Mercurial > repos > yufei-luo > s_mart

comparison SMART/Java/Python/Cpp/ncListCreator.cpp @ 18:94ab73e8a190

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author m-zytnicki
date Mon, 29 Apr 2013 03:20:15 -0400
parents
children
comparison
equal deleted inserted replaced
17:b0e8584489e6 18:94ab73e8a190
1 #include "ncListCreator.hpp"
2
3 NCListCreator::NCListCreator(string inputFileName): inputFileName(inputFileName) {}
4
5 string NCListCreator::getFileName(string chromosome) {
6 return inputFileName.substr(0, inputFileName.find_last_of('.')) + "_" + chromosome + ".bed";
7 }
8
9 void NCListCreator::splitFile () {
10 Chromosomes chromosomes;
11 GenomicInterval interval;
12 map <string, ofstream> splittedFiles;
13 map <string, ofstream>::iterator it;
14 ifstream file;
15 string line, fileName;
16 string chromosome;
17 file.open(inputFileName.c_str());
18 if (file.is_open()) {
19 while (file.good()) {
20 getline(file, line);
21 if (line.size() > 0) {
22 interval.parseFromLine(line);
23 chromosomes.insert(interval.chromosome);
24 fileName = getFileName(interval.chromosome);
25 it = splittedFiles.find(interval.chromosome);
26 if (it == splittedFiles.end()) {
27 ofstream outputFile;
28 outputFile.open(fileName.c_str(), ios::out | ios::binary);
29 interval.writeBinary(outputFile);
30 splittedFiles[chromosome] = outputFile;
31 }
32 else {
33 it->second << line << "\n";
34 }
35 }
36 }
37 file.close();
38 for (it = splittedFiles.begin(); it != splittedFiles.end(); it++) {
39 it->second.close();
40 }
41 }
42 else {
43 cout << "Unable to open file" << inputFileName;
44 }
45 }
46
47 void NCListCreator::run() {
48 for (Chromosomes::iterator it = chromosomes.begin(); splittedFiles != chromosomes.end(); splittedFiles++) {
49 buildLists(*it);
50 }
51 }
52
53 void NCListCreator::buildLists(string chromosome) {
54 createTables(chromosome);
55 labelLists();
56 computeSubStart();
57 computeAbsPosition();
58 cleanFiles();
59 transfer(chromosome);
60 }
61
62 void NCListCreator::createTables(string chromosome) {
63 initLists(chromosome);
64 h = new Table(H_CELL_SIZE, nbLists);
65 t = new Table(T_CELL_SIZE, nbLines);
66 l = new Table(L_CELL_SIZE, nbLines);
67 fillTables(chromosome);
68 }
69
70 void NCListCreator::initLists (string chromosome) {
71 nbLists = 0;
72 nbLines = 0;
73 ifstream file;
74 file.open(getFileName(chromosome).c_str(), ios::in | ios::binary);
75 Interval currentInterval, previousInterval;
76 if (file.is_open()) {
77 while (file.good()) {
78 if (currentInterval.parseBinary(file)) {
79 nbLines++;
80 if (previousInterval.include(currentInterval)) {
81 nbLists++;
82 }
83 }
84 previousInterval = currentInterval;
85 }
86 }
87 file.close();
88 }
89
90 void NCListCreator::fillTables (string chromosome) {
91 ifstream file;
92 file.open(getFileName(chromosome).c_str(), ios::in | ios::binary);
93 Interval currentInterval, previousInterval;
94 unsigned int i = 0;
95 if (file.is_open()) {
96 while (file.good()) {
97 if (currentInterval.parseBinary(file)) {
98 t->write(currentInterval.start, i, 0);
99 t->writeHere(currentInterval.end);
100 t->writeHere(-1);
101 t->writeHere(-1);
102 }
103 i++;
104 }
105 file.close();
106 }
107 t->write(SENTINEL, LIST, -1);
108 l->write(SENTINEL, LIST, 0);
109 t->write(SENTINEL, NEW, -1);
110 }
111
112 void NCListCreator::labelLists () {
113 unsigned int nextL = 0, thisL, length;
114 unsigned int p;
115 Interval current, parent;
116 for (unsigned int i = 0; i < nbLines; i++) {
117 p = i - 1;
118 t->moveTo(p, 0);
119 parent.readBinary(t->file);
120 t->moveTo(i, 0);
121 current.readBinary(t->file);
122 while ((p != SENTINEL) && (! parent.include(current))) {
123 p = t->read(p, PARENT);
124 t->moveTo(p, 0);
125 parent.readBinary(t->file);
126 }
127 thisL = t->read(p, LIST);
128 if (thisL == SENTINEL) {
129 thisL = nextL;
130 nextL++;
131 length = 0;
132 t->write(p, LIST, thisL);
133 }
134 else {
135 length = h->read(thisL, LENGTH);
136 }
137 t->write(i, PARENT, p);
138 h->write(thisL, LENGTH, length+1);
139 }
140 }
141
142 void NCListCreator::computeSubStart () {
143 unsigned int total = 0;
144 for (unsigned int i = 0; i < nbLists; i++) {
145 h->write(i, START, total);
146 total += h->read(i, LENGTH);
147 h->write(i, LENGTH, 0);
148 }
149 }
150
151 void NCListCreator::computeAbsPosition () {
152 Value s, e, pt, hp, pl, nb, lp;
153 for (unsigned int i = 0; i < nbLines; i++) {
154 s = t->read(i, START);
155 e = t->read(i, END);
156 pt = t->read(i, PARENT);
157 hp = t->read(pt, LIST);
158 pl = t->read(pt, NEW);
159 nb = h->read(hp, LENGTH);
160 lp = h->read(hp, START) + nb;
161 t->write(i, NEW, lp);
162 l->write(lp, START, s);
163 l->write(lp, END, e);
164 l->write(lp, LIST, SENTINEL);
165 l->write(lp, PARENT, pl);
166 h->write(lp, LENGTH, nb+1);
167 if (nb == 0) {
168 l->write(pl, LIST, hp);
169 }
170 }
171 }
172
173 void NCListCreator::cleanFiles () {
174 t->destroy();
175 }
176
177 void NCListCreator::transfer (string chromosome) {
178 ncLists[chromosome] = NCList(h, l);
179 }