annotate spp/src/maqread.cpp @ 6:ce08b0efa3fd draft

Uploaded
author zzhou
date Tue, 27 Nov 2012 16:11:40 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
1 #include "pc.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
2 #include <vector>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
3 #include <string.h>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
4 #include <iostream>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
5 #include <fstream>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
6 #include <sstream>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
7 #include <strstream>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
8 #include <algorithm>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
9 #include <string>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
10 #include <functional>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
11 #include <utility>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
12 #include <zlib.h>
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
13
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
14 extern "C" {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
15 #include "R.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
16 #include "Rmath.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
17 #include "Rinternals.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
18 #include "Rdefines.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
19 #include "maqmap.h"
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
20 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
21
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
22 using namespace std;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
23 using namespace __gnu_cxx;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
24
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
25
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
26 class lessAbsoluteValue {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
27 public:
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
28 bool operator()(int a, int b) const {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
29 return abs(a) < abs(b);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
30 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
31 };
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
32
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
33
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
34
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
35 //#define DEBUG 1
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
36
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
37 extern "C" {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
38
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
39 // read in text version of maq map
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
40 SEXP read_binmaqmap(SEXP filename,SEXP read_tag_names_R) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
41
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
42 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
43 Rprintf("start\n");
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
44 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
45 const char* fname=CHAR(asChar(filename));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
46 int read_names=*(INTEGER(read_tag_names_R));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
47 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
48 Rprintf("fname=%s\n",fname);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
49 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
50
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
51 // main data vector
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
52 // chr - pos
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
53 vector< vector<int> > pos;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
54 vector< vector<int> > posnm; // number of mismatches
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
55 vector< vector<string> > tagnames;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
56
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
57 // chromosome map
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
58 hash_map<string, int, hash<string>,equal_to<string> > cind_map;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
59 vector<string> cnames;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
60
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
61
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
62 gzFile f=gzopen(fname,"r");
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
63
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
64 maqmap_t *m = maqmap_read_header(f);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
65 maqmap1_t *m1, mm1;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
66 m1 = &mm1;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
67
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
68 if (!f) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
69 cout<<"can't open input file \""<<fname<<"\"\n";
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
70 } else {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
71 Rprintf("opened %s\n",fname);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
72
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
73 // read in bed line
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
74 string line;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
75 int fcount=0;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
76 while(maqmap_read1(f, m1)) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
77 string tagname=string(m1->name);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
78 string chr=string(m->ref_name[m1->seqid]);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
79 int len=m1->size;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
80 int fpos=(m1->pos>>1) + 1;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
81 if(m1->pos&1) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
82 fpos=-1*(fpos+len-1);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
83 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
84 int nm=m1->info1&0xf;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
85
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
86 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
87 Rprintf("read in map line chr=%s tagname=%s fpos=%d, nm=%d, len=%d\n",chr.c_str(),tagname.c_str(),fpos,nm,len);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
88 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
89
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
90
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
91 // determine the chromosome index
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
92 hash_map<string, int, hash<string>,equal_to<string> >::const_iterator li=cind_map.find(chr);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
93 int cind=-1;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
94 if(li==cind_map.end()) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
95 // register new chromosome
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
96 cind=cnames.size();
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
97 cnames.push_back(chr);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
98 cind_map[chr]=cind;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
99 // allocate new pos vector
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
100 pos.push_back(vector<int>());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
101 posnm.push_back(vector<int>());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
102 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
103 tagnames.push_back(vector<string>());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
104 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
105 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
106 Rprintf("registered new chromosome %s with cind=%d, pos.size=%d\n",chr.c_str(),cind,pos.size());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
107 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
108 } else {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
109 cind=li->second;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
110 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
111 fcount++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
112 (pos[cind]).push_back(fpos);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
113 (posnm[cind]).push_back(nm);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
114 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
115 (tagnames[cind]).push_back(tagname);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
116 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
117 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
118 Rprintf("read in position chr=%s cind=%d fpos=%d, nm=%d, len=%d\n",chr.c_str(),cind,fpos,nm,len);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
119 if(fcount>30) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
120 break;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
121 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
122 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
123
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
124 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
125 gzclose(f);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
126 Rprintf("done. read %d fragments\n",fcount);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
127 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
128
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
129
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
130 // construct output structures
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
131 SEXP chnames;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
132 int np=0; // number of protections
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
133 PROTECT(chnames = allocVector(STRSXP, cnames.size()));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
134 for(vector<string>::const_iterator csi=cnames.begin();csi!=cnames.end();++csi) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
135 SET_STRING_ELT(chnames, csi-cnames.begin(), mkChar(csi->c_str()));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
136 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
137 np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
138
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
139 // sort
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
140 //for(vector<vector<int> >::iterator csi=pos.begin();csi!=pos.end();++csi) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
141 // sort(csi->begin(), csi->end(), lessAbsoluteValue());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
142 //}
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
143
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
144 SEXP ans;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
145 PROTECT(ans = allocVector(VECSXP, cnames.size())); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
146 vector<vector<int> >::const_iterator nsi;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
147 vector<vector<string> >::const_iterator ssi;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
148 for(vector<vector<int> >::const_iterator csi=pos.begin();csi!=pos.end();++csi) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
149 nsi=posnm.begin()+(csi-pos.begin());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
150
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
151 SEXP dv,dnames_R;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
152 PROTECT(dnames_R = allocVector(STRSXP, 2+read_names)); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
153 SET_STRING_ELT(dnames_R, 0, mkChar("t"));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
154 SET_STRING_ELT(dnames_R, 1, mkChar("n"));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
155 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
156 SET_STRING_ELT(dnames_R, 2, mkChar("s"));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
157 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
158
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
159
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
160
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
161 SEXP tv,nv,sv;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
162 PROTECT(tv=allocVector(INTSXP,csi->size())); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
163 PROTECT(nv=allocVector(INTSXP,csi->size())); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
164 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
165 PROTECT(sv=allocVector(STRSXP,csi->size())); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
166 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
167 int* i_tv=INTEGER(tv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
168 int* i_nv=INTEGER(nv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
169
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
170 int i=0;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
171 vector<int>::const_iterator ini=nsi->begin();
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
172 for(vector<int> ::const_iterator pi=csi->begin();pi!=csi->end();++pi) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
173 i_tv[i]=*pi;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
174 i_nv[i]=*ini++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
175 i++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
176 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
177 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
178 int i=0;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
179 ssi=tagnames.begin()+(csi-pos.begin());
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
180 for(vector<string>::const_iterator si=ssi->begin();si!=ssi->end();++si) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
181 SET_STRING_ELT(sv,i,mkChar(si->c_str()));
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
182 i++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
183 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
184 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
185 PROTECT(dv = allocVector(VECSXP, 2+read_names)); np++;
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
186 SET_VECTOR_ELT(dv, 0, tv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
187 SET_VECTOR_ELT(dv, 1, nv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
188 if(read_names) {
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
189 SET_VECTOR_ELT(dv, 2, sv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
190 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
191 setAttrib(dv, R_NamesSymbol, dnames_R);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
192
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
193 SET_VECTOR_ELT(ans, csi-pos.begin(), dv);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
194 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
195
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
196 setAttrib(ans,R_NamesSymbol,chnames);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
197
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
198 #ifdef DEBUG
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
199 Rprintf("unprotecting %d elements\n",np);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
200 #endif
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
201
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
202 UNPROTECT(np);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
203 return(ans);
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
204 }
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
205
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
206
ce08b0efa3fd Uploaded
zzhou
parents:
diff changeset
207 }