static struct joinedRow *jrRowAdd(struct joinedTables *joined, char **row,
int fieldCount, int keyCount)
/* Add new row to joinable table. */
{
if (joined->maxRowCount != 0 && joined->rowCount >= joined->maxRowCount)
{
warn("Stopping after %d rows, try restricting region or adding filter",
joined->rowCount);
return NULL;
}
else
{
struct joinedRow *jr;
int i;
struct lm *lm = joined->lm;
lmAllocVar(lm, jr);
lmAllocArray(lm, jr->fields, joined->fieldCount);
lmAllocArray(lm, jr->keys, joined->keyCount);
jr->passedFilter = TRUE;
for (i=0; i<fieldCount; ++i)
jr->fields[i] = lmSlName(lm, row[i]);
row += fieldCount;
for (i=0; i<keyCount; ++i)
jr->keys[i] = lmSlName(lm, row[i]);
slAddHead(&joined->rowList, jr);
joined->rowCount += 1;
return jr;
}
}
struct cdsEvidence *orfsOnRna(struct dnaSeq *seq, struct hash *nmdHash, struct hash *mafHash,
int otherSpeciesCount, boolean anyStart)
/* Return scored list of all ORFs on RNA. */
{
DNA *dna = seq->dna;
int lastPos = seq->size - 3;
int startPos;
struct cdsEvidence *orfList = NULL, *orf;
struct lm *lm = lmInit(64*1024);
/* Figure out the key piece of info for NMD. */
int lastIntronPos = findLastIntronPos(nmdHash, seq->name);
double orthoWeightPer = 0;
struct orthoCdsArray *orthoList = NULL;
/* Calculate stuff useful for orthology */
if (otherSpeciesCount > 0)
{
orthoWeightPer = 1.0/otherSpeciesCount;
struct mafAli *maf = hashFindVal(mafHash, seq->name);
if (maf != NULL)
{
orthoList = calcOrthoList(maf, lm);
// uglyf("%s: ", seq->name);
// dumpOrthoArray(orthoArray, uglyOut);
}
}
/* Allocate some arrays that keep track of bases in
* upstream. This dramatically speeds up processing
* of TTN and other long transcripts which otherwise
* can take almost a minute each. */
int *upAtgCount, *upKozakCount;
lmAllocArray(lm, upAtgCount, seq->size);
lmAllocArray(lm, upKozakCount, seq->size);
calcUpstreams(seq, upAtgCount, upKozakCount);
/* Go through sequence making up a record for each
* start codon we find. */
for (startPos=0; startPos<=lastPos; ++startPos)
{
if (startsWith("atg", dna+startPos) || (anyStart && startPos < 3))
{
int stopPos = orfEndInSeq(seq, startPos);
orf = createCds(seq, startPos, stopPos, upAtgCount, upKozakCount,
lastIntronPos, orthoList, orthoWeightPer);
slAddHead(&orfList, orf);
}
}
slReverse(&orfList);
/* Clean up and go home. */
lmCleanup(&lm);
return orfList;
}
struct pslSets *pslSetsNew(int numSets)
/* construct a new pslSets object */
{
struct pslSets *ps;
AllocVar(ps);
ps->lm = lmInit(1024*1024);
ps->numSets = numSets;
lmAllocArray(ps->lm, ps->sets, numSets);
lmAllocArray(ps->lm, ps->pending, numSets);
return ps;
}
struct annoRow *aggvIntergenicRow(struct annoGratorGpVar *self, struct variant *variant,
boolean *retRJFilterFailed, struct lm *callerLm)
/* If intergenic variants (no overlapping or nearby genes) are to be included in output,
* make an output row with empty genePred and a gpFx that is empty except for soNumber. */
{
struct annoGrator *gSelf = &(self->grator);
struct annoStreamer *sSelf = &(gSelf->streamer);
char **wordsOut;
lmAllocArray(self->lm, wordsOut, sSelf->numCols);
// Add empty strings for genePred string columns:
int gpColCount = gSelf->mySource->numCols;
int i;
for (i = 0; i < gpColCount; i++)
wordsOut[i] = "";
struct gpFx *intergenicGpFx;
lmAllocVar(self->lm, intergenicGpFx);
intergenicGpFx->allele = firstAltAllele(variant->alleles);
if (isAllNt(intergenicGpFx->allele, strlen(intergenicGpFx->allele)))
touppers(intergenicGpFx->allele);
intergenicGpFx->soNumber = intergenic_variant;
intergenicGpFx->detailType = none;
aggvStringifyGpFx(&wordsOut[gpColCount], intergenicGpFx, self->lm);
boolean rjFail = (retRJFilterFailed && *retRJFilterFailed);
return annoRowFromStringArray(variant->chrom, variant->chromStart, variant->chromEnd, rjFail,
wordsOut, sSelf->numCols, callerLm);
}
struct hash *allChainsHash(char *fileName)
/* Hash all the chains in a given file by their ids. */
{
struct hash *chainHash = newHash(18);
struct lineFile *lf = lineFileOpen(fileName, TRUE);
struct chain *chain;
char chainId[20];
struct lm *lm = chainHash->lm;
struct rbTreeNode **stack;
lmAllocArray(lm, stack, 128);
while ((chain = chainRead(lf)) != NULL)
{
struct indexedChain *ixc;
lmAllocVar(lm, ixc);
ixc->chain = chain;
#ifdef SOON
#endif /* SOON */
ixc->blockTree = rangeTreeNewDetailed(lm, stack);
struct cBlock *block;
for (block = chain->blockList; block != NULL; block = block->next)
{
struct range *r = rangeTreeAdd(ixc->blockTree, block->tStart, block->tEnd);
r->val = block;
}
safef(chainId, sizeof(chainId), "%x", chain->id);
hashAddUnique(chainHash, chainId, ixc);
}
lineFileClose(&lf);
return chainHash;
}
static void rowBufInit(struct rowBuf *rowBuf, int size)
/* Clean up rowBuf and give it a new lm and buffer[size]. */
{
resetRowBuf(rowBuf);
rowBuf->lm = lmInit(0);
rowBuf->size = size;
lmAllocArray(rowBuf->lm, rowBuf->buf, size);
}
static struct pslMatches *pslMatchesAlloc(struct pslSets *ps)
/* allocate a matches object, either new or from the recycled list */
{
struct pslMatches *pm = slPopHead(&ps->matchesPool);
if (pm == NULL)
{
lmAllocVar(ps->lm, pm);
lmAllocArray(ps->lm, pm->psls, ps->numSets);
}
pm->numSets = ps->numSets;
return pm;
}
static struct bwgFixedStepPacked *
createFixedStepItems(double *score, int len, struct lm *lm)
{
struct bwgFixedStepPacked *packed;
lmAllocArray(lm, packed, len);
int i;
for (i=0; i<len; ++i)
{
packed[i].val = score[i];
}
return packed;
}
static struct bwgVariableStepPacked *
createVariableStepItems(int *start, double *score, int len, struct lm *lm)
{
struct bwgVariableStepPacked *packed;
lmAllocArray(lm, packed, len);
int i;
for (i=0; i<len; ++i)
{
packed[i].start = start[i] - 1;
packed[i].val = score[i];
}
return packed;
}
static struct ffAli *ffFindExtendNmers(char *nStart, char *nEnd, char *hStart, char *hEnd,
int seedSize)
/* Find perfectly matching n-mers and extend them. */
{
struct lm *lm = lmInit(32*1024);
struct seqHashEl **hashTable, *hashEl, **hashSlot;
struct ffAli *ffList = NULL, *ff;
char *n = nStart, *h = hStart, *ne = nEnd - seedSize, *he = hEnd - seedSize;
/* Hash the needle. */
lmAllocArray(lm, hashTable, 4*1024);
while (n <= ne)
{
if (!totalDegenerateN(n, seedSize))
{
hashSlot = ffHashFuncN(n, seedSize) + hashTable;
lmAllocVar(lm, hashEl);
hashEl->seq = n;
slAddHead(hashSlot, hashEl);
}
++n;
}
/* Scan the haystack adding hits. */
while (h <= he)
{
for (hashEl = hashTable[ffHashFuncN(h, seedSize)];
hashEl != NULL; hashEl = hashEl->next)
{
if (memcmp(hashEl->seq, h, seedSize) == 0)
{
AllocVar(ff);
ff->hStart = h;
ff->hEnd = h + seedSize;
ff->nStart = hashEl->seq;
ff->nEnd = hashEl->seq + seedSize;
extendExactLeft(ff->nStart - nStart, ff->hStart - hStart,
&ff->nStart, &ff->hStart);
extendExactRight(nEnd - ff->nEnd, hEnd - ff->hEnd, &ff->nEnd, &ff->hEnd);
ff->left = ffList;
ffList = ff;
}
}
++h;
}
ffList = ffMakeRightLinks(ffList);
ffList = ffMergeClose(ffList, nStart, hStart);
lmCleanup(&lm);
return ffList;
}
struct annoRow *annoRowFromStringArray(char *chrom, uint start, uint end, boolean rightJoinFail,
char **wordsIn, int numCols, struct lm *lm)
/* Allocate & return an annoRow with words cloned from wordsIn. */
{
struct annoRow *aRow;
lmAllocVar(lm, aRow);
aRow->chrom = lmCloneString(lm, chrom);
aRow->start = start;
aRow->end = end;
aRow->rightJoinFail = rightJoinFail;
char **words;
lmAllocArray(lm, words, numCols);
int i;
for (i = 0; i < numCols; i++)
words[i] = lmCloneString(lm, wordsIn[i]);
aRow->data = words;
return aRow;
}
struct orthoCdsArray *calcOrthoList(struct mafAli *maf, struct lm *lm)
/* Given maf, figure out orthoCdsArray list, one for each other
* species. (Assume first species is native.) */
{
struct orthoCdsArray *array, *arrayList = NULL;
struct mafComp *nativeComp = maf->components;
int nativeSize = nativeComp->size;
struct mafComp *comp;
for (comp = maf->components->next; comp != NULL; comp = comp->next)
{
AllocVar(array);
array->species = lmCloneString(lm, comp->src);
array->arraySize = nativeSize;
lmAllocArray(lm, array->cdsArray, nativeSize);
fillInArrayFromPair(lm, nativeComp, comp, array->cdsArray, nativeSize, maf->textSize);
slAddHead(&arrayList, array);
}
slReverse(&arrayList);
return arrayList;
}
static struct annoRow *aggvEffectToRow(struct annoGratorGpVar *self, struct gpFx *effect,
struct annoRow *rowIn, struct lm *callerLm)
// convert a single genePred annoRow and gpFx record to an augmented genePred annoRow;
{
struct annoGrator *gSelf = &(self->grator);
struct annoStreamer *sSelf = &(gSelf->streamer);
assert(sSelf->numCols > gSelf->mySource->numCols);
char **wordsOut;
lmAllocArray(self->lm, wordsOut, sSelf->numCols);
// copy the genePred fields over
int gpColCount = gSelf->mySource->numCols;
char **wordsIn = (char **)rowIn->data;
memcpy(wordsOut, wordsIn, sizeof(char *) * gpColCount);
// stringify the gpFx structure
aggvStringifyGpFx(&wordsOut[gpColCount], effect, callerLm);
return annoRowFromStringArray(rowIn->chrom, rowIn->start, rowIn->end, rowIn->rightJoinFail,
wordsOut, sSelf->numCols, callerLm);
}
static void parseVariableStepSection(struct lineFile *lf, boolean clipDontDie, struct lm *lm,
int itemsPerSlot, char *chrom, int chromSize, bits32 span, struct bwgSection **pSectionList)
/* Read the single column data in section until get to end. */
{
struct lm *lmLocal = lmInit(0);
/* Stream through section until get to end of file or next section,
* adding values from single column to list. */
char *words[2];
char *line;
struct bwgVariableStepItem *item, *nextItem, *itemList = NULL;
int originalSectionSize = 0;
while (lineFileNextReal(lf, &line))
{
if (steppedSectionEnd(line, 2))
{
lineFileReuse(lf);
break;
}
chopLine(line, words);
lmAllocVar(lmLocal, item);
int start = lineFileNeedNum(lf, words, 0);
if (start <= 0)
{
errAbort("line %d of %s: zero or negative chromosome coordinate not allowed",
lf->lineIx, lf->fileName);
}
item->start = start - 1;
item->val = lineFileNeedDouble(lf, words, 1);
if (item->start + span > chromSize)
{
warn("line %d of %s: chromosome %s has %u bases, but item ends at %u",
lf->lineIx, lf->fileName, chrom, chromSize, item->start + span);
if (!clipDontDie)
noWarnAbort();
}
else
{
slAddHead(&itemList, item);
++originalSectionSize;
}
}
slSort(&itemList, bwgVariableStepItemCmp);
/* Make sure no overlap between items. */
if (itemList != NULL)
{
item = itemList;
for (nextItem = item->next; nextItem != NULL; nextItem = nextItem->next)
{
if (item->start + span > nextItem->start)
errAbort("Overlap on %s between items starting at %d and %d.\n"
"Please remove overlaps and try again",
chrom, item->start, nextItem->start);
item = nextItem;
}
}
/* Break up into sections of no more than items-per-slot size. */
int sizeLeft = originalSectionSize;
for (item = itemList; item != NULL; )
{
/* Figure out size of this section */
int sectionSize = sizeLeft;
if (sectionSize > itemsPerSlot)
sectionSize = itemsPerSlot;
sizeLeft -= sectionSize;
/* Convert from list to array representation. */
struct bwgVariableStepPacked *packed, *p;
p = lmAllocArray(lm, packed, sectionSize);
int i;
for (i=0; i<sectionSize; ++i)
{
p->start = item->start;
p->val = item->val;
item = item->next;
++p;
}
/* Fill in section and add it to list. */
struct bwgSection *section;
lmAllocVar(lm, section);
section->chrom = chrom;
section->start = packed[0].start;
section->end = packed[sectionSize-1].start + span;
section->type = bwgTypeVariableStep;
section->items.variableStepPacked = packed;
section->itemSpan = span;
section->itemCount = sectionSize;
slAddHead(pSectionList, section);
}
lmCleanup(&lmLocal);
}
static void parseFixedStepSection(struct lineFile *lf, boolean clipDontDie, struct lm *lm,
int itemsPerSlot, char *chrom, bits32 chromSize, bits32 span, bits32 sectionStart,
bits32 step, struct bwgSection **pSectionList)
/* Read the single column data in section until get to end. */
{
struct lm *lmLocal = lmInit(0);
/* Stream through section until get to end of file or next section,
* adding values from single column to list. */
char *words[1];
char *line;
struct bwgFixedStepItem *item, *itemList = NULL;
int originalSectionSize = 0;
bits32 sectionEnd = sectionStart;
while (lineFileNextReal(lf, &line))
{
if (steppedSectionEnd(line, 1))
{
lineFileReuse(lf);
break;
}
chopLine(line, words);
lmAllocVar(lmLocal, item);
item->val = lineFileNeedDouble(lf, words, 0);
if (sectionEnd + span > chromSize)
{
warn("line %d of %s: chromosome %s has %u bases, but item ends at %u",
lf->lineIx, lf->fileName, chrom, chromSize, sectionEnd + span);
if (!clipDontDie)
noWarnAbort();
}
else
{
slAddHead(&itemList, item);
++originalSectionSize;
}
sectionEnd += step;
}
slReverse(&itemList);
/* Break up into sections of no more than items-per-slot size, and convert to packed format. */
int sizeLeft = originalSectionSize;
for (item = itemList; item != NULL; )
{
/* Figure out size of this section */
int sectionSize = sizeLeft;
if (sectionSize > itemsPerSlot)
sectionSize = itemsPerSlot;
sizeLeft -= sectionSize;
/* Allocate and fill in section. */
struct bwgSection *section;
lmAllocVar(lm, section);
section->chrom = chrom;
section->start = sectionStart;
sectionStart += sectionSize * step;
section->end = sectionStart - step + span;
section->type = bwgTypeFixedStep;
section->itemStep = step;
section->itemSpan = span;
section->itemCount = sectionSize;
/* Allocate array for data, and copy from list to array representation */
struct bwgFixedStepPacked *packed; /* An array */
section->items.fixedStepPacked = lmAllocArray(lm, packed, sectionSize);
int i;
for (i=0; i<sectionSize; ++i)
{
packed->val = item->val;
item = item->next;
++packed;
}
/* Add section to list. */
slAddHead(pSectionList, section);
}
lmCleanup(&lmLocal);
}
void fillInArrayFromPair(struct lm *lm, struct mafComp *native, struct mafComp *xeno,
struct orthoCds *array, int arraySize, int symCount)
/* Figure out the CDS in xeno for each position in native. */
{
char *nText = native->text, *xText = xeno->text;
int nSize = arraySize, xSize = symCount - countChars(xText, '-');
/* Create an array that for each point in native gives you the index of corresponding
* point in xeno, and another array that does the opposite. */
int *nToX, *xToN;
lmAllocArray(lm, nToX, nSize+1);
lmAllocArray(lm, xToN, xSize+1);
int i;
int nIx = 0, xIx = 0;
for (i=0; i<symCount; ++i)
{
char n = nText[i], x = xText[i];
if (n == '.')
errAbort("Dot in native component %s of maf. Can't handle it.", native->src);
nToX[nIx] = xIx;
xToN[xIx] = nIx;
if (n != '-')
{
array[nIx].base = x;
nToX[nIx] = xIx;
++nIx;
}
if (x != '-')
++xIx;
}
assert(xIx == xSize);
assert(nIx == nSize);
/* Put an extra value at end of arrays to simplify logic. */
nToX[nSize] = xSize;
xToN[xSize] = nSize;
/* Create xeno sequence without the '-' chars */
char *xDna = lmCloneString(lm, xText);
tolowers(xDna);
stripChar(xDna, '-');
#ifdef DEBUG
uglyf("xToN:");
for (i=0; i<xSize; ++i) uglyf(" %d", xToN[i]);
uglyf("\n");
#endif /* DEBUG */
/* Step through this, one frame at a time, looking for best ORF */
int frame;
for (frame=0; frame<3; ++frame)
{
/* Calculate some things constant for this frame, and deal with
* ORF that starts at beginning (may not have ATG) */
int lastPos = xSize-3;
int frameDnaSize = xSize-frame;
int start = frame, end = findOrfEnd(xDna, frameDnaSize, frame);
applyOrf(start, end, xDna, xToN, array, arraySize);
for (start = end; start<=lastPos; )
{
// uglyf("start %d %c%c%c\n", start, xDna[start], xDna[start+1], xDna[start+2]);
if (startsWith("atg", xDna+start))
{
end = findOrfEnd(xDna, frameDnaSize, start);
applyOrf(start, end, xDna, xToN, array, arraySize);
start = end;
}
else
start += 3;
}
}
}
struct wordTree *wordTreeForChainsInFile(char *fileName, int chainSize, struct lm *lm)
/* Return a wordTree of all chains-of-words of length chainSize seen in file.
* Allocate the structure in local memory pool lm. */
{
/* Stuff for processing file a line at a time. */
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *line, *word;
/* We'll keep a chain of three or so words in a doubly linked list. */
struct dlNode *node;
struct dlList *chain = dlListNew();
int curSize = 0;
/* We'll build up the tree starting with an empty root node. */
struct wordTree *wt = wordTreeNew("");
int wordCount = 0;
/* Save time/space by sharing stack between all "following" rbTrees. */
struct rbTreeNode **stack;
lmAllocArray(lm, stack, 256);
/* Loop through each line of input file, lowercasing the whole line, and then
* looping through each word of line, stripping out special chars, and finally
* processing each word. */
while (lineFileNext(lf, &line, NULL))
{
if (lower)
tolowers(line);
while ((word = nextWord(&line)) != NULL)
{
if (unpunc)
{
stripChar(word, ',');
stripChar(word, '.');
stripChar(word, ';');
stripChar(word, '-');
stripChar(word, '"');
stripChar(word, '?');
stripChar(word, '!');
stripChar(word, '(');
stripChar(word, ')');
if (word[0] == 0)
continue;
}
verbose(2, "%s\n", word);
/* We come to this point in the code for each word in the file.
* Here we want to maintain a chain of sequential words up to
* chainSize long. We do this with a doubly-linked list structure.
* For the first few words in the file we'll just build up the list,
* only adding it to the tree when we finally do get to the desired
* chain size. Once past the initial section of the file we'll be
* getting rid of the first link in the chain as well as adding a new
* last link in the chain with each new word we see. */
if (curSize < chainSize)
{
dlAddValTail(chain, cloneString(word));
++curSize;
if (curSize == chainSize)
addChainToTree(wt, chain, lm, stack);
}
else
{
/* Reuse doubly-linked-list node, but give it a new value, as we move
* it from head to tail of list. */
node = dlPopHead(chain);
freeMem(node->val);
node->val = cloneString(word);
dlAddTail(chain, node);
addChainToTree(wt, chain, lm, stack);
}
++wordCount;
}
}
/* Handle last few words in file, where can't make a chain of full size. Need
* a special case for file that has fewer than chain size words too. */
if (curSize < chainSize)
addChainToTree(wt, chain, lm, stack);
while ((node = dlPopHead(chain)) != NULL)
{
addChainToTree(wt, chain, lm, stack);
freeMem(node->val);
freeMem(node);
}
dlListFree(&chain);
lineFileClose(&lf);
return wt;
}
void chainNet(char *chainFile, char *tSizes, char *qSizes,
char *tNet, char *qNet)
/* chainNet - Make alignment nets out of chains. */
{
struct lineFile *lf = lineFileOpen(chainFile, TRUE);
struct hash *qHash, *tHash;
struct chrom *qChromList, *tChromList, *tChrom, *qChrom;
struct chain *chain;
double lastScore = -1;
struct lm *lm = lmInit(0);
struct rbTreeNode **rbStack;
FILE *tNetFile = mustOpen(tNet, "w");
FILE *qNetFile = mustOpen(qNet, "w");
lmAllocArray(lm, rbStack, 256);
makeChroms(qSizes, lm, rbStack, &qHash, &qChromList);
makeChroms(tSizes, lm, rbStack, &tHash, &tChromList);
verbose(1, "Got %d chroms in %s, %d in %s\n", slCount(tChromList), tSizes,
slCount(qChromList), qSizes);
lineFileSetMetaDataOutput(lf, tNetFile);
lineFileSetMetaDataOutput(lf, qNetFile);
/* Loop through chain file building up net. */
while ((chain = chainRead(lf)) != NULL)
{
/* Make sure that input is really sorted. */
if (lastScore >= 0 && chain->score > lastScore)
errAbort("%s must be sorted in order of score", chainFile);
lastScore = chain->score;
if (chain->score < minScore)
{
break;
}
verbose(2, "chain %f (%d els) %s %d-%d %c %s %d-%d\n",
chain->score, slCount(chain->blockList),
chain->tName, chain->tStart, chain->tEnd,
chain->qStrand, chain->qName, chain->qStart, chain->qEnd);
qChrom = hashMustFindVal(qHash, chain->qName);
if (qChrom->size != chain->qSize)
errAbort("%s is %d in %s but %d in %s", chain->qName,
chain->qSize, chainFile,
qChrom->size, qSizes);
tChrom = hashMustFindVal(tHash, chain->tName);
if (tChrom->size != chain->tSize)
errAbort("%s is %d in %s but %d in %s", chain->tName,
chain->tSize, chainFile,
tChrom->size, tSizes);
if (!inclQuery(chain))
verbose(2, "skipping chain on query %s\n", chain->qName);
else
{
addChain(qChrom, tChrom, chain);
verbose(2, "%s has %d inserts, %s has %d\n", tChrom->name,
tChrom->spaces->n, qChrom->name, qChrom->spaces->n);
}
}
/* Build up other side of fills. It's just for historical
* reasons this is not done during the main build up.
* It's a little less efficient this way, but to change it
* some hard reverse strand issues would have to be juggled. */
verbose(1, "Finishing nets\n");
finishNet(qChromList, TRUE);
finishNet(tChromList, FALSE);
/* Write out basic net files. */
verbose(1, "writing %s\n", tNet);
outputNetSide(tChromList, tNetFile, FALSE);
verbose(1, "writing %s\n", qNet);
outputNetSide(qChromList, qNetFile, TRUE);
/* prevent SIGPIPE in preceding process if input is a pipe, consume remainder
* of input file since we stop before EOF. */
if (isPipe(lf->fd))
{
char *line;
while(lineFileNext(lf, &line, NULL))
continue;
}
lineFileClose(&lf);
if (verboseLevel() > 1)
printMem(stderr);
}
struct bed *bedFromRow(
char *chrom, /* Chromosome bed is on. */
char **row, /* Row with other data for bed. */
int fieldCount, /* Number of fields in final bed. */
boolean isPsl, /* True if in PSL format. */
boolean isGenePred, /* True if in GenePred format. */
boolean isBedWithBlocks, /* True if BED with block list. */
boolean *pslKnowIfProtein,/* Have we figured out if psl is protein? */
boolean *pslIsProtein, /* True if we know psl is protien. */
struct lm *lm) /* Local memory pool */
/* Create bed from a database row when we already understand
* the format pretty well. The bed is allocated inside of
* the local memory pool lm. Generally use this in conjunction
* with the results of a SQL query constructed with the aid
* of the bedSqlFieldsExceptForChrom function. */
{
char *strand, tStrand, qStrand;
struct bed *bed;
int i, blockCount;
lmAllocVar(lm, bed);
bed->chrom = chrom;
bed->chromStart = sqlUnsigned(row[0]);
bed->chromEnd = sqlUnsigned(row[1]);
if (fieldCount < 4)
return bed;
bed->name = lmCloneString(lm, row[2]);
if (fieldCount < 5)
return bed;
bed->score = atoi(row[3]);
if (fieldCount < 6)
return bed;
strand = row[4];
qStrand = strand[0];
tStrand = strand[1];
if (tStrand == 0)
bed->strand[0] = qStrand;
else
{
/* psl: use XOR of qStrand,tStrand if both are given. */
if (tStrand == qStrand)
bed->strand[0] = '+';
else
bed->strand[0] = '-';
}
if (fieldCount < 8)
return bed;
bed->thickStart = sqlUnsigned(row[5]);
bed->thickEnd = sqlUnsigned(row[6]);
if (fieldCount < 12)
return bed;
bed->blockCount = blockCount = sqlUnsigned(row[7]);
lmAllocArray(lm, bed->blockSizes, blockCount);
sqlUnsignedArray(row[8], bed->blockSizes, blockCount);
lmAllocArray(lm, bed->chromStarts, blockCount);
sqlUnsignedArray(row[9], bed->chromStarts, blockCount);
if (isGenePred)
{
/* Translate end coordinates to sizes. */
for (i=0; i<bed->blockCount; ++i)
bed->blockSizes[i] -= bed->chromStarts[i];
}
else if (isPsl)
{
if (!*pslKnowIfProtein)
{
/* Figure out if is protein using a rather elaborate but
* working test I think Angie or Brian must have figured out. */
if (tStrand == '-')
{
int tSize = sqlUnsigned(row[10]);
*pslIsProtein =
(bed->chromStart ==
tSize - (3*bed->blockSizes[bed->blockCount - 1] +
bed->chromStarts[bed->blockCount - 1]));
}
else
{
*pslIsProtein = (bed->chromEnd ==
3*bed->blockSizes[bed->blockCount - 1] +
bed->chromStarts[bed->blockCount - 1]);
}
*pslKnowIfProtein = TRUE;
}
if (*pslIsProtein)
{
/* if protein then blockSizes are in protein space */
for (i=0; i<blockCount; ++i)
bed->blockSizes[i] *= 3;
}
if (tStrand == '-')
{
/* psl: if target strand is '-', flip the coords.
* (this is the target part of pslRcBoth from src/lib/psl.c) */
int tSize = sqlUnsigned(row[10]);
for (i=0; i<blockCount; ++i)
{
bed->chromStarts[i] = tSize -
(bed->chromStarts[i] + bed->blockSizes[i]);
}
reverseInts(bed->chromStarts, bed->blockCount);
reverseInts(bed->blockSizes, bed->blockCount);
}
}
if (!isBedWithBlocks)
{
/* non-bed: translate absolute starts to relative starts */
for (i=0; i < bed->blockCount; i++)
bed->chromStarts[i] -= bed->chromStart;
}
return bed;
}
void refSeparateButJoined(struct txGraph *graph, FILE *f)
/* Flag graphs that have two non-overlapping refSeqs. */
{
int sourceIx;
boolean foundIt = FALSE;
struct lm *lm = lmInit(0);
struct rbTreeNode **stack;
lmAllocArray(lm, stack, 128);
/* Loop through sources looking for reference type. */
for (sourceIx=0; sourceIx<graph->sourceCount; ++sourceIx)
{
struct txSource *source = &graph->sources[sourceIx];
if (sameString(source->type, refType))
{
/* Create a rangeTree including all exons of source. */
struct rbTree *tree = rangeTreeNewDetailed(lm, stack);
struct txEdge *edge;
for (edge = graph->edgeList; edge != NULL; edge = edge->next)
{
if (edge->type == ggExon && evOfSourceOnList(edge->evList, sourceIx))
rangeTreeAdd(tree, graph->vertices[edge->startIx].position,
graph->vertices[edge->endIx].position);
}
/* Go through remaining reference sources looking for no overlap. */
int i;
for (i=0; i<graph->sourceCount; ++i)
{
if (i == sourceIx)
continue;
struct txSource *s = &graph->sources[i];
if (sameString(s->type, refType))
{
boolean gotOverlap = FALSE;
for (edge = graph->edgeList; edge != NULL; edge = edge->next)
{
if (edge->type == ggExon && evOfSourceOnList(edge->evList, i))
{
if (rangeTreeOverlaps(tree,
graph->vertices[edge->startIx].position,
graph->vertices[edge->endIx].position))
{
gotOverlap = TRUE;
break;
}
}
}
if (!gotOverlap)
{
foundIt = TRUE;
break;
}
}
}
freez(&tree);
}
if (foundIt)
break;
}
if (foundIt)
{
fprintf(f, "%s\t%d\t%d\t%s\t0\t%s\n", graph->tName,
graph->tStart, graph->tEnd, "refJoined", graph->strand);
}
lmCleanup(&lm);
}
