struct vcfRecord *vcfRecordFromRow(struct vcfFile *vcff, char **words)
/* Parse words from a VCF data line into a VCF record structure. */
{
struct vcfRecord *record = vcfFileAlloc(vcff, sizeof(struct vcfRecord));
record->file = vcff;
record->chrom = vcfFilePooledStr(vcff, words[0]);
record->chromStart = lineFileNeedNum(vcff->lf, words, 1) - 1;
// chromEnd may be overwritten by parseRefAndAlt and parseInfoColumn.
record->chromEnd = record->chromStart+1;
record->name = vcfFilePooledStr(vcff, words[2]);
parseRefAndAlt(vcff, record, words[3], words[4]);
record->qual = vcfFilePooledStr(vcff, words[5]);
parseFilterColumn(vcff, record, words[6]);
// ADDED BY BO PENG to get whole INFO column
record->unparsedInfoElements = vcfFilePooledStr(vcff, words[7]);
parseInfoColumn(vcff, record, words[7]);
if (vcff->genotypeCount > 0)
{
record->format = vcfFilePooledStr(vcff, words[8]);
record->genotypeUnparsedStrings = vcfFileAlloc(vcff,
vcff->genotypeCount * sizeof(char *));
int i;
// Don't bother actually parsing all these until & unless we need the info:
for (i = 0; i < vcff->genotypeCount; i++)
record->genotypeUnparsedStrings[i] = vcfFileCloneStr(vcff, words[9+i]);
}
return record;
}
static void vcfParseData(struct vcfFile *vcff, int maxRecords)
/* Given a vcfFile into which the header has been parsed, and whose lineFile is positioned
* at the beginning of a data row, parse and store all data rows from lineFile. */
{
if (vcff == NULL)
return;
int recCount = 0, expected = 8;
if (vcff->genotypeCount > 0)
expected = 9 + vcff->genotypeCount;
char *words[VCF_MAX_COLUMNS];
int wordCount;
while ((wordCount = lineFileChop(vcff->lf, words)) > 0)
{
if (maxRecords >= 0 && recCount >= maxRecords)
break;
lineFileExpectWords(vcff->lf, expected, wordCount);
struct vcfRecord *record;
AllocVar(record);
record->file = vcff;
record->chrom = vcfFilePooledStr(vcff, words[0]);
record->chromStart = lineFileNeedNum(vcff->lf, words, 1) - 1;
// chromEnd may be overwritten by parseRefAndAlt and parseInfoColumn.
record->chromEnd = record->chromStart+1;
record->name = vcfFilePooledStr(vcff, words[2]);
parseRefAndAlt(vcff, record, words[3], words[4]);
record->qual = vcfFilePooledStr(vcff, words[5]);
parseFilterColumn(vcff, record, words[6]);
parseInfoColumn(vcff, record, words[7]);
if (vcff->genotypeCount > 0)
{
record->format = vcfFilePooledStr(vcff, words[8]);
record->genotypeUnparsedStrings = vcfFileAlloc(vcff,
vcff->genotypeCount * sizeof(char *));
int i;
// Don't bother actually parsing all these until & unless we need the info:
for (i = 0; i < vcff->genotypeCount; i++)
record->genotypeUnparsedStrings[i] = vcfFileCloneStr(vcff, words[9+i]);
}
slAddHead(&(vcff->records), record);
recCount++;
}
slReverse(&(vcff->records));
lineFileClose(&(vcff->lf));
}
