1 | #ifndef PROBE_H |
---|
2 | #define PROBE_H |
---|
3 | |
---|
4 | #ifndef __LIST__ |
---|
5 | #include <list> |
---|
6 | #endif |
---|
7 | #ifndef __SET__ |
---|
8 | #include <set> |
---|
9 | #endif |
---|
10 | |
---|
11 | #ifndef ARBDB_H |
---|
12 | #include <arbdb.h> |
---|
13 | #endif |
---|
14 | #ifndef PT_COM_H |
---|
15 | #include <PT_com.h> |
---|
16 | #endif |
---|
17 | #ifndef AISC_GEN_SERVER_INCLUDED |
---|
18 | #include <PT_server.h> |
---|
19 | #endif |
---|
20 | |
---|
21 | #ifndef ARB_ASSERT_H |
---|
22 | #include <arb_assert.h> |
---|
23 | #endif |
---|
24 | |
---|
25 | #define pt_assert(bed) arb_assert(bed) |
---|
26 | |
---|
27 | #if defined(DEBUG) |
---|
28 | #define PTM_DEBUG |
---|
29 | #endif // DEBUG |
---|
30 | |
---|
31 | typedef unsigned long ulong; |
---|
32 | typedef unsigned int uint; |
---|
33 | typedef unsigned char uchar; |
---|
34 | |
---|
35 | #define min(a, b) ((a) < (b)) ? (a) : (b) |
---|
36 | #define max(a, b) ((a) > (b)) ? (a) : (b) |
---|
37 | |
---|
38 | #define PT_CORE *(int *)0 = 0; |
---|
39 | |
---|
40 | #if defined(DEVEL_KAI) |
---|
41 | #define PT_MAX_MATCHES 1024*256 |
---|
42 | #else |
---|
43 | #define PT_MAX_MATCHES 100000 |
---|
44 | #endif // DEVEL_KAI |
---|
45 | |
---|
46 | // #define PT_MAX_IDENTS 10000 |
---|
47 | #define PT_POS_TREE_HEIGHT 20 |
---|
48 | #define PT_POS_SECURITY 10 |
---|
49 | #define MIN_PROBE_LENGTH 6 |
---|
50 | #define PT_NAME_HASH_SIZE 10000 |
---|
51 | |
---|
52 | enum PT_MATCH_TYPE { |
---|
53 | PT_MATCH_TYPE_INTEGER = 0, |
---|
54 | PT_MATCH_TYPE_WEIGHTED_PLUS_POS = 1, |
---|
55 | PT_MATCH_TYPE_WEIGHTED = -1 |
---|
56 | }; |
---|
57 | |
---|
58 | |
---|
59 | |
---|
60 | #define MATCHANSWER 50 /* private msg type: match result answer */ |
---|
61 | #define CREATEANSWER 51 /* private msg type: create result answer */ |
---|
62 | #define FINDANSWER 52 /* private msg type: find result answer */ |
---|
63 | |
---|
64 | extern ulong physical_memory; |
---|
65 | struct Hs_struct; |
---|
66 | extern char *pt_error_buffer; |
---|
67 | |
---|
68 | typedef enum type_types_type { |
---|
69 | t_int = 1, |
---|
70 | t_string = 0, |
---|
71 | t_float = 2 |
---|
72 | } type_types; |
---|
73 | |
---|
74 | typedef enum PT_bases_enum { |
---|
75 | PT_QU = 0, |
---|
76 | PT_N = 1, |
---|
77 | PT_A, |
---|
78 | PT_C, |
---|
79 | PT_G, |
---|
80 | PT_T, |
---|
81 | PT_B_MAX |
---|
82 | } PT_BASES; |
---|
83 | |
---|
84 | /* POS TREE */ |
---|
85 | typedef enum enum_PT_NODE_TYPE { |
---|
86 | PT_NT_LEAF = 0, |
---|
87 | PT_NT_CHAIN = 1, |
---|
88 | PT_NT_NODE = 2, |
---|
89 | PT_NT_SINGLE_NODE = 3, /* stage 3 */ |
---|
90 | PT_NT_SAVED = 3, /* stage 1+2 */ |
---|
91 | PT_NT_UNDEF = 4 |
---|
92 | } PT_NODE_TYPE; |
---|
93 | |
---|
94 | typedef struct POS_TREE_struct { |
---|
95 | uchar flags; |
---|
96 | char data; |
---|
97 | } POS_TREE; |
---|
98 | |
---|
99 | typedef struct PTMM_struct { |
---|
100 | char *base; |
---|
101 | int stage1; |
---|
102 | int stage2; |
---|
103 | int stage3; |
---|
104 | int mode; |
---|
105 | } PTM2; |
---|
106 | |
---|
107 | |
---|
108 | |
---|
109 | /* Probe search */ |
---|
110 | |
---|
111 | |
---|
112 | struct probe_statistic { |
---|
113 | int match_count; /* Counter for matches */ |
---|
114 | double rel_match_count; /* match_count / (seq_len - probe_len + 1) */ |
---|
115 | }; |
---|
116 | |
---|
117 | struct probe_input_data { /* every taxas own data */ |
---|
118 | /******* name and sequence *********/ |
---|
119 | char *data; |
---|
120 | long checksum; |
---|
121 | int size; |
---|
122 | char *name; |
---|
123 | char *fullname; |
---|
124 | GBDATA *gbd; |
---|
125 | |
---|
126 | /********* probe design ************/ |
---|
127 | int is_group; /* -1: nevermind |
---|
128 | 0: no group |
---|
129 | 1: group */ |
---|
130 | |
---|
131 | /* probe design (match) */ |
---|
132 | PT_probematch *match; /* best hit for PT_new_design */ |
---|
133 | |
---|
134 | /********* find family ************/ |
---|
135 | struct probe_statistic stat; |
---|
136 | |
---|
137 | /********* free pointer ************/ |
---|
138 | int next; |
---|
139 | }; |
---|
140 | |
---|
141 | struct probe_statistic_struct { |
---|
142 | int cut_offs; /* statistic of chains */ |
---|
143 | int single_node; |
---|
144 | int short_node; |
---|
145 | int long_node; |
---|
146 | int longs; |
---|
147 | int shorts; |
---|
148 | int shorts2; |
---|
149 | int chars; |
---|
150 | |
---|
151 | }; |
---|
152 | |
---|
153 | class BI_ecoli_ref; |
---|
154 | |
---|
155 | extern struct probe_struct_global { |
---|
156 | GBDATA *gb_main; /* ARBDB interface */ |
---|
157 | GBDATA *gb_species_data; |
---|
158 | GBDATA *gb_extended_data; |
---|
159 | char *alignment_name; |
---|
160 | GB_HASH *namehash; /* name to int */ |
---|
161 | |
---|
162 | struct probe_input_data *data; /* the internal database */ |
---|
163 | |
---|
164 | char *ecoli; /* the ecoli sequenz */ |
---|
165 | BI_ecoli_ref *bi_ecoli; |
---|
166 | |
---|
167 | int data_count; |
---|
168 | int max_size; /* maximum sequence len */ |
---|
169 | long char_count; /* number of all 'acgtuACGTU' */ |
---|
170 | |
---|
171 | int mismatches; /* chain handle in match */ |
---|
172 | double wmismatches; |
---|
173 | int N_mismatches; |
---|
174 | int w_N_mismatches[PT_POS_TREE_HEIGHT+PT_POS_SECURITY+1]; |
---|
175 | |
---|
176 | int reversed; /* tell the matcher whether probe is reversed */ |
---|
177 | |
---|
178 | double *pos_to_weight; /* position to weight */ |
---|
179 | char complement[256]; /* complement */ |
---|
180 | |
---|
181 | int deep; /* for probe matching */ |
---|
182 | int height; |
---|
183 | int length; |
---|
184 | int apos; |
---|
185 | |
---|
186 | int sort_by; |
---|
187 | |
---|
188 | char *probe; /* probe design + chains */ |
---|
189 | char *main_probe; |
---|
190 | |
---|
191 | char *server_name; /* name of this server */ |
---|
192 | aisc_com *link; |
---|
193 | T_PT_MAIN main; |
---|
194 | struct Hs_struct *com_so; /* the communication socket */ |
---|
195 | POS_TREE *pt; |
---|
196 | PTM2 *ptmain; |
---|
197 | |
---|
198 | probe_statistic_struct stat; |
---|
199 | |
---|
200 | } psg; |
---|
201 | |
---|
202 | class gene_struct { |
---|
203 | char *gene_name; |
---|
204 | const char *arb_species_name; // pointers into 'gene_name' |
---|
205 | const char *arb_gene_name; |
---|
206 | |
---|
207 | void init(const char *gene_name_, const char *arb_species_name_, const char *arb_gene_name_) { |
---|
208 | int gene_name_len = strlen(gene_name_); |
---|
209 | int arb_species_name_len = strlen(arb_species_name_); |
---|
210 | int arb_gene_name_len = strlen(arb_gene_name_); |
---|
211 | |
---|
212 | int fulllen = gene_name_len+1+arb_species_name_len+1+arb_gene_name_len+1; |
---|
213 | gene_name = new char[fulllen]; |
---|
214 | strcpy(gene_name, gene_name_); |
---|
215 | arb_species_name = gene_name+(gene_name_len+1); |
---|
216 | strcpy((char*)arb_species_name, arb_species_name_); |
---|
217 | arb_gene_name = arb_species_name+(arb_species_name_len+1); |
---|
218 | strcpy((char*)arb_gene_name, arb_gene_name_); |
---|
219 | } |
---|
220 | |
---|
221 | public: |
---|
222 | gene_struct(const char *gene_name_, const char *arb_species_name_, const char *arb_gene_name_) { |
---|
223 | init(gene_name_, arb_species_name_, arb_gene_name_); |
---|
224 | } |
---|
225 | gene_struct(const gene_struct& other) { |
---|
226 | if (&other != this) { |
---|
227 | init(other.get_internal_gene_name(), other.get_arb_species_name(), other.get_arb_gene_name()); |
---|
228 | } |
---|
229 | } |
---|
230 | gene_struct& operator = (const gene_struct& other) { |
---|
231 | if (&other != this) { |
---|
232 | delete [] gene_name; |
---|
233 | init(other.get_internal_gene_name(), other.get_arb_species_name(), other.get_arb_gene_name()); |
---|
234 | } |
---|
235 | return *this; |
---|
236 | } |
---|
237 | |
---|
238 | ~gene_struct() { |
---|
239 | delete [] gene_name; |
---|
240 | } |
---|
241 | |
---|
242 | const char *get_internal_gene_name() const { return gene_name; } |
---|
243 | const char *get_arb_species_name() const { return arb_species_name; } |
---|
244 | const char *get_arb_gene_name() const { return arb_gene_name; } |
---|
245 | }; |
---|
246 | |
---|
247 | extern int gene_flag; // if 'gene_flag' == 1 -> we are a gene pt server |
---|
248 | |
---|
249 | struct ltByArbName { |
---|
250 | bool operator()(const gene_struct *gs1, const gene_struct *gs2) const { |
---|
251 | int cmp = strcmp(gs1->get_arb_species_name(), gs2->get_arb_species_name()); |
---|
252 | if (cmp == 0) { cmp = strcmp(gs1->get_arb_gene_name(), gs2->get_arb_gene_name()); } |
---|
253 | return cmp<0; |
---|
254 | } |
---|
255 | }; |
---|
256 | struct ltByInternalName { |
---|
257 | bool operator()(const gene_struct *gs1, const gene_struct *gs2) const { |
---|
258 | int cmp = strcmp(gs1->get_internal_gene_name(), gs2->get_internal_gene_name()); |
---|
259 | return cmp<0; |
---|
260 | } |
---|
261 | }; |
---|
262 | |
---|
263 | typedef std::list<gene_struct> gene_struct_list; |
---|
264 | typedef std::set<const gene_struct *, ltByInternalName> gene_struct_index_internal; |
---|
265 | typedef std::set<const gene_struct *, ltByArbName> gene_struct_index_arb; |
---|
266 | |
---|
267 | extern gene_struct_list all_gene_structs; // stores all gene_structs |
---|
268 | extern gene_struct_index_arb gene_struct_arb2internal; // sorted by arb speces+gene name |
---|
269 | extern gene_struct_index_internal gene_struct_internal2arb; // sorted by internal name |
---|
270 | |
---|
271 | #define PT_base_string_counter_eof(str) (*(unsigned char *)(str) == 255) |
---|
272 | |
---|
273 | #else |
---|
274 | #error probe.h included twice |
---|
275 | #endif |
---|