int main() {
VCFInputFile vin("noindex.vcf.gz");
vin.setRangeList("1:0");
int lineNo = 0;
while (vin.readRecord()) {
lineNo ++;
VCFRecord& r = vin.getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
printf("%s:%d\t", r.getChrom(), r.getPos());
// e.g.: get TAG from INFO field
// fprintf(stderr, "%s\n", r.getInfoTag("ANNO"));
// e.g.: Loop each (selected) people in the same order as in the VCF
for (int i = 0; i < people.size(); i++) {
indv = people[i];
// get GT index. if you are sure the index will not change, call this function only once!
int GTidx = r.getFormatIndex("GT");
if (GTidx >= 0)
printf("%s ", indv->justGet(0).toStr()); // [0] meaning the first field of each individual
else
fprintf(stderr, "Cannot find GT field!\n");
}
printf("\n");
};
fprintf(stdout, "Total %d VCF records have converted successfully\n", lineNo);
};
int loadMarkerFromVCF(const std::string& fileName, const std::string& marker,
std::vector<std::string>* rowLabel, Matrix* genotype) {
if (!rowLabel || !genotype) {
// invalid parameter
return -1;
}
Matrix& m = *genotype;
int col = 0;
VCFInputFile vin(fileName);
vin.setRangeList(marker);
while (vin.readRecord()) {
VCFRecord& r = vin.getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
m.Dimension(people.size(), col + 1);
int GTidx = r.getFormatIndex("GT");
for (int i = 0; i < (int)people.size(); i++) {
indv = people[i];
// get GT index. if you are sure the index will not change,
// call this function only once!
if (GTidx >= 0) {
// printf("%s ", indv->justGet(0).toStr()); // [0] meaning the first
// field of each individual
m[i][col] = indv->justGet(GTidx).getGenotype();
} else {
logger->error("Cannot find GT field!");
return -1;
}
}
if (col == 0) {
// set-up names
rowLabel->resize(people.size());
for (size_t i = 0; i < people.size(); ++i) {
(*rowLabel)[i] = people[i]->getName();
}
}
std::string colLabel = r.getChrom();
colLabel += ":";
colLabel += r.getPosStr();
m.SetColumnLabel(col, colLabel.c_str());
++col;
}
return 0;
}
int loadGenotype(VCFInputFile& vin, AllConcordanceType* input, int idx) {
AllConcordanceType& data = *input;
std::string key;
int lineNo = 0;
while (vin.readRecord()) {
lineNo++;
key.clear();
VCFRecord& r = vin.getVCFRecord();
key += r.getChrom();
key += ":";
key += r.getPosStr();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
int GTidx = r.getFormatIndex("GT");
if (GTidx < 0) continue;
for (size_t i = 0; i < people.size(); ++i) {
indv = people[i];
const VCFValue& v = indv->justGet(GTidx);
int a1 = v.getAllele1();
int a2 = v.getAllele2();
if (a1 == MISSING_GENOTYPE || a2 == MISSING_GENOTYPE) {
data[indv->getName()][key][idx] = Value::MISSING;
} else if (a1 == 0) {
if (a2 == 0) {
data[indv->getName()][key][idx] = Value::HOMREF;
} else if (a2 == 1) {
data[indv->getName()][key][idx] = Value::HET;
}
} else if (a1 == 1) {
if (a2 == 0) {
data[indv->getName()][key][idx] = Value::HET;
} else if (a2 == 1) {
data[indv->getName()][key][idx] = Value::HOMALT;
}
}
}
};
fprintf(stderr, "Total %d VCF records have read successfully\n", lineNo);
return lineNo;
};
int PlinkOutputFile::writeRecord(VCFRecord* r) {
// write BIM
if (isMultiAllelic(r->getRef()) || isMultiAllelic(r->getAlt())) {
fprintf(stdout, "%s:%d Skip with ref = [ %s ] and alt= [ %s ]\n", __FILE__,
__LINE__, r->getRef(), r->getAlt());
return -1;
}
this->writeBIM(r->getChrom(), r->getID(), 0, r->getPos(), r->getRef(),
r->getAlt());
// write BED
int GTidx = r->getFormatIndex("GT");
VCFPeople& people = r->getPeople();
unsigned char c = 0;
VCFIndividual* indv;
int offset;
for (unsigned int i = 0; i < people.size(); i++) {
indv = people[i];
offset = i & (4 - 1);
if (indv->justGet(GTidx).isHaploid()) { // 0: index of GT
int a1 = indv->justGet(GTidx).getAllele1();
if (a1 == 0)
setGenotype(&c, offset, HOM_REF);
else if (a1 == 1)
setGenotype(&c, offset, HET);
else
setGenotype(&c, offset, MISSING);
} else {
int a1 = indv->justGet(GTidx).getAllele1();
int a2 = indv->justGet(GTidx).getAllele2();
if (a1 == 0) {
if (a2 == 0) {
// h**o ref: 0b00
} else if (a2 == 1) {
setGenotype(&c, offset, HET); // het: 0b01
} else {
setGenotype(&c, offset, MISSING); // missing 0b10
}
} else if (a1 == 1) {
if (a2 == 0) {
setGenotype(&c, offset, HET); // het: 0b01
} else if (a2 == 1) {
setGenotype(&c, offset, HOM_ALT); // hom alt: 0b11
} else {
setGenotype(&c, offset, MISSING); // missing
}
} else {
// NOTE: Plink does not support tri-allelic
// so have to set genotype as missing.
setGenotype(&c, offset, MISSING); // missing
}
}
if (offset == 3) { // 3: 4 - 1, so every 4 genotype we will flush
fwrite(&c, sizeof(char), 1, this->fpBed);
c = 0;
}
}
if (people.size() % 4 != 0) { // remaining some bits
fwrite(&c, sizeof(char), 1, this->fpBed);
}
return 0;
}
int GenotypeExtractor::extractMultipleGenotype(Matrix* g) {
static Matrix m; // make it static to reduce memory allocation
int row = 0;
std::vector<std::string> colNames;
std::string name;
this->hemiRegion.clear();
GenotypeCounter genoCounter;
while (this->vin->readRecord()) {
VCFRecord& r = this->vin->getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
m.Dimension(row + 1, people.size());
genoCounter.reset();
int genoIdx;
const bool useDosage = (!this->dosageTag.empty());
if (useDosage) {
genoIdx = r.getFormatIndex(dosageTag.c_str());
} else {
genoIdx = r.getFormatIndex("GT");
}
int GDidx = r.getFormatIndex("GD");
int GQidx = r.getFormatIndex("GQ");
assert(this->parRegion);
bool hemiRegion = this->parRegion->isHemiRegion(r.getChrom(), r.getPos());
// e.g.: Loop each (selected) people in the same order as in the VCF
const int numPeople = (int)people.size();
for (int i = 0; i < numPeople; i++) {
indv = people[i];
// get GT index. if you are sure the index will not change, call this
// function only once!
if (genoIdx >= 0) {
// printf("%s ", indv->justGet(0).toStr()); // [0] meaning the first
// field of each individual
if (useDosage) {
if (!hemiRegion) {
m[row][i] = indv->justGet(genoIdx).toDouble();
} else {
// for male hemi region, imputated dosage is usually between 0 and 1
// need to multiply by 2.0
if ((*sex)[i] == PLINK_MALE) {
m[row][i] = indv->justGet(genoIdx).toDouble() * 2.0;
}
}
} else {
if (!hemiRegion) {
m[row][i] = indv->justGet(genoIdx).getGenotype();
} else {
if ((*sex)[i] == PLINK_MALE) {
m[row][i] = indv->justGet(genoIdx).getMaleNonParGenotype02();
} else if ((*sex)[i] == PLINK_FEMALE) {
m[row][i] = indv->justGet(genoIdx).getGenotype();
} else {
m[row][i] = MISSING_GENOTYPE;
}
}
}
if (!checkGD(indv, GDidx) || !checkGQ(indv, GQidx)) {
m[row][i] = MISSING_GENOTYPE;
}
genoCounter.add(m[row][i]);
} else {
logger->error("Cannot find %s field!",
this->dosageTag.empty() ? "GT" : dosageTag.c_str());
return -1;
}
}
// check frequency cutoffs
// int numNonMissingPeople = 0;
// double maf = 0.;
// for (int i = 0; i < numPeople; ++i) {
// if (m[row][i] < 0) continue;
// maf += m[row][i];
// ++numNonMissingPeople;
// }
// if (numNonMissingPeople) {
// maf = maf / (2. * numNonMissingPeople);
// } else {
// maf = 0.0;
// }
// if (maf > .5) {
// maf = 1.0 - maf;
// }
const double maf = genoCounter.getMAF();
if (this->freqMin > 0. && this->freqMin > maf) continue;
if (this->freqMax > 0. && this->freqMax < maf) continue;
// store genotype results
name = r.getChrom();
name += ":";
name += r.getPosStr();
colNames.push_back(name);
++row;
assert(this->parRegion);
if (this->parRegion &&
this->parRegion->isHemiRegion(r.getChrom(), r.getPos())) {
this->hemiRegion.push_back(true);
} else {
this->hemiRegion.push_back(false);
}
this->counter.push_back(genoCounter);
} // end while (this->vin->readRecord())
// delete rows (ugly code here, as we may allocate extra row in previous
// loop)
m.Dimension(row, m.cols);
// now transpose (marker by people -> people by marker)
g->Transpose(m);
for (int i = 0; i < row; ++i) {
g->SetColumnLabel(i, colNames[i].c_str());
}
return SUCCEED;
} // end GenotypeExtractor
int GenotypeExtractor::extractSingleGenotype(Matrix* g, Result* b) {
Matrix& genotype = *g;
Result& buf = *b;
bool hasRead = this->vin->readRecord();
if (!hasRead) return FILE_END;
VCFRecord& r = this->vin->getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
buf.updateValue("CHROM", r.getChrom());
buf.updateValue("POS", r.getPosStr());
buf.updateValue("REF", r.getRef());
buf.updateValue("ALT", r.getAlt());
genotype.Dimension(people.size(), 1);
counter.resize(1);
// get GT index. if you are sure the index will not change, call this
// function only once!
const bool useDosage = (!this->dosageTag.empty());
int genoIdx;
if (useDosage) {
genoIdx = r.getFormatIndex(dosageTag.c_str());
} else {
genoIdx = r.getFormatIndex("GT");
}
// int GTidx = r.getFormatIndex("GT");
int GDidx = r.getFormatIndex("GD");
int GQidx = r.getFormatIndex("GQ");
bool hemiRegion = this->parRegion->isHemiRegion(r.getChrom(), r.getPos());
// e.g.: Loop each (selected) people in the same order as in the VCF
const int numPeople = (int)people.size();
for (int i = 0; i < numPeople; i++) {
indv = people[i];
if (genoIdx >= 0) {
// printf("%s ", indv->justGet(0).toStr()); // [0] meaning the first
// field of each individual
if (useDosage) {
genotype[i][0] = indv->justGet(genoIdx).toDouble();
} else {
if (!hemiRegion) {
genotype[i][0] = indv->justGet(genoIdx).getGenotype();
} else {
if ((*sex)[i] == PLINK_MALE) {
genotype[i][0] = indv->justGet(genoIdx).getMaleNonParGenotype02();
} else if ((*sex)[i] == PLINK_FEMALE) {
genotype[i][0] = indv->justGet(genoIdx).getGenotype();
} else {
genotype[i][0] = MISSING_GENOTYPE;
}
}
}
if (!checkGD(indv, GDidx) || !checkGQ(indv, GQidx)) {
genotype[i][0] = MISSING_GENOTYPE;
}
counter[0].add(genotype[i][0]);
// logger->info("%d ", int(genotype[i][0]));
} else {
std::string s;
indv->toStr(&s);
logger->error(
"Cannot find [ %s ] field when read individual information [ %s ]!",
this->dosageTag.empty() ? "GT" : this->dosageTag.c_str(), s.c_str());
return ERROR;
}
}
// check frequency cutoffs
// double maf = 0.;
// if (this->freqMin > 0.0 || this->freqMax > 0.) {
// for (int i = 0; i < numPeople; ++i) {
// maf += genotype[i][0];
// }
// maf = maf / (2. * numPeople);
// if (maf > .5) {
// maf = 1.0 - maf;
// }
// }
const double maf = counter[0].getMAF();
if (this->freqMin > 0. && this->freqMin > maf) return FAIL_FILTER;
if (this->freqMax > 0. && this->freqMax < maf) return FAIL_FILTER;
std::string label = r.getChrom();
label += ':';
label += r.getPosStr();
genotype.SetColumnLabel(0, label.c_str());
this->hemiRegion.resize(1);
assert(this->parRegion);
if (this->parRegion &&
this->parRegion->isHemiRegion(r.getChrom(), r.getPos())) {
this->hemiRegion[0] = true;
} else {
this->hemiRegion[0] = false;
}
return SUCCEED;
} // end extractSingleGenotype()
int main(int argc, char** argv) {
time_t currentTime = time(0);
fprintf(stderr, "Analysis started at: %s", ctime(¤tTime));
PARSE_PARAMETER(argc, argv);
PARAMETER_STATUS();
if (FLAG_REMAIN_ARG.size() > 0) {
fprintf(stderr, "Unparsed arguments: ");
for (unsigned int i = 0; i < FLAG_REMAIN_ARG.size(); i++) {
fprintf(stderr, " %s", FLAG_REMAIN_ARG[i].c_str());
}
fprintf(stderr, "\n");
abort();
}
REQUIRE_STRING_PARAMETER(FLAG_inVcf,
"Please provide input file using: --inVcf");
const char* fn = FLAG_inVcf.c_str();
VCFInputFile vin(fn);
// set range filters here
// e.g.
// vin.setRangeList("1:69500-69600");
vin.setRangeList(FLAG_rangeList.c_str());
vin.setRangeFile(FLAG_rangeFile.c_str());
// set people filters here
if (FLAG_peopleIncludeID.size() || FLAG_peopleIncludeFile.size()) {
vin.excludeAllPeople();
vin.includePeople(FLAG_peopleIncludeID.c_str());
vin.includePeopleFromFile(FLAG_peopleIncludeFile.c_str());
}
vin.excludePeople(FLAG_peopleExcludeID.c_str());
vin.excludePeopleFromFile(FLAG_peopleExcludeFile.c_str());
// let's write it out.
if (FLAG_updateId != "") {
int ret = vin.updateId(FLAG_updateId.c_str());
fprintf(stdout, "%d samples have updated id.\n", ret);
}
// load gene ranges
std::map<std::string, std::string> geneRange;
if (FLAG_geneName.size()) {
if (FLAG_geneFile.size() == 0) {
fprintf(stderr, "Have to provide --geneFile to extract by gene.\n");
abort();
}
LineReader lr(FLAG_geneFile);
std::vector<std::string> fd;
while (lr.readLineBySep(&fd, "\t ")) {
if (FLAG_geneName != fd[0]) continue;
fd[2] = chopChr(fd[2]); // chop "chr1" to "1"
if (geneRange.find(fd[0]) == geneRange.end()) {
geneRange[fd[0]] = fd[2] + ":" + fd[4] + "-" + fd[5];
} else {
geneRange[fd[0]] += "," + fd[2] + ":" + fd[4] + "-" + fd[5];
}
};
}
std::string range;
for (std::map<std::string, std::string>::iterator it = geneRange.begin();
it != geneRange.end(); it++) {
if (range.size() > 0) {
range += ",";
}
range += it->second;
};
fprintf(stderr, "range = %s\n", range.c_str());
vin.setRangeList(range.c_str());
Regex regex;
if (FLAG_annoType.size()) {
regex.readPattern(FLAG_annoType);
}
// print header
std::vector<std::string> names;
vin.getVCFHeader()->getPeopleName(&names);
printf("CHROM\tPOS");
for (unsigned int i = 0; i < names.size(); i++) {
printf("\t%s", names[i].c_str());
}
printf("\n");
// real working part
int nonVariantSite = 0;
while (vin.readRecord()) {
VCFRecord& r = vin.getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
if (FLAG_variantOnly) {
bool hasVariant = false;
int geno;
int GTidx = r.getFormatIndex("GT");
for (size_t i = 0; i < people.size(); i++) {
indv = people[i];
geno = indv->justGet(GTidx).getGenotype();
if (geno != 0 && geno != MISSING_GENOTYPE) hasVariant = true;
}
if (!hasVariant) {
nonVariantSite++;
continue;
}
}
if (FLAG_annoType.size()) {
bool isMissing = false;
const char* tag = r.getInfoTag("ANNO", &isMissing).toStr();
if (isMissing) continue;
// fprintf(stdout, "ANNO = %s", tag);
bool match = regex.match(tag);
// fprintf(stdout, " %s \t", match ? "match": "noMatch");
// fprintf(stdout, " %s \n", exists ? "exists": "missing");
if (!match) {
continue;
}
}
fprintf(stdout, "%s\t%s", r.getChrom(), r.getPosStr());
for (size_t i = 0; i < people.size(); i++) {
indv = people[i];
fprintf(stdout, "\t%d", indv->justGet(0).getGenotype());
}
fprintf(stdout, "\n");
};
currentTime = time(0);
fprintf(stderr, "Analysis ends at: %s", ctime(¤tTime));
return 0;
};
