unsigned int FieldElement::totNumLocalField() const
{
unsigned int ret = 0;
unsigned int nd = numLocalData();
for ( unsigned int i = 0; i < nd; ++i )
ret += numField( i );
return ret;
}
void liftAgp(char *destFile, struct hash *liftHash, int sourceCount, char *sources[])
/* Lift up coordinates in .agp file. */
{
FILE *dest = mustOpen(destFile, "w");
char *source;
int i;
struct lineFile *lf;
int lineSize, wordCount;
char *line, *words[32];
char *s;
struct liftSpec *spec;
int start = 0;
int end = 0;
int ix = 0;
char newDir[256], newName[128], newExt[64];
struct bigInsert *bi;
struct chromInserts *chromInserts;
struct hash *insertHash = newHash(8);
struct hash *contigsHash = newHash(10);
boolean firstContig = TRUE;
char lastContig[256];
char *contig;
int lastEnd = 0;
if (sourceCount < 2)
usage();
if (how == carryMissing)
warn("'carry' doesn't work for .agp files, ignoring");
splitPath(destFile, newDir, newName, newExt);
/* Read in inserts file and process it. */
chromInsertsRead(sources[0], insertHash);
chromInserts = hashFindVal(insertHash, newName);
strcpy(lastContig, "");
for (i=1; i<sourceCount; ++i)
{
source = sources[i];
verbose(1, "Lifting %s\n", source);
lf = lineFileMayOpen(source, TRUE);
if (lf != NULL)
{
while (lineFileNext(lf, &line, &lineSize))
{
/* Check for comments and just pass them through. */
s = skipLeadingSpaces(line);
if (s[0] == '#')
{
fprintf(dest, "%s\n", line);
continue;
}
/* Parse line, adjust offsets, write */
wordCount = chopLine(line, words);
if (wordCount != 8 && wordCount != 9)
malformedAgp(lf);
contig = words[0];
if (!sameString(contig, lastContig))
{
char *gapType = "contig";
char *ctg = rmChromPrefix(contig);
int gapSize = chromInsertsGapSize(chromInserts,
ctg, firstContig);
if (hashLookup(contigsHash, contig))
errAbort("Contig repeated line %d of %s", lf->lineIx, lf->fileName);
hashAdd(contigsHash, contig, NULL);
if (gapSize != 0)
{
if ((bi = bigInsertBeforeContig(chromInserts, ctg)) != NULL)
{
gapType = bi->type;
}
fprintf(dest, "%s\t%d\t%d\t%d\tN\t%d\t%s\tno\n",
newName, end+1, end+gapSize, ++ix, gapSize, gapType);
}
firstContig = FALSE;
strcpy(lastContig, contig);
}
spec = findLift(liftHash, contig, lf);
cantHandleSpecRevStrand(spec);
start = numField(words, 1, 0, lf) + spec->offset;
end = numField(words, 2, 0, lf) + spec->offset;
if (end > lastEnd) lastEnd = end;
if (!sameString(newName, spec->newName))
errAbort("Mismatch in new name between %s and %s", newName, spec->newName);
fprintf(dest, "%s\t%d\t%d\t%d\t%s\t%s\t%s\t%s",
newName, start, end, ++ix,
words[4], words[5], words[6], words[7]);
if (wordCount == 9)
fprintf(dest, "\t%s", words[8]);
fputc('\n', dest);
}
lineFileClose(&lf);
if (dots)
verbose(1, "\n");
}
}
if (chromInserts != NULL)
{
if ((bi = chromInserts->terminal) != NULL)
{
fprintf(dest, "%s\t%d\t%d\t%d\tN\t%d\t%s\tno\n",
newName, lastEnd+1, lastEnd+bi->size, ++ix, bi->size, bi->type);
}
}
if (ferror(dest))
errAbort("error writing %s", destFile);
fclose(dest);
}
void liftOut(char *destFile, struct hash *liftHash, int sourceCount, char *sources[])
/* Lift up coordinates in .out file. Add offset to id (15th) column to
* maintain non-overlapping id ranges for different input files. */
{
FILE *dest = mustOpen(destFile, "w");
char *source;
int i;
struct lineFile *lf;
int lineSize, wordCount;
char *line, *words[32];
char *s;
int begin, end, left;
char leftString[18];
int highestIdSoFar = 0, idOffset = 0;
char idStr[32];
struct liftSpec *spec;
char *newName;
rmskOutWriteHead(dest);
for (i=0; i<sourceCount; ++i)
{
source = sources[i];
verbose(1, "Lifting %s\n", source);
if (!fileExists(source))
{
warn("%s does not exist\n", source);
continue;
}
lf = lineFileOpen(source, TRUE);
if (!lineFileNext(lf, &line, &lineSize))
{
warn("%s is empty\n", source);
lineFileClose(&lf);
continue;
}
if (startsWith("There were no", line))
{
lineFileClose(&lf);
continue;
}
skipLines(lf, 2);
while (lineFileNext(lf, &line, &lineSize))
{
wordCount = chopLine(line, words);
// 16 becomes 17, new field in RMasker June23 '03 - Hiram
if (wordCount < 14 || wordCount > 17)
errAbort("Expecting 14-17 words (found %d) line %d of %s", wordCount, lf->lineIx, lf->fileName);
if (wordCount >= 15)
{
if (words[14][0] == '*')
{
warn("Warning: 15th column has * (should be a numeric ID).\n");
idStr[0] = '\0';
}
else
{
int numId = sqlUnsigned(words[14]) + idOffset;
if (numId > highestIdSoFar)
highestIdSoFar = numId;
safef(idStr, sizeof(idStr), "%d", numId);
}
}
else
idStr[0] = '\0';
begin = numField(words, 5, 0, lf);
end = numField(words, 6, 0, lf);
s = words[7];
if (s[0] != '(')
errAbort("Expecting parenthesized field 8 line %d of %s", lf->lineIx, lf->fileName);
left = numField(words, 7, 1, lf);
spec = findLift(liftHash, words[4], lf);
if (spec == NULL)
{
if (how == carryMissing)
newName = words[4];
else
continue;
}
else
{
cantHandleSpecRevStrand(spec);
begin += spec->offset;
end += spec->offset;
left = spec->newSize - end;
newName = spec->newName;
}
sprintf(leftString, "(%d)", left);
fprintf(dest,
"%5s %5s %4s %4s %-9s %7d %7d %9s %1s %-14s %-19s %6s %4s %6s %6s\n",
words[0], words[1], words[2], words[3], newName,
begin, end, leftString,
words[8], words[9], words[10], words[11], words[12], words[13], idStr);
}
lineFileClose(&lf);
idOffset = highestIdSoFar;
}
if (ferror(dest))
errAbort("error writing %s", destFile);
fclose(dest);
}
void liftGap(char *destFile, struct hash *liftHash, int sourceCount, char *sources[])
/* Lift up coordinates in .gap file (just the gaps from .agp). Negative strad allowed */
{
FILE *dest = mustOpen(destFile, "w");
char *source;
int i;
struct lineFile *lf;
int lineSize, wordCount;
char *line, *words[32];
char *s;
struct liftSpec *spec;
int start = 0;
int end = 0;
int ix = 0;
char newDir[256], newName[128], newExt[64];
char lastContig[256];
char *contig;
int lastEnd = 0;
int fragStart, fragEnd;
if (how == carryMissing)
warn("'carry' doesn't work for .gap files, ignoring");
splitPath(destFile, newDir, newName, newExt);
strcpy(lastContig, "");
for (i=0; i<sourceCount; ++i)
{
source = sources[i];
verbose(1, "Lifting %s\n", source);
lf = lineFileMayOpen(source, TRUE);
if (lf != NULL)
{
while (lineFileNext(lf, &line, &lineSize))
{
/* Check for comments and just pass them through. */
s = skipLeadingSpaces(line);
if (s[0] == '#')
{
fprintf(dest, "%s\n", line);
continue;
}
/* Parse line, adjust offsets, write */
wordCount = chopLine(line, words);
if (wordCount != 8 && wordCount != 9)
malformedAgp(lf);
if (words[4][0] != 'N' && words[4][0] != 'U')
errAbort("Found non-gap in .gap file: %s", words[4]);
contig = words[0];
spec = findLift(liftHash, contig, lf);
start = fragStart = numField(words, 1, 0, lf);
end = fragEnd = numField(words, 2, 0, lf);
end = fragEnd;
if (spec->strand == '-')
{
start = spec->oldSize - fragEnd + 1;
end = spec->oldSize - fragStart + 1;
}
start += spec->offset;
end += spec->offset;
if (end > lastEnd) lastEnd = end;
fprintf(dest, "%s\t%d\t%d\t%d\t%s\t%s\t%s\t%s",
spec->newName, start, end, ++ix,
words[4], words[5], words[6], words[7]);
if (wordCount == 9)
fprintf(dest, "\t%s", words[8]);
fputc('\n', dest);
}
lineFileClose(&lf);
if (dots)
verbose(1, "\n");
}
}
if (ferror(dest))
errAbort("error writing %s", destFile);
fclose(dest);
}
