#include<cstring>
#include<cassert>
#include<string>
-#include<map>
#include<sstream>
+#include<iostream>
+#include <stdint.h>
#include "sam/bam.h"
#include "sam/sam.h"
+#include "sam_rsem_aux.h"
+#include "sam_rsem_cvt.h"
#include "utils.h"
+
#include "SingleHit.h"
#include "PairedEndHit.h"
class BamWriter {
public:
- BamWriter(char, const char*, const char*, const char*, const char*);
+ BamWriter(char, const char*, const char*, const char*, Transcripts&);
~BamWriter();
- void work(HitWrapper<SingleHit>, Transcripts&);
- void work(HitWrapper<PairedEndHit>, Transcripts&);
+ void work(HitWrapper<SingleHit>);
+ void work(HitWrapper<PairedEndHit>);
private:
samfile_t *in, *out;
+ Transcripts& transcripts;
- std::map<std::string, int> refmap;
- std::map<std::string, int>::iterator iter;
-
- struct SingleEndT {
- bam1_t *b;
-
- SingleEndT(bam1_t *b = NULL) {
- this->b = b;
- }
-
- bool operator< (const SingleEndT& o) const {
- int strand1, strand2;
- uint32_t *p1, *p2;
-
- if (b->core.tid != o.b->core.tid) return b->core.tid < o.b->core.tid;
- if (b->core.pos != o.b->core.pos) return b->core.pos < o.b->core.pos;
- strand1 = b->core.flag & 0x0010; strand2 = o.b->core.flag & 0x0010;
- if (strand1 != strand2) return strand1 < strand2;
- if (b->core.n_cigar != o.b->core.n_cigar) return b->core.n_cigar < o.b->core.n_cigar;
- p1 = bam1_cigar(b); p2 = bam1_cigar(o.b);
- for (int i = 0; i < (int)b->core.n_cigar; i++) {
- if (*p1 != *p2) return *p1 < *p2;
- ++p1; ++p2;
- }
- return false;
- }
- };
-
- //b is mate 1, b2 is mate 2
- struct PairedEndT {
- bam1_t *b, *b2;
-
- PairedEndT() { b = NULL; b2 = NULL;}
-
- PairedEndT(bam1_t *b, bam1_t *b2) {
- this->b = b;
- this->b2 = b2;
- }
-
- bool operator< (const PairedEndT& o) const {
- int strand1, strand2;
- uint32_t *p1, *p2;
-
- //compare b
- if (b->core.tid != o.b->core.tid) return b->core.tid < o.b->core.tid;
- if (b->core.pos != o.b->core.pos) return b->core.pos < o.b->core.pos;
- strand1 = b->core.flag & 0x0010; strand2 = o.b->core.flag & 0x0010;
- if (strand1 != strand2) return strand1 < strand2;
- if (b->core.n_cigar != o.b->core.n_cigar) return b->core.n_cigar < o.b->core.n_cigar;
- p1 = bam1_cigar(b); p2 = bam1_cigar(o.b);
- for (int i = 0; i < (int)b->core.n_cigar; i++) {
- if (*p1 != *p2) return *p1 < *p2;
- ++p1; ++p2;
- }
-
- //compare b2
- if (b2->core.tid != o.b2->core.tid) return b2->core.tid < o.b2->core.tid;
- if (b2->core.pos != o.b2->core.pos) return b2->core.pos < o.b2->core.pos;
- strand1 = b2->core.flag & 0x0010; strand2 = o.b2->core.flag & 0x0010;
- if (strand1 != strand2) return strand1 < strand2;
- if (b2->core.n_cigar != o.b2->core.n_cigar) return b2->core.n_cigar < o.b2->core.n_cigar;
- p1 = bam1_cigar(b2); p2 = bam1_cigar(o.b2);
- for (int i = 0; i < (int)b2->core.n_cigar; i++) {
- if (*p1 != *p2) return *p1 < *p2;
- ++p1; ++p2;
- }
-
- return false;
- }
- };
-
- uint8_t getMAPQ(double val) {
- double err = 1.0 - val;
- if (err <= 1e-10) return 100;
- return (uint8_t)(-10 * log10(err) + .5); // round it
- }
-
- void push_qname(const uint8_t* qname, int l_qname, std::vector<uint8_t>& data) {
- for (int i = 0; i < l_qname; i++) data.push_back(*(qname + i));
- }
-
- void push_seq(const uint8_t* seq, int readlen, char strand, std::vector<uint8_t>& data) {
- int seq_len = (readlen + 1) / 2;
-
- switch (strand) {
- case '+': for (int i = 0; i < seq_len; i++) data.push_back(*(seq + i)); break;
- case '-':
- uint8_t code, base;
- code = 0; base = 0;
- for (int i = 0; i < readlen; i++) {
- switch (bam1_seqi(seq, readlen - i - 1)) {
- case 1: base = 8; break;
- case 2: base = 4; break;
- case 4: base = 2; break;
- case 8: base = 1; break;
- case 15: base = 15; break;
- default: assert(false);
- }
- code |= base << (4 * (1 - i % 2));
- if (i % 2 == 1) { data.push_back(code); code = 0; }
- }
-
- if (readlen % 2 == 1) { data.push_back(code); }
- break;
- default: assert(false);
- }
- }
-
- void push_qual(const uint8_t* qual, int readlen, char strand, std::vector<uint8_t>& data) {
- switch (strand) {
- case '+': for (int i = 0; i < readlen; i++) data.push_back(*(qual + i)); break;
- case '-': for (int i = readlen - 1; i >= 0; i--) data.push_back(*(qual + i)); break;
- default: assert(false);
- }
- }
-
- //convert transcript coordinate to chromosome coordinate and generate CIGAR string
- void tr2chr(const Transcript&, int, int, int&, int&, std::vector<uint8_t>&);
+ //convert bam1_t
+ void convert(bam1_t*, double);
};
//fn_list can be NULL
-BamWriter::BamWriter(char inpType, const char* inpF, const char* fn_list, const char* outF, const char* chr_list) {
+BamWriter::BamWriter(char inpType, const char* inpF, const char* fn_list, const char* outF, Transcripts& transcripts)
+ : transcripts(transcripts)
+{
switch(inpType) {
case 's': in = samopen(inpF, "r", fn_list); break;
case 'b': in = samopen(inpF, "rb", fn_list); break;
}
assert(in != 0);
- //generate output's header
- bam_header_t *out_header = NULL;
- refmap.clear();
+ //build mappings from external sid to internal sid
+ transcripts.buildMappings(in->header->n_targets, in->header->target_name);
- if (chr_list == NULL) {
- out_header = in->header;
- }
- else {
- out_header = sam_header_read2(chr_list);
+ //generate output's header
+ bam_header_t *out_header = bam_header_dwt(in->header);
- for (int i = 0; i < out_header->n_targets; i++) {
- refmap[out_header->target_name[i]] = i;
- }
- }
+ std::ostringstream strout;
+ strout<<"@HD\tVN:1.4\tSO:unknown\n@PG\tID:RSEM\n";
+ std::string content = strout.str();
+ append_header_text(out_header, content.c_str(), content.length());
out = samopen(outF, "wb", out_header);
assert(out != 0);
- if (chr_list != NULL) { bam_header_destroy(out_header); }
+ bam_header_destroy(out_header);
}
BamWriter::~BamWriter() {
samclose(out);
}
-void BamWriter::work(HitWrapper<SingleHit> wrapper, Transcripts& transcripts) {
+void BamWriter::work(HitWrapper<SingleHit> wrapper) {
bam1_t *b;
- std::string cqname; // cqname : current query name
- std::map<SingleEndT, double> hmap;
- std::map<SingleEndT, double>::iterator hmapIter;
SingleHit *hit;
- int cnt = 0;
+ HIT_INT_TYPE cnt = 0;
- cqname = "";
b = bam_init1();
- hmap.clear();
while (samread(in, b) >= 0) {
-
- if (verbose && cnt > 0 && cnt % 1000000 == 0) { printf("%d entries are finished!\n", cnt); }
++cnt;
+ if (verbose && cnt % 1000000 == 0) { std::cout<< cnt<< " alignment lines are loaded!"<< std::endl; }
- if (b->core.flag & 0x0004) continue;
-
- hit = wrapper.getNextHit();
- assert(hit != NULL);
-
- int sid = b->core.tid + 1;
- assert(sid == hit->getSid());
- const Transcript& transcript = transcripts.getTranscriptAt(sid);
-
- if (transcripts.getType() == 0) {
- int pos = b->core.pos;
- int readlen = b->core.l_qseq;
- uint8_t *qname = b->data, *seq = bam1_seq(b), *qual = bam1_qual(b);
- std::vector<uint8_t> data;
- data.clear();
-
- iter = refmap.find(transcript.getSeqName());
- assert(iter != refmap.end());
- b->core.tid = iter->second;
- b->core.qual = 255;
-
- uint16_t rstrand = b->core.flag & 0x0010; // read strand
- b->core.flag -= rstrand;
- rstrand = (!rstrand && transcript.getStrand() == '+' || rstrand && transcript.getStrand() == '-' ? 0 : 0x0010);
- b->core.flag += rstrand;
-
- push_qname(qname, b->core.l_qname, data);
- int core_pos, core_n_cigar;
- tr2chr(transcript, pos + 1, pos + readlen, core_pos, core_n_cigar, data);
- if (core_pos < 0) b->core.tid = -1;
- b->core.pos = core_pos;
- b->core.n_cigar = core_n_cigar;
- push_seq(seq, readlen, transcript.getStrand(), data);
- push_qual(qual, readlen, transcript.getStrand(), data);
-
- free(b->data);
- b->m_data = b->data_len = data.size() + 7; // 7 extra bytes for ZW tag
- b->l_aux = 7;
- b->data = (uint8_t*)malloc(b->m_data);
- for (int i = 0; i < b->data_len; i++) b->data[i] = data[i];
-
- b->core.bin = bam_reg2bin(b->core.pos, bam_calend(&(b->core), bam1_cigar(b)));
- }
- else {
- b->m_data = b->data_len = b->data_len - b->l_aux + 7; // 7 extra bytes for ZW tag
- b->l_aux = 7;
- b->data = (uint8_t*)realloc(b->data, b->m_data);
- }
-
+ if (!(b->core.flag & 0x0004)) {
+ hit = wrapper.getNextHit();
+ assert(hit != NULL);
- if (cqname != bam1_qname(b)) {
- if (!hmap.empty()) {
- for (hmapIter = hmap.begin(); hmapIter != hmap.end(); hmapIter++) {
- bam1_t *tmp_b = hmapIter->first.b;
- tmp_b->core.qual = getMAPQ(hmapIter->second);
- uint8_t *p = bam1_aux(tmp_b);
- *p = 'Z'; ++p; *p = 'W'; ++p; *p = 'f'; ++p;
- float val = (float)hmapIter->second;
- memcpy(p, &val, 4);
- samwrite(out, tmp_b);
- bam_destroy1(tmp_b); // now hmapIter->b makes no sense
- }
- hmap.clear();
- }
- cqname = bam1_qname(b);
+ assert(transcripts.getInternalSid(b->core.tid + 1) == hit->getSid());
+ convert(b, hit->getConPrb());
}
- hmapIter = hmap.find(SingleEndT(b));
- if (hmapIter == hmap.end()) {
- hmap[SingleEndT(bam_dup1(b))] = hit->getConPrb();
- }
- else {
- hmapIter->second += hit->getConPrb();
- }
+ //if (b->core.qual > 0) samwrite(out, b); // output only when MAPQ > 0
+ samwrite(out, b);
}
assert(wrapper.getNextHit() == NULL);
- if (!hmap.empty()) {
- for (hmapIter = hmap.begin(); hmapIter != hmap.end(); hmapIter++) {
- bam1_t *tmp_b = hmapIter->first.b;
- tmp_b->core.qual = getMAPQ(hmapIter->second);
- uint8_t *p = bam1_aux(tmp_b);
- *p = 'Z'; ++p; *p = 'W'; ++p; *p = 'f'; ++p;
- float val = (float)hmapIter->second;
- memcpy(p, &val, 4);
- samwrite(out, tmp_b);
- bam_destroy1(tmp_b); // now hmapIter->b makes no sense
- }
- hmap.clear();
- }
-
bam_destroy1(b);
- if (verbose) { printf("Bam output file is generated!\n"); }
+ if (verbose) { std::cout<< "Bam output file is generated!"<< std::endl; }
}
-void BamWriter::work(HitWrapper<PairedEndHit> wrapper, Transcripts& transcripts) {
+void BamWriter::work(HitWrapper<PairedEndHit> wrapper) {
bam1_t *b, *b2;
- std::string cqname; // cqname : current query name
- std::map<PairedEndT, double> hmap;
- std::map<PairedEndT, double>::iterator hmapIter;
PairedEndHit *hit;
- int cnt = 0;
+ HIT_INT_TYPE cnt = 0;
- cqname = "";
b = bam_init1();
b2 = bam_init1();
- hmap.clear();
while (samread(in, b) >= 0 && samread(in, b2) >= 0) {
+ cnt += 2;
+ if (verbose && cnt % 1000000 == 0) { std::cout<< cnt<< " alignment lines are loaded!"<< std::endl; }
- if (verbose && cnt > 0 && cnt % 1000000 == 0) { printf("%d entries are finished!\n", cnt); }
- ++cnt;
-
- if (!((b->core.flag & 0x0002) && (b2->core.flag & 0x0002))) continue;
-
- //swap if b is mate 2
- if (b->core.flag & 0x0080) {
- assert(b2->core.flag & 0x0040);
- bam1_t *tmp = b;
- b = b2; b2 = tmp;
- }
-
- hit = wrapper.getNextHit();
- assert(hit != NULL);
-
- int sid = b->core.tid + 1;
- assert(sid == hit->getSid());
- assert(sid == b2->core.tid + 1);
- const Transcript& transcript = transcripts.getTranscriptAt(sid);
-
- if (transcripts.getType() == 0) {
- int pos = b->core.pos, pos2 = b2->core.pos;
- int readlen = b->core.l_qseq, readlen2 = b2->core.l_qseq;
- uint8_t *qname = b->data, *seq = bam1_seq(b), *qual = bam1_qual(b);
- uint8_t *qname2 = b2->data, *seq2 = bam1_seq(b2), *qual2 = bam1_qual(b2);
- std::vector<uint8_t> data, data2;
+ assert((b->core.flag & 0x0001) && (b2->core.flag & 0x0001));
+ assert(((b->core.flag & 0x0040) && (b2->core.flag & 0x0080)) || ((b->core.flag & 0x0080) && (b2->core.flag & 0x0040)));
- data.clear();
- data2.clear();
+ //unalignable reads, skip
+ bool notgood = (b->core.flag & 0x0004) || (b2->core.flag & 0x0004);
- iter = refmap.find(transcript.getSeqName());
- assert(iter != refmap.end());
- b->core.tid = iter->second; b->core.mtid = iter->second;
- b2->core.tid = iter->second; b2->core.mtid = iter->second;
-
- uint16_t rstrand = b->core.flag & 0x0010;
- b->core.flag = b->core.flag - (b->core.flag & 0x0010) - (b->core.flag & 0x0020);
- b2->core.flag = b2->core.flag - (b2->core.flag & 0x0010) - (b2->core.flag & 0x0020);
-
- uint16_t add, add2;
- if (!rstrand && transcript.getStrand() == '+' || rstrand && transcript.getStrand() == '-') {
- add = 0x0020; add2 = 0x0010;
- }
- else {
- add = 0x0010; add2 = 0x0020;
+ if (!notgood) {
+ //swap if b is mate 2
+ if (b->core.flag & 0x0080) {
+ assert(b2->core.flag & 0x0040);
+ bam1_t *tmp = b;
+ b = b2; b2 = tmp;
}
- b->core.flag += add;
- b2->core.flag += add2;
- b->core.qual = b2->core.qual = 255;
+ hit = wrapper.getNextHit();
+ assert(hit != NULL);
- //Do I really need this? The insert size uses transcript coordinates
- if (transcript.getStrand() == '-') {
- b->core.isize = -b->core.isize;
- b2->core.isize = -b2->core.isize;
- }
+ assert(transcripts.getInternalSid(b->core.tid + 1) == hit->getSid());
+ assert(transcripts.getInternalSid(b2->core.tid + 1) == hit->getSid());
- push_qname(qname, b->core.l_qname, data);
- push_qname(qname2, b2->core.l_qname, data2);
- int core_pos, core_n_cigar;
- tr2chr(transcript, pos + 1, pos + readlen, core_pos, core_n_cigar, data);
- if (core_pos < 0) b->core.tid = -1;
- b->core.pos = core_pos; b->core.n_cigar = core_n_cigar;
- tr2chr(transcript, pos2 + 1, pos2 + readlen2, core_pos, core_n_cigar, data2);
- if (core_pos < 0) b2->core.tid = -1;
- b2->core.pos = core_pos; b2->core.n_cigar = core_n_cigar;
- b->core.mpos = b2->core.pos;
- b2->core.mpos = b->core.pos;
- push_seq(seq, readlen, transcript.getStrand(), data);
- push_seq(seq2, readlen2, transcript.getStrand(), data2);
- push_qual(qual, readlen, transcript.getStrand(), data);
- push_qual(qual2, readlen2, transcript.getStrand(), data2);
-
- free(b->data);
- b->m_data = b->data_len = data.size() + 7; // 7 extra bytes for ZW tag
- b->l_aux = 7;
- b->data = (uint8_t*)malloc(b->m_data);
- for (int i = 0; i < b->data_len; i++) b->data[i] = data[i];
-
- free(b2->data);
- b2->m_data = b2->data_len = data2.size() + 7; // 7 extra bytes for ZW tag
- b2->l_aux = 7;
- b2->data = (uint8_t*)malloc(b2->m_data);
- for (int i = 0; i < b2->data_len; i++) b2->data[i] = data2[i];
-
- b->core.bin = bam_reg2bin(b->core.pos, bam_calend(&(b->core), bam1_cigar(b)));
- b2->core.bin = bam_reg2bin(b2->core.pos, bam_calend(&(b2->core), bam1_cigar(b2)));
- }
- else {
- b->m_data = b->data_len = b->data_len - b->l_aux + 7; // 7 extra bytes for ZW tag
- b->l_aux = 7;
- b->data = (uint8_t*)realloc(b->data, b->m_data);
-
- b2->m_data = b2->data_len = b2->data_len - b2->l_aux + 7; // 7 extra bytes for ZW tag
- b2->l_aux = 7;
- b2->data = (uint8_t*)realloc(b2->data, b2->m_data);
- }
+ convert(b, hit->getConPrb());
+ convert(b2, hit->getConPrb());
- if (cqname != bam1_qname(b)) {
- if (!hmap.empty()) {
- for (hmapIter = hmap.begin(); hmapIter != hmap.end(); hmapIter++) {
- bam1_t *tmp_b = hmapIter->first.b;
- bam1_t *tmp_b2 = hmapIter->first.b2;
-
- tmp_b->core.qual = tmp_b2->core.qual = getMAPQ(hmapIter->second);
-
- uint8_t *p = bam1_aux(tmp_b), *p2 = bam1_aux(tmp_b2);
- *p = 'Z'; ++p; *p = 'W'; ++p; *p = 'f'; ++p;
- *p2 = 'Z'; ++p2; *p2 = 'W'; ++p2; *p2 = 'f'; ++p2;
-
- float val = (float)hmapIter->second;
- memcpy(p, &val, 4);
- memcpy(p2, &val, 4);
-
- samwrite(out, tmp_b);
- samwrite(out, tmp_b2);
-
- bam_destroy1(tmp_b);
- bam_destroy1(tmp_b2);
- }
- hmap.clear();
- }
- cqname = bam1_qname(b);
+ // omit "b->core.mpos = b2->core.pos; b2->core.mpos = b->core.pos" for the reason that it is possible that only one mate is alignable
}
- hmapIter = hmap.find(PairedEndT(b, b2));
- if (hmapIter == hmap.end()) {
- hmap[PairedEndT(bam_dup1(b), bam_dup1(b2))] = hit->getConPrb();
- }
- else {
- hmapIter->second += hit->getConPrb();
- }
+ samwrite(out, b);
+ samwrite(out, b2);
}
assert(wrapper.getNextHit() == NULL);
- if (!hmap.empty()) {
- for (hmapIter = hmap.begin(); hmapIter != hmap.end(); hmapIter++) {
- bam1_t *tmp_b = hmapIter->first.b;
- bam1_t *tmp_b2 = hmapIter->first.b2;
-
- tmp_b->core.qual = tmp_b2->core.qual = getMAPQ(hmapIter->second);
-
- uint8_t *p = bam1_aux(tmp_b), *p2 = bam1_aux(tmp_b2);
- *p = 'Z'; ++p; *p = 'W'; ++p; *p = 'f'; ++p;
- *p2 = 'Z'; ++p2; *p2 = 'W'; ++p2; *p2 = 'f'; ++p2;
-
- float val = (float)hmapIter->second;
- memcpy(p, &val, 4);
- memcpy(p2, &val, 4);
-
- samwrite(out, tmp_b);
- samwrite(out, tmp_b2);
-
- bam_destroy1(tmp_b);
- bam_destroy1(tmp_b2);
- }
- hmap.clear();
- }
-
bam_destroy1(b);
bam_destroy1(b2);
- if (verbose) { printf("Bam output file is generated!"); }
+ if (verbose) { std::cout<< "Bam output file is generated!"<< std::endl; }
}
-void BamWriter::tr2chr(const Transcript& transcript, int sp, int ep, int& pos, int& n_cigar, std::vector<uint8_t>& data) {
- int length = transcript.getLength();
- char strand = transcript.getStrand();
- const std::vector<Interval>& structure = transcript.getStructure();
-
- int s, i;
- int oldlen, curlen;
-
- uint32_t operation;
- uint8_t *p;
-
- n_cigar = 0;
- s = structure.size();
-
- if (strand == '-') {
- int tmp = sp;
- sp = length - ep + 1;
- ep = length - tmp + 1;
- }
-
- if (ep < 1 || sp > length) { // a read which align to polyA tails totally!
- pos = (sp > length ? structure[s - 1].end : structure[0].start - 1); // 0 based
-
- n_cigar = 1;
- operation = (ep - sp + 1) << BAM_CIGAR_SHIFT | BAM_CINS; //BAM_CSOFT_CLIP;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
-
- return;
- }
-
- if (sp < 1) {
- n_cigar++;
- operation = (1 - sp) << BAM_CIGAR_SHIFT | BAM_CINS; //BAM_CSOFT_CLIP;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
- sp = 1;
- }
-
- oldlen = curlen = 0;
-
- for (i = 0; i < s; i++) {
- oldlen = curlen;
- curlen += structure[i].end - structure[i].start + 1;
- if (curlen >= sp) break;
- }
- assert(i < s);
- pos = structure[i].start + (sp - oldlen - 1) - 1; // 0 based
-
- while (curlen < ep && i < s) {
- n_cigar++;
- operation = (curlen - sp + 1) << BAM_CIGAR_SHIFT | BAM_CMATCH;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
-
- ++i;
- if (i >= s) continue;
- n_cigar++;
- operation = (structure[i].start - structure[i - 1].end - 1) << BAM_CIGAR_SHIFT | BAM_CREF_SKIP;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
-
- oldlen = curlen;
- sp = oldlen + 1;
- curlen += structure[i].end - structure[i].start + 1;
- }
-
- if (i >= s) {
- n_cigar++;
- operation = (ep - length) << BAM_CIGAR_SHIFT | BAM_CINS; //BAM_CSOFT_CLIP;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
- }
- else {
- n_cigar++;
- operation = (ep - sp + 1) << BAM_CIGAR_SHIFT | BAM_CMATCH;
- p = (uint8_t*)(&operation);
- for (int j = 0; j < 4; j++) data.push_back(*(p + j));
- }
+void BamWriter::convert(bam1_t *b, double prb) {
+ b->core.qual = getMAPQ(prb);
+ float val = (float)prb;
+ bam_aux_append(b, "ZW", 'f', bam_aux_type2size('f'), (uint8_t*)&val);
}
#endif /* BAMWRITER_H_ */