", lf->lineIx, lf->fileName);
if (firstTime)
{
char *s = words[7];
if (s[0] != '(' || !isdigit(s[1]))
errAbort("Expected parenthesized number line %d of %s", lf->lineIx, lf->fileName);
if (seqSize != end + atoi(s+1))
errAbort("Size mismatch line %d of %s", lf->lineIx, lf->fileName);
firstTime = FALSE;
}
if (end > seqSize)
errAbort("End past bounds line %d of %s", lf->lineIx, lf->fileName);
bitSetRange(bits, start, end-start);
}
lineFileClose(&lf);
}
static void covAddRange(struct covStats *cov, int start, int end)
/* Add range to stats. */
{
struct region *r = cov->region;
if (end > r->start && start < r->end)
{
int unclippedSize, size;
unclippedSize = end - start;
if (start < r->start) start = r->start;
if (end > r->end) end = r->end;
size = end-start;
bitSetRange(cov->bits, start - r->start, size);
cov->sumBases += size;
cov->itemCount += 1;
if (unclippedSize > cov->maxBases) cov->maxBases = unclippedSize;
if (unclippedSize < cov->minBases) cov->minBases = unclippedSize;
}
}
static struct visiMatch *visiSearcherAdd(struct visiSearcher *searcher,
int imageId, double weight, int startWord, int wordCount)
/* Add given weight to match involving imageId, creating
* a fresh match if necessary for imageId. */
{
struct visiMatch key, *match;
key.imageId = imageId;
match = rbTreeFind(searcher->tree, &key);
if (match == NULL)
{
match = visiMatchNew(imageId, searcher->wordCount);
slAddHead(&searcher->matchList, match);
rbTreeAdd(searcher->tree, match);
}
match->weight += weight;
assert(startWord + wordCount <= searcher->wordCount);
bitSetRange(match->wordBits, startWord, wordCount);
return match;
}
void fbOrChain(Bits *acc, char *track, char *chrom, int chromSize)
/* Or in a chain file. */
{
struct lineFile *lf;
char fileName[512];
struct chain *chain;
struct cBlock *b;
chromFileName(track, chrom, fileName);
if (!fileExists(fileName))
return;
lf = lineFileOpen(fileName, TRUE);
while ((chain = chainRead(lf)) != NULL)
{
for (b = chain->blockList; b != NULL; b = b->next)
{
int s = b->tStart, e = b->tEnd;
if (s < 0) outOfRange(lf, chrom, chromSize);
if (e > chromSize) outOfRange(lf, chrom, chromSize);
bitSetRange(acc, b->tStart, b->tEnd - b->tStart);
}
chainFree(&chain);
}
}
static void fetchIntoBuf(struct bbiFile *bwf, char *chrom, bits32 start, bits32 end,
struct bigWigValsOnChrom *chromVals)
/* Get data for interval. Return list allocated out of lm. */
{
/* A lot of code duplicated with bigWigIntervalQuery, but here the clipping
* is simplified since always working across full chromosome, and the output is
* different. Since both of these are in inner loops and speed critical, it's hard
* to factor out without perhaps making it worse than the bit of duplication. */
if (bwf->typeSig != bigWigSig)
errAbort("Trying to do fetchIntoBuf on a non big-wig file.");
bbiAttachUnzoomedCir(bwf);
struct fileOffsetSize *blockList = bbiOverlappingBlocks(bwf, bwf->unzoomedCir,
chrom, start, end, NULL);
struct fileOffsetSize *block, *beforeGap, *afterGap;
struct udcFile *udc = bwf->udc;
boolean isSwapped = bwf->isSwapped;
float val;
int i;
Bits *covBuf = chromVals->covBuf;
double *valBuf = chromVals->valBuf;
/* Set up for uncompression optionally. */
char *uncompressBuf = NULL;
if (bwf->uncompressBufSize > 0)
uncompressBuf = needLargeMem(bwf->uncompressBufSize);
/* This loop is a little complicated because we merge the read requests for efficiency, but we
* have to then go back through the data one unmerged block at a time. */
for (block = blockList; block != NULL; )
{
/* Find contigious blocks and read them into mergedBuf. */
fileOffsetSizeFindGap(block, &beforeGap, &afterGap);
bits64 mergedOffset = block->offset;
bits64 mergedSize = beforeGap->offset + beforeGap->size - mergedOffset;
udcSeek(udc, mergedOffset);
char *mergedBuf = needLargeMem(mergedSize);
udcMustRead(udc, mergedBuf, mergedSize);
char *blockBuf = mergedBuf;
/* Loop through individual blocks within merged section. */
for (;block != afterGap; block = block->next)
{
/* Uncompress if necessary. */
char *blockPt, *blockEnd;
if (uncompressBuf)
{
blockPt = uncompressBuf;
int uncSize = zUncompress(blockBuf, block->size, uncompressBuf, bwf->uncompressBufSize);
blockEnd = blockPt + uncSize;
}
else
{
blockPt = blockBuf;
blockEnd = blockPt + block->size;
}
/* Deal with insides of block. */
struct bwgSectionHead head;
bwgSectionHeadFromMem(&blockPt, &head, isSwapped);
switch (head.type)
{
case bwgTypeBedGraph:
{
for (i=0; i<head.itemCount; ++i)
{
bits32 s = memReadBits32(&blockPt, isSwapped);
bits32 e = memReadBits32(&blockPt, isSwapped);
bitSetRange(covBuf, s, e-s);
val = memReadFloat(&blockPt, isSwapped);
bits32 j;
for (j=s; j<e; ++j)
valBuf[j] = val;
}
break;
}
case bwgTypeVariableStep:
{
for (i=0; i<head.itemCount; ++i)
{
bits32 s = memReadBits32(&blockPt, isSwapped);
val = memReadFloat(&blockPt, isSwapped);
bitSetRange(covBuf, s, head.itemSpan);
bits32 e = s + head.itemSpan;
bits32 j;
for (j=s; j<e; ++j)
valBuf[j] = val;
}
break;
}
case bwgTypeFixedStep:
{
/* Do a little optimization for the most common and worst case - step1/span1 */
if (head.itemStep == 1 && head.itemSpan == 1)
{
bits32 s = head.start;
bits32 e = head.end;
bitSetRange(covBuf, s, e-s);
bits32 j;
for (j=s; j<e; ++j)
valBuf[j] = memReadFloat(&blockPt, isSwapped);
}
else
{
bits32 s = head.start;
bits32 e = s + head.itemSpan;
for (i=0; i<head.itemCount; ++i)
{
bitSetRange(covBuf, s, head.itemSpan);
val = memReadFloat(&blockPt, isSwapped);
bits32 j;
for (j=s; j<e; ++j)
valBuf[j] = val;
s += head.itemStep;
e += head.itemStep;
}
}
break;
}
default:
internalErr();
break;
}
assert(blockPt == blockEnd);
blockBuf += block->size;
}
freeMem(mergedBuf);
}
freeMem(uncompressBuf);
slFreeList(&blockList);
}
