void getChromSizes(char *database, struct hash **retHash,
struct chromSizes **retList)
/* Return hash of chromSizes. Also calculates size without
* gaps. */
{
struct sqlConnection *conn = hAllocConn(database);
struct chromInfo *ci, *ciList = getAllChromInfo(database);
struct sqlResult *sr;
char **row;
struct chromSizes *cs, *csList = NULL;
struct hash *hash = newHash(8);
int rowOffset;
for (ci = ciList; ci != NULL; ci = ci->next)
{
AllocVar(cs);
hashAddSaveName(hash, ci->chrom, cs, &cs->name);
slAddHead(&csList, cs);
cs->totalSize = ci->size;
sr = hChromQuery(conn, "gold", ci->chrom, NULL, &rowOffset);
while ((row = sqlNextRow(sr)) != NULL)
{
struct agpFrag frag;
agpFragStaticLoad(row + rowOffset, &frag);
cs->seqSize += frag.chromEnd - frag.chromStart;
}
sqlFreeResult(&sr);
}
hFreeConn(&conn);
slReverse(&csList);
*retHash = hash;
*retList = csList;
}
void readPatch(char *fileName, struct hash *cloneHash,
struct ntContig **retNtList, struct hash **retNtHash)
/* Read nt.agp file into clone/hash. */
{
struct ntContig *ntList = NULL, *nt = NULL;
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[9];
struct agpFrag frag;
struct clone *clone, *ntClone, *lastClone = NULL;
struct cloneRef *ref;
struct hash *ntHash = newHash(0);
char cloneName[128];
char fragName[128];
char c;
int ntOrder = 0;
while (lineFileRow(lf, row))
{
agpFragStaticLoad(row, &frag);
// file is 1-based but agpFragLoad() now assumes 0-based:
frag.chromStart -= 1;
frag.fragStart -= 1;
if (nt == NULL || !sameString(frag.chrom, nt->name))
{
AllocVar(nt);
slAddHead(&ntList, nt);
if (hashLookup(ntHash, frag.chrom) != NULL)
errAbort("NT contig %s repeated line %d of %s", row[0], lf->lineIx, lf->fileName);
hashAddSaveName(ntHash, frag.chrom, nt, &nt->name);
lastClone = NULL;
ntOrder = 0;
}
strcpy(cloneName, frag.frag);
chopSuffix(cloneName);
clone = hashMustFindVal(cloneHash, cloneName);
clone->ntStart = frag.chromStart;
clone->ntEnd = frag.chromEnd;
if (clone->nt != NULL)
{
warn("Clone %s trying to be in two NT contigs (%s and %s) line %d of %s",
clone->name, clone->nt->name, nt->name, lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
clone->nt = nt;
c = frag.strand[0];
if (c == '-')
clone->ntOrientation = -1;
else if (c == '+')
clone->ntOrientation = +1;
else
errAbort("Expecting +1 or -1 field 5, line %d, file %s", lf->lineIx, lf->fileName);
c = frag.type[0];
if (c == 'F' || c == 'D' || c == 'P')
clone->seqType = c;
else
errAbort("Expecting F, D, or P field 6, line %d, file %s", lf->lineIx, lf->fileName);
sprintf(fragName, "%s_1", frag.frag);
clone->fragName = cloneString(fragName);
clone->goldStart = frag.fragStart;
clone->goldEnd = frag.fragEnd;
clone->ntOrder = ntOrder++;
/* Add ref to NT. */
AllocVar(ref);
ref->ref = clone;
slAddTail(&nt->cloneList, ref);
/* Do a few tests. */
if (clone->goldStart >= clone->goldEnd)
{
warn("Clone %s end before start (%d before %d) line %d of %s",
clone->name, clone->goldStart, clone->goldEnd, lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
if (clone->ntStart >= clone->ntEnd)
{
warn("Clone %s NT end before NT start line %d of %s",
clone->name, lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
if (clone->goldEnd > clone->size)
{
if (sameString(clone->startFrag, clone->endFrag))
{
warn("Clone %s end position %d, clone size %d, line %d of %s",
clone->name, clone->goldEnd, clone->size, lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
}
if (clone->ntEnd - clone->ntStart != clone->goldEnd - clone->goldStart)
{
warn("Size not the same in NT contig as in clone %s (%d vs %d) line %d of %s",
clone->name,
clone->ntEnd - clone->ntStart, clone->goldEnd-clone->goldStart,
lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
nt->sumSize += clone->goldEnd - clone->goldStart;
ntClone = hashFindVal(cloneHash, nt->name);
if (ntClone != NULL && clone->ntEnd > ntClone->size)
{
warn("Clone %s NT end position %d, NT size %d, line %d of %s",
clone->name, clone->ntEnd, ntClone->size, lf->lineIx, lf->fileName);
nt->problem = TRUE;
}
if (ntClone != NULL)
nt->size = ntClone->size;
else
nt->size = clone->size; /* This happens for single-clone NT contigs only. */
if (lastClone != NULL)
{
if (lastClone->ntEnd != clone->ntStart)
{
warn("last clone (%s)'s end doesn't match with current clone (%s)'s start line %d of %s",
lastClone->name, clone->name, lf->lineIx, lf->fileName);
}
}
lastClone = clone;
}
lineFileClose(&lf);
slReverse(&ntList);
for (nt = ntList; nt != NULL; nt = nt->next)
{
if (nt->sumSize != nt->size)
{
warn("Sum of fragments of %s is %d, but size is supposed to be %d",
nt->name, nt->sumSize, nt->size);
nt->problem = TRUE;
}
}
*retNtList = ntList;
*retNtHash = ntHash;
}
void splitAgp(char *agpName, char *goldFileName, char *gapFileName)
/* Split up agp file into gold and gap files. */
{
struct lineFile *lf;
char *words[16];
int wordCount;
FILE *goldTab, *gapTab;
/* Scan through .agp file splitting it into gold
* and gap components. */
goldTab = mustOpen(goldFileName, "w");
gapTab = mustOpen(gapFileName, "w");
lf = lineFileOpen(agpName, TRUE);
while ((wordCount = lineFileChop(lf, words)) > 0)
{
int start, end;
if (wordCount < 5)
errAbort("Short line %d of %s", lf->lineIx, lf->fileName);
int len = strlen(words[0]);
if (len > maxChromNameSize)
{
maxChromNameSize = len;
if (maxChromNameSize > 254)
errAbort("ERROR: chrom name size is over 254(%d) characters: "
"'%s'", maxChromNameSize, words[0]);
}
start = sqlUnsigned(words[1])-1;
end = sqlUnsigned(words[2]);
if (words[4][0] == 'N' || words[4][0] == 'U')
{
struct agpGap gap;
agpGapStaticLoad(words, &gap);
gap.chromStart -= 1;
fprintf(gapTab, "%u\t", hFindBin(start, end));
agpGapTabOut(&gap, gapTab);
verbose(3,"#GAP\t%s:%d-%d\n", gap.chrom, gap.chromStart, gap.chromEnd);
}
else
{
struct agpFrag gold;
agpFragStaticLoad(words, &gold);
agpFragValidate(&gold);
len = strlen(words[5]);
if (len > maxFragNameSize)
{
maxFragNameSize = len;
if (maxFragNameSize > 254)
errAbort("ERROR: fragment name size is over 254(%d) "
"characters: '%s'", maxFragNameSize, words[5]);
}
// file is 1-based. agpFragLoad() now assumes 0-based.
// and agpFragTabOut() will assume 1-based, but we will load
// the generated file straight into the database, so
// subtract 2:
gold.chromStart -= 2;
gold.fragStart -= 2;
fprintf(goldTab, "%u\t", hFindBin(start, end));
agpFragTabOut(&gold, goldTab);
}
}
lineFileClose(&lf);
carefulClose(&goldTab);
carefulClose(&gapTab);
}
