| 1 | |
|---|
| 2 | #include <PT_com.h> |
|---|
| 3 | #include <arbdb.h> |
|---|
| 4 | |
|---|
| 5 | #ifndef NDEBUG |
|---|
| 6 | # define pt_assert(bed) do { if (!(bed)) *(int *)0=0; } while (0) |
|---|
| 7 | # ifndef DEBUG |
|---|
| 8 | # error DEBUG is NOT defined - but it has to! |
|---|
| 9 | # endif |
|---|
| 10 | #else |
|---|
| 11 | # ifdef DEBUG |
|---|
| 12 | # error DEBUG is defined - but it should not! |
|---|
| 13 | # endif |
|---|
| 14 | # define pt_assert(bed) |
|---|
| 15 | #endif |
|---|
| 16 | |
|---|
| 17 | #define PTM_DEBUG |
|---|
| 18 | typedef unsigned long ulong; |
|---|
| 19 | typedef unsigned int uint; |
|---|
| 20 | typedef unsigned char uchar; |
|---|
| 21 | |
|---|
| 22 | #define min(a, b) ((a) < (b)) ? (a) : (b) |
|---|
| 23 | #define max(a, b) ((a) > (b)) ? (a) : (b) |
|---|
| 24 | #define PT_CORE *(int *)0=0; |
|---|
| 25 | #define PT_MAX_MATCHES 10000 |
|---|
| 26 | #define PT_MAX_IDENTS 10000 |
|---|
| 27 | #define PT_POS_TREE_HEIGHT 20 |
|---|
| 28 | #define PT_POS_SECURITY 10 |
|---|
| 29 | #define MIN_PROBE_LENGTH 8 |
|---|
| 30 | #define PT_NAME_HASH_SIZE 10000 |
|---|
| 31 | |
|---|
| 32 | enum PT_MATCH_TYPE { |
|---|
| 33 | PT_MATCH_TYPE_INTEGER = 0, |
|---|
| 34 | PT_MATCH_TYPE_WEIGHTED_PLUS_POS = 1, |
|---|
| 35 | PT_MATCH_TYPE_WEIGHTED = -1 |
|---|
| 36 | }; |
|---|
| 37 | |
|---|
| 38 | |
|---|
| 39 | |
|---|
| 40 | #define MATCHANSWER 50 /* private msg type: match result answer */ |
|---|
| 41 | #define CREATEANSWER 51 /* private msg type: create result answer */ |
|---|
| 42 | #define FINDANSWER 52 /* private msg type: find result answer */ |
|---|
| 43 | |
|---|
| 44 | extern ulong physical_memory; |
|---|
| 45 | struct Hs_struct; |
|---|
| 46 | extern char *pt_error_buffer; |
|---|
| 47 | |
|---|
| 48 | typedef enum type_types_type { |
|---|
| 49 | t_int = 1, |
|---|
| 50 | t_string=0, |
|---|
| 51 | t_float=2 |
|---|
| 52 | } type_types; |
|---|
| 53 | |
|---|
| 54 | typedef enum PT_bases_enum { |
|---|
| 55 | PT_QU = 0, |
|---|
| 56 | PT_N = 1, |
|---|
| 57 | PT_A, PT_C, PT_G, PT_T, PT_B_MAX } PT_BASES; |
|---|
| 58 | |
|---|
| 59 | /* POS TREE */ |
|---|
| 60 | typedef enum enum_PT_NODE_TYPE { |
|---|
| 61 | PT_NT_LEAF = 0, |
|---|
| 62 | PT_NT_CHAIN = 1, |
|---|
| 63 | PT_NT_NODE = 2, |
|---|
| 64 | PT_NT_SINGLE_NODE = 3, /* stage 3 */ |
|---|
| 65 | PT_NT_SAVED = 3, /* stage 1+2 */ |
|---|
| 66 | PT_NT_UNDEF = 4 |
|---|
| 67 | } PT_NODE_TYPE; |
|---|
| 68 | |
|---|
| 69 | typedef struct POS_TREE_struct { |
|---|
| 70 | uchar flags; |
|---|
| 71 | char data; |
|---|
| 72 | } POS_TREE; |
|---|
| 73 | |
|---|
| 74 | typedef struct PTMM_struct { |
|---|
| 75 | char *base; |
|---|
| 76 | int stage1; |
|---|
| 77 | int stage2; |
|---|
| 78 | int stage3; |
|---|
| 79 | int mode; |
|---|
| 80 | } PTM2; |
|---|
| 81 | |
|---|
| 82 | |
|---|
| 83 | |
|---|
| 84 | /* Probe search */ |
|---|
| 85 | |
|---|
| 86 | |
|---|
| 87 | struct probe_statistic { |
|---|
| 88 | int match_count; /* Counter for matches */ |
|---|
| 89 | double rel_match_count; /* match_count / (seq_len - probe_len + 1) */ |
|---|
| 90 | }; |
|---|
| 91 | |
|---|
| 92 | struct probe_input_data { /* every taxas own data */ |
|---|
| 93 | /******* name and sequence *********/ |
|---|
| 94 | char *data; |
|---|
| 95 | long checksum; |
|---|
| 96 | int size; |
|---|
| 97 | char *name; |
|---|
| 98 | char *fullname; |
|---|
| 99 | GBDATA *gbd; |
|---|
| 100 | |
|---|
| 101 | /********* probe design ************/ |
|---|
| 102 | int is_group; /* -1: nevermind |
|---|
| 103 | 0: no group |
|---|
| 104 | 1: group */ |
|---|
| 105 | /* probe design (match) */ |
|---|
| 106 | PT_probematch *match; /* best hit for PT_new_design */ |
|---|
| 107 | |
|---|
| 108 | /********* find family ************/ |
|---|
| 109 | struct probe_statistic stat; |
|---|
| 110 | |
|---|
| 111 | /********* free pointer ************/ |
|---|
| 112 | int next; |
|---|
| 113 | }; |
|---|
| 114 | |
|---|
| 115 | struct probe_statistic_struct { |
|---|
| 116 | int cut_offs; /* statistic of chains */ |
|---|
| 117 | int single_node; |
|---|
| 118 | int short_node; |
|---|
| 119 | int long_node; |
|---|
| 120 | int longs; |
|---|
| 121 | int shorts; |
|---|
| 122 | int shorts2; |
|---|
| 123 | int chars; |
|---|
| 124 | |
|---|
| 125 | }; |
|---|
| 126 | |
|---|
| 127 | extern struct probe_struct_global { |
|---|
| 128 | GBDATA *gb_main; /* ARBDB interface */ |
|---|
| 129 | GBDATA *gb_species_data; |
|---|
| 130 | GBDATA *gb_extended_data; |
|---|
| 131 | char *alignment_name; |
|---|
| 132 | GB_HASH *namehash; /* name to int */ |
|---|
| 133 | |
|---|
| 134 | struct probe_input_data *data; /* the internal database */ |
|---|
| 135 | char *ecoli; /* the ecoli sequenz */ |
|---|
| 136 | void *bi_ecoli; |
|---|
| 137 | |
|---|
| 138 | int data_count; |
|---|
| 139 | int max_size; /* maximum sequence len */ |
|---|
| 140 | long char_count; /* number of all 'acgtuACGTU' */ |
|---|
| 141 | |
|---|
| 142 | int mismatches; /* chain handle in match */ |
|---|
| 143 | double wmismatches; |
|---|
| 144 | int N_mismatches; |
|---|
| 145 | int w_N_mismatches[PT_POS_TREE_HEIGHT+PT_POS_SECURITY+1]; |
|---|
| 146 | |
|---|
| 147 | int reversed; /* tell the matcher whether probe is |
|---|
| 148 | reversed */ |
|---|
| 149 | |
|---|
| 150 | double *pos_to_weight; /* position to weight */ |
|---|
| 151 | char complement[256]; /* complement */ |
|---|
| 152 | |
|---|
| 153 | int deep; /* for probe matching */ |
|---|
| 154 | int height; |
|---|
| 155 | int length; |
|---|
| 156 | int apos; |
|---|
| 157 | |
|---|
| 158 | int sort_by; |
|---|
| 159 | |
|---|
| 160 | char *probe; /* probe design + chains */ |
|---|
| 161 | char *main_probe; |
|---|
| 162 | |
|---|
| 163 | char *server_name; /* name of this server */ |
|---|
| 164 | aisc_com *link; |
|---|
| 165 | T_PT_MAIN main; |
|---|
| 166 | struct Hs_struct *com_so; /* the communication socket */ |
|---|
| 167 | POS_TREE *pt; |
|---|
| 168 | PTM2 *ptmain; |
|---|
| 169 | probe_statistic_struct stat; |
|---|
| 170 | } psg; |
|---|
| 171 | |
|---|
| 172 | |
|---|
| 173 | #define PT_base_string_counter_eof(str) (*(unsigned char *)(str) == 255) |
|---|