int64_t writeAll(std::vector<the_t> &thetas, char *chr,BGZF *fp){
fprintf(stderr,"\tWriting: chr:%s with nSites:%zu\n",chr,thetas.size());
int64_t retVal =bgzf_tell(fp);
size_t clen=strlen(chr);
bgzf_write(fp,&clen,sizeof(size_t));//write len of chr
bgzf_write(fp,chr,clen);//write chr
size_t vLen = thetas.size();
bgzf_write(fp,&vLen,sizeof(size_t));//write len of positions;
int *posi = new int[thetas.size()];
static float **the = new float*[5];
for(int i=0;i<5;i++)
the[i] = new float[thetas.size()];
for(size_t i=0;i<thetas.size();i++){
posi[i] =thetas[i].posi;
for(int j=0;j<5;j++)
the[j][i] = thetas[i].vals[j];
delete [] thetas[i].vals;
}
bgzf_write(fp,posi,sizeof(int)*thetas.size());
for(int j=0;j<5;j++){
bgzf_write(fp,the[j],sizeof(float)*thetas.size());
delete [] the[j];
}
delete [] posi;
fprintf(stderr,"\tDone writing: %s\n",chr);
return retVal;
}
void abcSmartCounts::changeChr(int newRefId){
if(doSmartCounts==0)
return;
// fprintf(stderr,"cur:%d new:%d\n",curChr,newRefId);
if(curChr!=-1){
int64_t retVal =bgzf_tell(fbin);
int clen = strlen(header->name[curChr]);
bgzf_write(fbin,&clen,sizeof(int));
bgzf_write(fbin,header->name[curChr],clen);
bgzf_write(fbin,&len,sizeof(int));
for(int i=0;i<4;i++)
bgzf_write(fbin,counts[i],len);//write len of chr
//write index stuff
fprintf(stderr,"Writing index for chr: %s\n",header->name[curChr]);
fwrite(&clen,sizeof(int),1,fidx);
fwrite(header->name[curChr] ,sizeof(char),clen,fidx);
fwrite(&len,sizeof(int),1,fidx);
fwrite(&retVal,sizeof(int64_t),1,fidx);
}
curChr = newRefId;
len = header->l_ref[curChr];
for(int i=0;i<4;i++){
delete [] counts[i];
counts[i] = new unsigned char[len];
memset(counts[i],0,len);
}
}
void
populate_index(uint64_t *table, cmph_t *hash, BGZF *fastq_file)
{
while( 1 )
{
/* Find @ */
char c;
while( ( c = bgzf_getc( fastq_file ) ) != '@' && c >= 0 )
{
}
long pos = bgzf_tell( fastq_file );
if( pos == -1 )
{
break;
}
char *accession = NULL;
cmph_uint32 accession_length;
if( read_one_line( &accession, &accession_length, fastq_file ) != 1 )
{
break;
}
/* Next char is sequence, save pos */
unsigned int id = cmph_search( hash, accession, accession_length );
table[ id ] = (uint64_t) pos;
}
}
abcSmartCounts::~abcSmartCounts(){
if(doSmartCounts==0)
return;
int64_t retVal =bgzf_tell(fbin);
int clen = strlen(header->name[curChr]);
bgzf_write(fbin,&clen,sizeof(int));
bgzf_write(fbin,header->name[curChr],clen);
bgzf_write(fbin,&len,sizeof(int));
for(int i=0;i<4;i++)
bgzf_write(fbin,counts[i],len);//write len of chr
//write index stuff
fwrite(&clen,sizeof(int),1,fidx);
fwrite(header->name[curChr],sizeof(char),clen,fidx);
fwrite(&len,sizeof(int),1,fidx);
fwrite(&retVal,sizeof(int64_t),1,fidx);
for(int i=0;i<4;i++)
delete [] counts[i];
delete [] counts;
fclose(fidx);
bgzf_close(fbin);
}
BamFilePrivate(const std::string& fn)
: filename_(fn)
, firstAlignmentOffset_(-1)
{
// ensure we've updated htslib verbosity with requested verbosity here
hts_verbose = ( PacBio::BAM::HtslibVerbosity == -1 ? 0 : PacBio::BAM::HtslibVerbosity);
// attempt open
auto f = RawOpen();
#if !defined (PBBAM_NO_CHECK_EOF) || PBBAM_AUTOVALIDATE
// sanity check on file
const int eofCheck = bgzf_check_EOF(f->fp.bgzf);
if (eofCheck <= 0 ) {
// 1: EOF present & correct
// 2: not seekable (e.g. reading from stdin)
// 0: EOF absent
// -1: some other error
std::stringstream e;
if (eofCheck == 0)
e << fn << " : is missing EOF block" << std::endl;
else
e << fn << " : unknown error while checking EOF block" << std::endl;
throw std::runtime_error(e.str());
}
#endif
// attempt fetch header
std::unique_ptr<bam_hdr_t, internal::HtslibHeaderDeleter> hdr(sam_hdr_read(f.get()));
header_ = internal::BamHeaderMemory::FromRawData(hdr.get());
// cache first alignment offset
firstAlignmentOffset_ = bgzf_tell(f->fp.bgzf);
}
uint64_t read_chunk(double **chunk_data, params *pars, uint64_t chunk) {
uint64_t total_elems_read = 0;
if(chunk >= pars->n_chunks)
error("invalid chunk number!");
// Define chunk start and end positions
uint64_t start_pos = chunk * pars->max_chunk_size;
uint64_t end_pos = start_pos + pars->max_chunk_size - 1;
if(end_pos >= pars->n_sites) end_pos = pars->n_sites - 1;
uint64_t chunk_size = end_pos - start_pos + 1;
if( pars->verbose >= 6 ) printf("\tReading chunk %lu from position %lu to %lu (%lu)\n", chunk+1, start_pos, end_pos, chunk_size);
// Search start position
#ifdef _USE_BGZF
if( bgzf_seek(pars->in_glf_fh, pars->chunks_voffset[chunk], SEEK_SET) < 0 )
error("cannot seek GLF file (BGZF)!");
#endif
// Read data from file
for(uint64_t c = 0; c < chunk_size; c++) {
#ifdef _USE_BGZF
int bytes_read = bgzf_read(pars->in_glf_fh, chunk_data[c], (int) pars->n_ind * 3 * sizeof(double));
if(pars->call_geno)
call_geno(chunk_data[c], pars->n_ind, 3);
uint64_t elems_read = (uint64_t) bytes_read / sizeof(double);
#else
chunk_data[c] = pars->data[start_pos+c];
uint64_t elems_read = pars->n_ind * 3;
#endif
if( elems_read != pars->n_ind * 3 )
error("cannot read GLF file!");
total_elems_read += elems_read;
}
#ifdef _USE_BGZF
// Update index for next chunk
if( chunk+1 != pars->n_chunks && pars->chunks_voffset[chunk+1] == 0 )
pars->chunks_voffset[chunk+1] = bgzf_tell(pars->in_glf_fh);
#endif
return( total_elems_read/(pars->n_ind * 3) );
}
BgzfFileType::BgzfFileType(const char * filename, const char * mode)
{
// If the file is for write and is '-', then write to stdout.
if(((mode[0] == 'w') || (mode[0] == 'W')) &&
(strcmp(filename, "-") == 0))
{
// Write to stdout.
bgzfHandle = bgzf_fdopen(fileno(stdout), mode);
}
else if(((mode[0] == 'r') || (mode[0] == 'R')) &&
(strcmp(filename, "-") == 0))
{
// read from stdin
bgzfHandle = bgzf_fdopen(fileno(stdin), mode);
}
else
{
bgzfHandle = bgzf_open(filename, mode);
}
myStartPos = 0;
if (bgzfHandle != NULL)
{
// Check to see if the file is being opened for read, if the eof block
// is required, and if it is, if it is there.
if ((mode[0] == 'r' || mode[0] == 'R') && ourRequireEofBlock &&
(bgzf_check_EOF(bgzfHandle) == 0))
{
std::cerr << "BGZF EOF marker is missing in " << filename << std::endl;
// the block is supposed to be there, but isn't, so close the file.
close();
}
else
{
// Successfully opened a properly formatted file, so get the start
// position.
myStartPos = bgzf_tell(bgzfHandle);
}
}
myEOF = false;
}
//return zero if fine.
int writeDat(char *last,mmap &mm,tary<char> *keep,tary<char> *major,tary<char> *minor,BGZF *BFP,FILE *fp,int doCompl){
assert(last!=NULL);
if((major!=NULL) ^ (minor!=NULL)){
fprintf(stderr,"major and minor should be the same\n");
return 1;
}
int hasMajMin =0;
if(major!=NULL)
hasMajMin =1;
fprintf(stderr,"\t-> Writing chr:\'%s\' \n",last);
mmap::iterator it=mm.find(last);
if(it!=mm.end()){
return 1;
}else
mm[strdup(last)]=1;
//write data and index stuff
int64_t retVal =bgzf_tell(BFP);//now contains the offset to which we should point.
//write chrname
int clen=strlen(last)+1;
fwrite(&clen,1,sizeof(int),fp);
fwrite(last,clen,sizeof(char),fp);
fwrite(&retVal,1,sizeof(int64_t),fp);
for(int i=0;doCompl&&i<keep->l;i++)
if(keep->d[i]==0)
keep->d[i]=1;
else
keep->d[i]=0;
fwrite(&keep->l,sizeof(size_t),1,fp);//write len of chr
fwrite(&hasMajMin,1,sizeof(int),fp);
aio::bgzf_write(BFP,keep->d,keep->l);//write keep
if(hasMajMin){
aio::bgzf_write(BFP,major->d,major->l);//write maj
aio::bgzf_write(BFP,minor->d,minor->l);//write min
}
return 0;
}
int fst_index(int argc,char **argv){
if(argc<1){
fprintf(stderr,"Must supply afile.saf.idx [chrname, write more info]\n");
return 0;
}
args *arg = getArgs(argc,argv);
if(!arg->fstout){
fprintf(stderr,"\t-> Must supply -fstout for doing fstindex\n");
return 0;
}
std::vector<persaf *> &saf =arg->saf;
//assert(saf.size()==2);
size_t nSites = arg->nSites;
if(nSites == 0){//if no -nSites is specified
nSites = 100000;//<- set default to 100k sites, no need to load everything...
// nSites=nsites(saf,arg);
}
fprintf(stderr,"\t-> nSites: %lu\n",nSites);
std::vector<Matrix<float> *> gls;
for(int i=0;i<saf.size();i++)
gls.push_back(alloc<float>(nSites,saf[i]->nChr+1));
// int ndim= parspace(saf);
if(arg->sfsfname.size()!=choose(saf.size(),2)){
fprintf(stderr,"\t-> You have supplied: %lu populations, that is %d pairs\n",saf.size(),choose(saf.size(),2));
fprintf(stderr,"\t-> You therefore need to supply %d 2dsfs priors instead of:%lu\n",choose(saf.size(),2),arg->sfsfname.size());
exit(0);
}
std::vector<double *> sfs;
int inc =0;
for(int i=0;i<saf.size();i++)
for(int j=i+1;j<saf.size();j++){
size_t pairdim = (saf[i]->nChr+1)*(saf[j]->nChr+1);
double *ddd=new double[pairdim];
readSFS(arg->sfsfname[inc],pairdim,ddd);
normalize(ddd,pairdim);
sfs.push_back(ddd);
inc++;
}
double **a1,**b1;
a1=new double*[choose(saf.size(),2)];
b1=new double*[choose(saf.size(),2)];
inc=0;
for(int i=0;i<saf.size();i++)
for(int j=i+1;j<saf.size();j++){
calcCoef((int)saf[i]->nChr,(int)saf[j]->nChr,&a1[inc],&b1[inc]);
// fprintf(stderr,"a1[%d]:%p b1[%d]:%p\n",inc,&a1[inc][0],inc,&b1[inc][0]);
inc++;
}
BGZF *fstbg = openFileBG(arg->fstout,".fst.gz");
FILE *fstfp = openFile(arg->fstout,".fst.idx");
char buf[8]="fstv1";
bgzf_write(fstbg,buf,8);
fwrite(buf,1,8,fstfp);
#if 0
for(int i=0;i<ndim;i++)
fprintf(stdout,"%f %f\n",a1[i],b1[i]);
exit(0);
#endif
#if 1
size_t nsafs=saf.size();
fwrite(&nsafs,sizeof(size_t),1,fstfp);
for(int i=0;i<nsafs;i++){
size_t clen= strlen(saf[i]->fname);
fwrite(&clen,sizeof(size_t),1,fstfp);
fwrite(saf[i]->fname,1,clen,fstfp);
}
#endif
int asdf = choose(saf.size(),2);
std::vector<double> *ares = new std::vector<double> [choose(saf.size(),2)];
std::vector<double> *bres = new std::vector<double> [choose(saf.size(),2)];
// for(int i=0;i<3;i++)
// fprintf(stderr,"ares.size():%lu bres.size():%lu sfs:%p\n",ares[i].size(),bres[i].size(),&sfs[i][0]);
std::vector<int> posi;
setGloc(saf,nSites);
int *posiToPrint = new int[nSites];
for(myMap::iterator it = saf[0]->mm.begin();it!=saf[0]->mm.end();++it) {
// fprintf(stderr,"doing chr:%s\n",it->first);
if(arg->chooseChr!=NULL){
it = saf[0]->mm.find(arg->chooseChr);
if(it==saf[0]->mm.end()){
fprintf(stderr,"Problem finding chr: %s\n",arg->chooseChr);
break;
}
}
for(int i=0;i<choose(saf.size(),2);i++){
ares[i].clear();
bres[i].clear();
}
posi.clear();
while(1) {
int ret=readdata(saf,gls,nSites,it->first,arg->start,arg->stop,posiToPrint,NULL);//read nsites from data
// fprintf(stderr,"ret:%d glsx:%lu\n",ret,gls[0]->x);
//if(gls[0]->x!=nSites&&arg->chooseChr==NULL&&ret!=-3){
//fprintf(stderr,"continue continue\n");
// continue;
//}
fprintf(stderr,"\t-> Will now do fst temp dump using a chunk of %lu\n",gls[0]->x);
int inc=0;
for(int i=0;i<saf.size();i++)
for(int j=i+1;j<saf.size();j++){
// fprintf(stderr,"i:%d j:%d inc:%d gls[i]:%p gls[j]:%p sfs:%p a1:%p b1:%p\n",i,j,inc,gls[i],gls[j],sfs[i],&a1[inc][0],&a1[inc][0]);
block_coef(gls[i],gls[j],sfs[inc],a1[inc],b1[inc],ares[inc],bres[inc]);
inc++;
}
for(int i=0;i<gls[0]->x;i++)
posi.push_back(posiToPrint[i]);
for(int i=0;i<gls.size();i++)
gls[i]->x =0;
if(ret==-2)//no more data in files or in chr, eith way we break;
break;
}
size_t clen = strlen(it->first);
fwrite(&clen,sizeof(size_t),1,fstfp);
fwrite(it->first,1,clen,fstfp);
size_t nit=posi.size();
assert(1==fwrite(&nit,sizeof(size_t),1,fstfp));
int64_t tell = bgzf_tell(fstbg);
fwrite(&tell,sizeof(int64_t),1,fstfp);
bgzf_write(fstbg,&posi[0],posi.size()*sizeof(int));
int inc =0;
for(int i=0;i<saf.size();i++)
for(int j=i+1;j<saf.size();j++){
bgzf_write(fstbg,&(ares[inc][0]),ares[inc].size()*sizeof(double));
bgzf_write(fstbg,&(bres[inc][0]),bres[inc].size()*sizeof(double));
inc++;
}
if(arg->chooseChr!=NULL)
break;
}
delGloc(saf,nSites);
destroy(gls,nSites);
destroy_args(arg);
for(int i=0;i<sfs.size();i++)
delete [] sfs[i];
#if 0
fprintf(stderr,"\n\t-> NB NB output is no longer log probs of the frequency spectrum!\n");
fprintf(stderr,"\t-> Output is now simply the expected values! \n");
fprintf(stderr,"\t-> You can convert to the old format simply with log(norm(x))\n");
#endif
bgzf_close(fstbg);
fclose(fstfp);
fprintf(stderr,"\t-> fst index finished with no errors!\n");
return 0;
}
int main(int argc, char *argv[]) {
if (argc <= 1) {
fprintf(stderr, "Usage: thrash_threads4 input.bam\n");
exit(1);
}
// Find a valid seek location ~64M into the file
int i;
ssize_t got;
BGZF *fpin = bgzf_open(argv[1], "r");
uint64_t upos = 0, uend = 0;
char buf[100000];
for (i = 0; i < 100; i++) {
if ((got = bgzf_read(fpin, buf, 65536)) < 0)
abort();
upos += got;
}
int64_t pos = bgzf_tell(fpin);
while ((got = bgzf_read(fpin, buf, 65536)) > 0) {
uend += got;
}
if (got < 0) abort();
int64_t end = bgzf_tell(fpin);
bgzf_close(fpin);
// Ensure input is big enough to avoid case 3,4 below going off the end
// of the file
if (uend < upos + 10000000) {
fprintf(stderr, "Please supply a bigger input file\n");
exit(1);
}
#define N 1000
// Spam random seeks & reads
for (i = 0; i < 1000; i++) {
printf("i=%d\t", i);
fpin = bgzf_open(argv[1], "r");
int j, eof = 0, mt = 0;
for (j = 0; j < 80; j++) {
int n = rand() % 7;
putchar('0'+n); fflush(stdout);
switch (n) {
case 0: // start
if (bgzf_seek(fpin, 0LL, SEEK_SET) < 0) puts("!");//abort();
eof = 0;
break;
case 1: // mid
if (bgzf_seek(fpin, pos, SEEK_SET) < 0) puts("!");//abort();
eof = 0;
break;
case 2: // eof
if (bgzf_seek(fpin, end, SEEK_SET) < 0) puts("!");//abort();
eof = 1;
break;
case 3: case 4: {
int l = rand()%(n==3?100000:100);
if (bgzf_read(fpin, buf, l) != l*(1-eof)) abort();
break;
}
case 5:
usleep(N);
break;
case 6:
if (!mt)
bgzf_mt(fpin, 8, 256);
mt = 1;
break;
}
}
printf("\n");
if (bgzf_close(fpin))
abort();
}
return 0;
}
value caml_bgzf_tell(value bgzf) {
CAMLparam1(bgzf);
CAMLreturn(copy_int64(bgzf_tell(BGZF_val(bgzf))));
}
/**
* Create single chromosome index file
* the file content is a 2-column matrix of int64_t type
* line1: num_sample num_marker
* line2: 0 bgzf_offset_for_#CHROM_line
* line3: var_1_pos bgzf_offset_for_var_1
* ...
*/
int SingleChromosomeBCFIndex::createIndex() {
// const char* fn = bcfFile_.c_str();
BGZF* fp = fBcfFile_; // bgzf_open(fn, "rb");
bgzf_seek(fp, 0, SEEK_SET);
// check magic number
char magic[5];
if (5 != bgzf_read(fp, magic, 5)) {
return -1; // exit(1);
}
if (!(magic[0] == 'B' && magic[1] == 'C' && magic[2] == 'F' &&
magic[3] == 2 && (magic[4] == 1 || magic[4] == 2))) {
return -1; // exit(1);
}
// read header
uint32_t l_text;
if (4 != bgzf_read(fp, &l_text, 4)) {
return -1; // exit(1);
}
Rprintf("l_text = %d\n", l_text);
std::string s;
int64_t bgzf_offset_before_header = bgzf_tell(fp); // the beginning of header block
s.resize(l_text);
if (bgzf_read(fp, (void*)s.data(), l_text) != l_text) {
REprintf( "Read failed!\n");
}
BCFHeader bcfHeader;
if (bcfHeader.parseHeader(s,
&bcfHeader.header_contig_id,
&bcfHeader.header_id,
&bcfHeader.header_number,
&bcfHeader.header_type,
&bcfHeader.header_description)) {
REprintf( "Parse header failed!\n");
return -1; // exit(1);
}
// locate #CHROM line
int64_t bgzf_offset_after_header = bgzf_tell(fp); // the end of header block
size_t ptr_chrom_line = s.find("#CHROM"); // the index of "#CHROM", also the size between beginning of header to '#CHROM'
if (ptr_chrom_line == std::string::npos) {
REprintf( "Cannot find the \"#CHROM\" line!\n");
return -1; // exit(1);
}
Rprintf("offset_header = %d\n", (int) ptr_chrom_line);
bgzf_seek(fp, bgzf_offset_before_header, SEEK_SET); // rewind fp to the beginning of header
s.resize(ptr_chrom_line);
int64_t before_chrom_size = bgzf_read(fp, (void*) s.data(), ptr_chrom_line);
int64_t bgzf_offset_before_chrom = bgzf_tell(fp); // the offset to #CHROM
s.resize(l_text - before_chrom_size);
int64_t after_chrom_size = bgzf_read(fp, (void*) s.data(), l_text - before_chrom_size);
// load sample names
while (s.back() == '\n' || s.back() == '\0') {
s.resize(s.size() - 1);
}
stringTokenize(s, "\t", &bcfHeader.sample_names);
const int64_t num_sample = (int)bcfHeader.sample_names.size() - 9; // vcf header has 9 columns CHROM...FORMAT before actual sample names
Rprintf("sample size = %ld\n", num_sample);
Rprintf("last character is s[after_chrom_size-1] = %d\n", s[after_chrom_size - 1]); // should be 0, the null terminator character
// quality check
if (bgzf_offset_after_header != bgzf_tell(fp)) {
REprintf( "Messed up bgzf header\n");
return -1; // exit(1);
}
// create index file
FILE* fIndex = fopen(indexFile_.c_str(), "wb");
int64_t num_marker = 0;
int64_t pos = 0;
fwrite(&num_sample, sizeof(int64_t), 1, fIndex);
fwrite(&num_marker, sizeof(int64_t), 1, fIndex);
fwrite(&pos, sizeof(int64_t), 1, fIndex);
fwrite(&bgzf_offset_before_chrom, sizeof(int64_t), 1, fIndex);
uint32_t l_shared;
uint32_t l_indiv;
std::vector<char> data;
int64_t offset;
do {
offset = bgzf_tell(fp);
if (4 != bgzf_read(fp, &l_shared, sizeof(uint32_t))) {
break; // REprintf( "Wrong read!\n"); exit(1);
}
if (4 != bgzf_read(fp, &l_indiv, sizeof(uint32_t))) {
break; // REprintf( "Wrong read!\n"); exit(1);
}
data.resize(l_shared + l_indiv);
if (l_shared + l_indiv != bgzf_read(fp, data.data(), (l_shared+l_indiv) * sizeof(char))) {
break; // REprintf( "Wrong read!\n"); exit(1);
}
memcpy(&pos, data.data() + 4, 4);
fwrite(&pos, sizeof(int64_t), 1, fIndex);
fwrite(&offset, sizeof(int64_t), 1, fIndex);
num_marker++;
if (num_marker % 10000 == 0) {
Rprintf("\rprocessed %ld markers", num_marker);
}
} while (true);
if (fseek(fIndex, 0, SEEK_SET)) {
REprintf( "fseek failed\n!");
}
fwrite(&num_sample, sizeof(int64_t), 1, fIndex);
fwrite(&num_marker, sizeof(int64_t), 1, fIndex);
fclose(fIndex);
Rprintf("Indexing finished with %ld samples and %ld markers\n", num_sample, num_marker);
return 0;
}
