| 1 |
// |
| 2 |
// FastaTools.cpp |
| 3 |
// TopHat |
| 4 |
// |
| 5 |
// Created by Harold Pimentel on 10/27/11. |
| 6 |
// |
| 7 |
|
| 8 |
#include "FastaTools.h" |
| 9 |
|
| 10 |
FastaReader::FastaReader() |
| 11 |
{ |
| 12 |
isPrimed_ = false; |
| 13 |
} |
| 14 |
|
| 15 |
FastaReader::FastaReader(std::string fname) |
| 16 |
{ |
| 17 |
isPrimed_ = false; |
| 18 |
init(fname); |
| 19 |
} |
| 20 |
|
| 21 |
FastaReader::~FastaReader() |
| 22 |
{ |
| 23 |
ifstream_.close(); |
| 24 |
} |
| 25 |
|
| 26 |
void FastaReader::init(std::string fname) |
| 27 |
{ |
| 28 |
if (isPrimed_) { |
| 29 |
std::cerr << "Warning: object has already FastaReader has already been " |
| 30 |
<< "initialized with file: " << fname_ << std::endl; |
| 31 |
return; |
| 32 |
} |
| 33 |
|
| 34 |
fname_ = fname; |
| 35 |
ifstream_.open(fname_.c_str(), std::ios::in); |
| 36 |
if (!ifstream_.good()) { |
| 37 |
std::cerr << "ERROR: Could not open file " << fname_ << |
| 38 |
" in FastaReader" << std::endl; |
| 39 |
exit(1); |
| 40 |
} |
| 41 |
|
| 42 |
// Check the first character to see if it is valid |
| 43 |
char c = ifstream_.peek(); |
| 44 |
if (c != '>') |
| 45 |
{ |
| 46 |
std::cerr << "ERROR: Invalid format for FASTA file. Begins with a '" << |
| 47 |
c << "'instead of a '>'" << std::endl; |
| 48 |
exit(1); |
| 49 |
} |
| 50 |
|
| 51 |
|
| 52 |
isPrimed_ = true; |
| 53 |
|
| 54 |
} |
| 55 |
|
| 56 |
|
| 57 |
bool FastaReader::good() const |
| 58 |
{ |
| 59 |
return ifstream_.good() && !ifstream_.eof(); |
| 60 |
} |
| 61 |
|
| 62 |
// TODO: Make it read the description |
| 63 |
// Right now only reads the ID and sequence |
| 64 |
// Up to caller to allocate memory. |
| 65 |
// Only deallocates memory when there are no more records left |
| 66 |
void FastaReader::next(FastaRecord* rec) |
| 67 |
{ |
| 68 |
if (!isPrimed_) |
| 69 |
{ |
| 70 |
std::cerr << "ERROR: Stream has not been primed (FastaReader)" |
| 71 |
<< std::endl; |
| 72 |
exit(1); |
| 73 |
} |
| 74 |
// Get the entire first line and description |
| 75 |
ifstream_.getline(line_buf_, LINE_BUF_SIZE); |
| 76 |
sscanf(line_buf_, "%s", id_buf_); |
| 77 |
char *cur_id = new char[strlen(id_buf_) + 1]; |
| 78 |
strncpy(cur_id, id_buf_ + 1, strlen(id_buf_)); |
| 79 |
std::string id(static_cast<const char*>(cur_id)); |
| 80 |
delete cur_id; |
| 81 |
|
| 82 |
std::string seq; |
| 83 |
std::string cur_line; |
| 84 |
|
| 85 |
if (!good()) { |
| 86 |
delete rec; |
| 87 |
rec = NULL; |
| 88 |
return; |
| 89 |
} |
| 90 |
|
| 91 |
// Read until you see another ">" |
| 92 |
while (ifstream_.peek() != '>' && ifstream_.good() && !ifstream_.eof()) { |
| 93 |
ifstream_ >> cur_line >> std::ws; |
| 94 |
seq += cur_line; |
| 95 |
} |
| 96 |
|
| 97 |
rec->id_ = id; |
| 98 |
rec->seq_ = seq; |
| 99 |
|
| 100 |
return; |
| 101 |
} |
| 102 |
|
| 103 |
|
| 104 |
FastaWriter::FastaWriter() |
| 105 |
{ |
| 106 |
isPrimed_ = false; |
| 107 |
} |
| 108 |
|
| 109 |
FastaWriter::FastaWriter(std::string fname) |
| 110 |
{ |
| 111 |
isPrimed_ = false; |
| 112 |
init(fname); |
| 113 |
} |
| 114 |
|
| 115 |
FastaWriter::~FastaWriter() |
| 116 |
{ |
| 117 |
ofstream_.close(); |
| 118 |
} |
| 119 |
|
| 120 |
void FastaWriter::init(std::string fname) |
| 121 |
{ |
| 122 |
if (isPrimed_) |
| 123 |
{ |
| 124 |
std::cerr << "Warning: Cannot allocate FastaWriter to file '" << |
| 125 |
fname << "'. It has already been allocated to file '" |
| 126 |
<< fname_ << "'" << std::endl; |
| 127 |
return; |
| 128 |
} |
| 129 |
|
| 130 |
ofstream_.open(fname.c_str(), std::ios::out); |
| 131 |
if (!ofstream_.good()) |
| 132 |
{ |
| 133 |
std::cerr << "ERROR: Could not open " << fname << " for writing in " |
| 134 |
<< "FastaWriter" << std::endl; |
| 135 |
exit(1); |
| 136 |
} |
| 137 |
|
| 138 |
fname_ = fname; |
| 139 |
|
| 140 |
isPrimed_ = true; |
| 141 |
} |
| 142 |
|
| 143 |
void FastaWriter::write(FastaRecord* rec, size_t column_size) |
| 144 |
{ |
| 145 |
if (rec->seq_.length() == 0) |
| 146 |
return; //don't write empty records |
| 147 |
ofstream_ << ">" << rec->id_; //<< std::endl; |
| 148 |
if (rec->desc_.length()) { |
| 149 |
ofstream_ << " " << rec->desc_; |
| 150 |
} |
| 151 |
ofstream_ << std::endl; |
| 152 |
|
| 153 |
// iterate throught the string and print out the string |
| 154 |
size_t start = 0; |
| 155 |
while (start < rec->seq_.length()) |
| 156 |
{ |
| 157 |
ofstream_ << rec->seq_.substr(start, column_size) << std::endl; |
| 158 |
start += column_size; |
| 159 |
} |
| 160 |
} |
