int vcfTabixBatchRead(struct vcfFile *vcff, char *chrom, int start, int end,
int maxErr, int maxRecords)
// Reads a batch of records from an opened and indexed VCF file, adding them to
// vcff->records and returning the count of new records added in this batch.
// Note: vcff->records will continue to be sorted, even if batches are loaded
// out of order. Additionally, resulting vcff->records will contain no duplicates
// so returned count refects only the new records added, as opposed to all records
// in range. If maxErr >= zero, then continue to parse until there are maxErr+1
// errors. A maxErr less than zero does not stop and reports all errors. Set
// maxErr to VCF_IGNORE_ERRS for silence.
{
int oldCount = slCount(vcff->records);
if (lineFileSetTabixRegion(vcff->lf, chrom, start, end))
{
struct vcfRecord *records = vcfParseData(vcff, maxRecords);
if (records)
{
struct vcfRecord *lastRec = vcff->records;
if (lastRec == NULL)
vcff->records = records;
else
{
// Considered just asserting the batches were in order, but a problem may
// result when non-overlapping location windows pick up the same long variant.
slSortMergeUniq(&(vcff->records), records, vcfRecordCmp, NULL);
}
}
}
return slCount(vcff->records) - oldCount;
}
struct vcfFile *vcfFileMayOpen(char *fileOrUrl, int maxErr, int maxRecords)
/* Parse a VCF file into a vcfFile object. If maxErr not zero, then
* continue to parse until this number of error have been reached. A maxErr
* less than zero does not stop and reports all errors. */
{
struct lineFile *lf = NULL;
if (startsWith("http://", fileOrUrl) || startsWith("ftp://", fileOrUrl) ||
startsWith("https://", fileOrUrl))
lf = netLineFileOpen(fileOrUrl);
else
lf = lineFileMayOpen(fileOrUrl, TRUE);
struct vcfFile *vcff = vcfFileHeaderFromLineFile(lf, maxErr);
vcfParseData(vcff, maxRecords);
return vcff;
}
struct vcfFile *vcfTabixFileMayOpen(char *fileOrUrl, char *chrom, int start, int end,
int maxErr, int maxRecords)
/* Parse header and rows within the given position range from a VCF file that has been
* compressed and indexed by tabix into a vcfFile object; return NULL if or if file has
* no items in range.
* If maxErr not zero, then continue to parse until this number of error have been reached.
* A maxErr less than zero does not stop and reports all errors. */
{
struct lineFile *lf = lineFileTabixMayOpen(fileOrUrl, TRUE);
struct vcfFile *vcff = vcfFileHeaderFromLineFile(lf, maxErr);
if (vcff == NULL)
return NULL;
if (isNotEmpty(chrom) && start != end)
{
if (lineFileSetTabixRegion(lf, chrom, start, end))
vcfParseData(vcff, maxRecords);
}
return vcff;
}
struct vcfFile *vcfFileMayOpen(char *fileOrUrl, int maxErr, int maxRecords, boolean parseAll)
/* Open fileOrUrl and parse VCF header; return NULL if unable.
* If parseAll, then read in all lines, parse and store in
* vcff->records; if maxErr >= zero, then continue to parse until
* there are maxErr+1 errors. A maxErr less than zero does not stop
* and reports all errors. Set maxErr to VCF_IGNORE_ERRS for silence */
{
struct lineFile *lf = NULL;
if (startsWith("http://", fileOrUrl) || startsWith("ftp://", fileOrUrl) ||
startsWith("https://", fileOrUrl))
lf = netLineFileOpen(fileOrUrl);
else
lf = lineFileMayOpen(fileOrUrl, TRUE);
struct vcfFile *vcff = vcfFileHeaderFromLineFile(lf, maxErr);
if (parseAll)
{
vcff->records = vcfParseData(vcff, maxRecords);
lineFileClose(&(vcff->lf)); // Not sure why it is closed. Angie?
}
return vcff;
}
struct vcfFile *vcfTabixFileMayOpen(char *fileOrUrl, char *chrom, int start, int end,
int maxErr, int maxRecords)
/* Open a VCF file that has been compressed and indexed by tabix and
* parse VCF header, or return NULL if unable. If chrom is non-NULL,
* seek to the position range and parse all lines in range into
* vcff->records. If maxErr >= zero, then continue to parse until
* there are maxErr+1 errors. A maxErr less than zero does not stop
* and reports all errors. Set maxErr to VCF_IGNORE_ERRS for silence */
{
struct lineFile *lf = lineFileTabixMayOpen(fileOrUrl, TRUE);
struct vcfFile *vcff = vcfFileHeaderFromLineFile(lf, maxErr);
if (vcff == NULL)
return NULL;
if (isNotEmpty(chrom) && start != end)
{
if (lineFileSetTabixRegion(lf, chrom, start, end))
{
vcff->records = vcfParseData(vcff, maxRecords);
lineFileClose(&(vcff->lf)); // Not sure why it is closed. Angie?
}
}
return vcff;
}
