1 | // =============================================================== // |
---|
2 | // // |
---|
3 | // File : AP_codon_table.hxx // |
---|
4 | // Purpose : // |
---|
5 | // // |
---|
6 | // Coded by Ralf Westram (coder@reallysoft.de) // |
---|
7 | // Institute of Microbiology (Technical University Munich) // |
---|
8 | // http://www.arb-home.de/ // |
---|
9 | // // |
---|
10 | // =============================================================== // |
---|
11 | |
---|
12 | #ifndef AP_CODON_TABLE_HXX |
---|
13 | #define AP_CODON_TABLE_HXX |
---|
14 | |
---|
15 | #ifndef ARB_ASSERT_H |
---|
16 | #include <arb_assert.h> |
---|
17 | #endif |
---|
18 | #ifndef STATIC_ASSERT_H |
---|
19 | #include <static_assert.h> |
---|
20 | #endif |
---|
21 | #ifndef _STDINT_H |
---|
22 | #include <stdint.h> |
---|
23 | #endif |
---|
24 | |
---|
25 | #define pn_assert(cond) arb_assert(cond) |
---|
26 | |
---|
27 | // -------------------------------------------------------------------------------- |
---|
28 | |
---|
29 | struct AWT_Codon_Code_Definition { |
---|
30 | const char *name; |
---|
31 | const char *aa; // amino-codes |
---|
32 | const char *starts; |
---|
33 | int embl_feature_transl_table; // number of transl_table-entry in EMBL/GENEBANK features list |
---|
34 | }; |
---|
35 | |
---|
36 | #define AWT_CODON_TABLES 17 // number of different Amino-Translation-Tables |
---|
37 | #define AWT_MAX_CODONS 64 // maximum of possible codon ( = 4^3) |
---|
38 | |
---|
39 | const int AWAR_PROTEIN_TYPE_bacterial_code_index = 8; // contains the index of the bacterial code table |
---|
40 | |
---|
41 | // -------------------------------------------------------------------------------- |
---|
42 | |
---|
43 | #define ALL_TABLES_MASK ((1<<AWT_CODON_TABLES)-1) |
---|
44 | |
---|
45 | class TransTables { |
---|
46 | uint32_t allowed; |
---|
47 | |
---|
48 | static uint32_t bitmask(int nr) { pn_assert(nr >= 0 && nr<AWT_CODON_TABLES); return 1<<nr; } |
---|
49 | |
---|
50 | public: |
---|
51 | TransTables() { allowAll(); } |
---|
52 | |
---|
53 | bool is_allowed(int nr) const { return (allowed & bitmask(nr)) != 0; } |
---|
54 | bool any() const { return allowed; } |
---|
55 | bool none() const { return !allowed; } |
---|
56 | bool all() const { return allowed == ALL_TABLES_MASK; } |
---|
57 | |
---|
58 | void allow(int nr) { allowed |= bitmask(nr); } |
---|
59 | void forbid(int nr) { allowed &= ~bitmask(nr); } |
---|
60 | void forbid(const TransTables& other) { allowed &= ~other.allowed; } |
---|
61 | |
---|
62 | void allowAll() { |
---|
63 | STATIC_ASSERT(sizeof(allowed)*8 > AWT_CODON_TABLES); // one bit wasted |
---|
64 | allowed = ALL_TABLES_MASK; |
---|
65 | } |
---|
66 | void forbidAll() { allowed = 0; } |
---|
67 | void forbidAllBut(int nr) { allowed &= bitmask(nr); } |
---|
68 | |
---|
69 | int explicit_table() const { |
---|
70 | // return explicit table number (or -1 if not exactly one table is allowed) |
---|
71 | if (any()) { |
---|
72 | uint32_t tabs = allowed; |
---|
73 | for (int t = 0; t<AWT_CODON_TABLES; ++t) { |
---|
74 | if (tabs&1) return (tabs == 1) ? t : -1; |
---|
75 | tabs >>= 1; |
---|
76 | } |
---|
77 | pn_assert(0); |
---|
78 | } |
---|
79 | return -1; |
---|
80 | } |
---|
81 | |
---|
82 | bool is_subset_of(const TransTables& other) const { return (allowed&other.allowed) == allowed; } |
---|
83 | |
---|
84 | const char *to_string() const { |
---|
85 | const int MAX_LEN = 42; |
---|
86 | static char buffer[MAX_LEN]; |
---|
87 | char *out = buffer; |
---|
88 | |
---|
89 | for (int a = 0; a<AWT_CODON_TABLES; ++a) { |
---|
90 | if (is_allowed(a)) { |
---|
91 | out += sprintf(out, ",%i", a); |
---|
92 | } |
---|
93 | } |
---|
94 | pn_assert((out-buffer)<MAX_LEN); |
---|
95 | out[0] = 0; |
---|
96 | return buffer+1; |
---|
97 | } |
---|
98 | |
---|
99 | }; |
---|
100 | |
---|
101 | // -------------------------------------------------------------------------------- |
---|
102 | |
---|
103 | void AP_initialize_codon_tables(); |
---|
104 | |
---|
105 | int AWT_embl_transl_table_2_arb_code_nr(int embl_code_nr); |
---|
106 | int AWT_arb_code_nr_2_embl_transl_table(int arb_code_nr); |
---|
107 | |
---|
108 | bool AWT_is_codon(char protein, const char *const dna, const TransTables& allowed, TransTables& allowed_left, const char **fail_reason_ptr = 0); |
---|
109 | const char *AP_get_codons(char protein, int code_nr); |
---|
110 | |
---|
111 | char AWT_is_start_codon(const char *dna, int arb_code_nr); |
---|
112 | |
---|
113 | const char *AP_get_protein_name(char protein); |
---|
114 | const char* AWT_get_codon_code_name(int code); |
---|
115 | |
---|
116 | #ifdef DEBUG |
---|
117 | void AWT_dump_codons(); |
---|
118 | void test_AWT_get_codons(); |
---|
119 | #endif |
---|
120 | |
---|
121 | // -------------------------------------------------------------------------------- |
---|
122 | |
---|
123 | #else |
---|
124 | #error AP_codon_table.hxx included twice |
---|
125 | #endif // AP_CODON_TABLE_HXX |
---|