1 // ***************************************************************************
2 // BamConstants.h (c) 2011 Derek Barnett
3 // Marth Lab, Department of Biology, Boston College
4 // ---------------------------------------------------------------------------
5 // Last modified: 4 October 2011 (DB)
6 // ---------------------------------------------------------------------------
7 // Provides basic constants for handling BAM files.
8 // ***************************************************************************
10 #ifndef BAM_CONSTANTS_H
11 #define BAM_CONSTANTS_H
13 #include <api/api_global.h>
17 /*! \namespace BamTools::Constants
18 \brief Provides basic constants for handling BAM files.
24 const int BAM_SIZEOF_INT = 4;
26 // header magic number
27 const char* const BAM_HEADER_MAGIC = "BAM\1";
28 const unsigned int BAM_HEADER_MAGIC_LENGTH = 4;
30 // BAM alignment core size
31 const int BAM_CORE_SIZE = 32;
32 const int BAM_CORE_BUFFER_SIZE = 8;
34 // BAM alignment flags
35 const int BAM_ALIGNMENT_PAIRED = 0x0001;
36 const int BAM_ALIGNMENT_PROPER_PAIR = 0x0002;
37 const int BAM_ALIGNMENT_UNMAPPED = 0x0004;
38 const int BAM_ALIGNMENT_MATE_UNMAPPED = 0x0008;
39 const int BAM_ALIGNMENT_REVERSE_STRAND = 0x0010;
40 const int BAM_ALIGNMENT_MATE_REVERSE_STRAND = 0x0020;
41 const int BAM_ALIGNMENT_READ_1 = 0x0040;
42 const int BAM_ALIGNMENT_READ_2 = 0x0080;
43 const int BAM_ALIGNMENT_SECONDARY = 0x0100;
44 const int BAM_ALIGNMENT_QC_FAILED = 0x0200;
45 const int BAM_ALIGNMENT_DUPLICATE = 0x0400;
48 const char* const BAM_CIGAR_LOOKUP = "MIDNSHP=X";
49 const int BAM_CIGAR_MATCH = 0;
50 const int BAM_CIGAR_INS = 1;
51 const int BAM_CIGAR_DEL = 2;
52 const int BAM_CIGAR_REFSKIP = 3;
53 const int BAM_CIGAR_SOFTCLIP = 4;
54 const int BAM_CIGAR_HARDCLIP = 5;
55 const int BAM_CIGAR_PAD = 6;
56 const int BAM_CIGAR_SEQMATCH = 7;
57 const int BAM_CIGAR_MISMATCH = 8;
59 const char BAM_CIGAR_MATCH_CHAR = 'M';
60 const char BAM_CIGAR_INS_CHAR = 'I';
61 const char BAM_CIGAR_DEL_CHAR = 'D';
62 const char BAM_CIGAR_REFSKIP_CHAR = 'N';
63 const char BAM_CIGAR_SOFTCLIP_CHAR = 'S';
64 const char BAM_CIGAR_HARDCLIP_CHAR = 'H';
65 const char BAM_CIGAR_PAD_CHAR = 'P';
66 const char BAM_CIGAR_SEQMATCH_CHAR = '=';
67 const char BAM_CIGAR_MISMATCH_CHAR = 'X';
69 const int BAM_CIGAR_SHIFT = 4;
70 const int BAM_CIGAR_MASK = ((1 << BAM_CIGAR_SHIFT) - 1);
73 const char BAM_TAG_TYPE_ASCII = 'A';
74 const char BAM_TAG_TYPE_UINT8 = 'c';
75 const char BAM_TAG_TYPE_INT8 = 'C';
76 const char BAM_TAG_TYPE_UINT16 = 's';
77 const char BAM_TAG_TYPE_INT16 = 'S';
78 const char BAM_TAG_TYPE_UINT32 = 'i';
79 const char BAM_TAG_TYPE_INT32 = 'I';
80 const char BAM_TAG_TYPE_FLOAT = 'f';
81 const char BAM_TAG_TYPE_STRING = 'Z';
82 const char BAM_TAG_TYPE_HEX = 'H';
83 const char BAM_TAG_TYPE_ARRAY = 'B';
85 const size_t BAM_TAG_TAGSIZE = 2;
86 const size_t BAM_TAG_TYPESIZE = 1;
87 const int BAM_TAG_ARRAYBASE_SIZE = 8;
90 const char* const BAM_DNA_LOOKUP = "=ACMGRSVTWYHKDBN";
91 const unsigned char BAM_BASECODE_EQUAL = 0;
92 const unsigned char BAM_BASECODE_A = 1;
93 const unsigned char BAM_BASECODE_C = 2;
94 const unsigned char BAM_BASECODE_G = 4;
95 const unsigned char BAM_BASECODE_T = 8;
96 const unsigned char BAM_BASECODE_N = 15;
98 const char BAM_DNA_EQUAL = '=';
99 const char BAM_DNA_A = 'A';
100 const char BAM_DNA_C = 'C';
101 const char BAM_DNA_G = 'G';
102 const char BAM_DNA_T = 'T';
103 const char BAM_DNA_N = 'N';
104 const char BAM_DNA_DEL = '-';
105 const char BAM_DNA_PAD = '*';
108 const int GZIP_ID1 = 31;
109 const int GZIP_ID2 = 139;
110 const int CM_DEFLATE = 8;
111 const int FLG_FEXTRA = 4;
112 const int OS_UNKNOWN = 255;
113 const int BGZF_XLEN = 6;
114 const int BGZF_ID1 = 66;
115 const int BGZF_ID2 = 67;
116 const int BGZF_LEN = 2;
117 const int GZIP_WINDOW_BITS = -15;
118 const int Z_DEFAULT_MEM_LEVEL = 8;
121 const int BGZF_BLOCK_HEADER_LENGTH = 18;
122 const int BGZF_BLOCK_FOOTER_LENGTH = 8;
123 const int BGZF_MAX_BLOCK_SIZE = 65536;
124 const int BGZF_DEFAULT_BLOCK_SIZE = 65536;
126 } // namespace Constants
128 // -------------------------
129 // tag-type helper structs
130 // -------------------------
132 // fail on any types not specified below
134 struct TagTypeHelper {
135 static bool CanConvertFrom(const char) { assert(false); return false; }
136 static bool CanConvertTo(const char) { assert(false); return false; }
137 static char TypeCode(void) { assert(false); return 0; }
141 struct TagTypeHelper<uint8_t> {
142 static bool CanConvertFrom(const char c) {
143 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
144 c == Constants::BAM_TAG_TYPE_UINT8 );
146 static bool CanConvertTo(const char c) {
147 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
148 c == Constants::BAM_TAG_TYPE_UINT8 ||
149 c == Constants::BAM_TAG_TYPE_UINT16 ||
150 c == Constants::BAM_TAG_TYPE_UINT32 );
153 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_UINT8; }
157 struct TagTypeHelper<int8_t> {
158 static bool CanConvertFrom(const char c) {
159 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
160 c == Constants::BAM_TAG_TYPE_INT8 );
162 static bool CanConvertTo(const char c) {
163 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
164 c == Constants::BAM_TAG_TYPE_INT8 ||
165 c == Constants::BAM_TAG_TYPE_INT16 ||
166 c == Constants::BAM_TAG_TYPE_INT32 );
168 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_INT8; }
172 struct TagTypeHelper<uint16_t> {
173 static bool CanCovnertFrom(const char c) {
174 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
175 c == Constants::BAM_TAG_TYPE_UINT8 ||
176 c == Constants::BAM_TAG_TYPE_UINT16 );
178 static bool CanConvertTo(const char c) {
179 return ( c == Constants::BAM_TAG_TYPE_UINT16 ||
180 c == Constants::BAM_TAG_TYPE_UINT32);
182 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_UINT16; }
186 struct TagTypeHelper<int16_t> {
187 static bool CanConvertFrom(const char c) {
188 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
189 c == Constants::BAM_TAG_TYPE_INT8 ||
190 c == Constants::BAM_TAG_TYPE_INT16 );
192 static bool CanConvertTo(const char c) {
193 return ( c == Constants::BAM_TAG_TYPE_INT16 ||
194 c == Constants::BAM_TAG_TYPE_INT32);
196 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_INT16; }
200 struct TagTypeHelper<uint32_t> {
201 static bool CanConvertFrom(const char c) {
202 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
203 c == Constants::BAM_TAG_TYPE_UINT8 ||
204 c == Constants::BAM_TAG_TYPE_UINT16 ||
205 c == Constants::BAM_TAG_TYPE_UINT32 );
207 static bool CanConvertTo(const char c) {
208 return ( c == Constants::BAM_TAG_TYPE_UINT32 );
210 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_UINT32; }
214 struct TagTypeHelper<int32_t> {
215 static bool CanConvertFrom(const char c) {
216 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
217 c == Constants::BAM_TAG_TYPE_INT8 ||
218 c == Constants::BAM_TAG_TYPE_INT16 ||
219 c == Constants::BAM_TAG_TYPE_INT32 );
221 static bool CanConvertTo(const char c) {
222 return ( c == Constants::BAM_TAG_TYPE_INT32 );
224 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_INT32; }
228 struct TagTypeHelper<float> {
229 static bool CanConvertFrom(const char c) {
230 return ( c == Constants::BAM_TAG_TYPE_ASCII ||
231 c == Constants::BAM_TAG_TYPE_UINT8 ||
232 c == Constants::BAM_TAG_TYPE_INT8 ||
233 c == Constants::BAM_TAG_TYPE_UINT16 ||
234 c == Constants::BAM_TAG_TYPE_INT16 ||
235 c == Constants::BAM_TAG_TYPE_UINT32 ||
236 c == Constants::BAM_TAG_TYPE_INT32 ||
237 c == Constants::BAM_TAG_TYPE_FLOAT);
239 static bool CanConvertTo(const char c) {
240 return ( c == Constants::BAM_TAG_TYPE_FLOAT );
242 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_FLOAT; }
246 struct TagTypeHelper<std::string> {
247 static bool CanConvertFrom(const char c) {
248 return ( c == Constants::BAM_TAG_TYPE_HEX ||
249 c == Constants::BAM_TAG_TYPE_STRING );
251 static bool CanConvertTo(const char c) {
252 return ( c == Constants::BAM_TAG_TYPE_HEX ||
253 c == Constants::BAM_TAG_TYPE_STRING );
255 static char TypeCode(void) { return Constants::BAM_TAG_TYPE_STRING; }
258 } // namespace BamTools
260 #endif // BAM_CONSTANTS_H