| 1 | // =============================================================== // |
|---|
| 2 | // // |
|---|
| 3 | // File : PRD_Design.hxx // |
|---|
| 4 | // Purpose : // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Wolfram Foerster in February 2001 // |
|---|
| 7 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 8 | // http://www.arb-home.de/ // |
|---|
| 9 | // // |
|---|
| 10 | // =============================================================== // |
|---|
| 11 | |
|---|
| 12 | #ifndef PRD_DESIGN_HXX |
|---|
| 13 | #define PRD_DESIGN_HXX |
|---|
| 14 | |
|---|
| 15 | #include "PRD_Range.hxx" |
|---|
| 16 | #include "PRD_Node.hxx" |
|---|
| 17 | #include "PRD_Pair.hxx" |
|---|
| 18 | |
|---|
| 19 | #include <deque> |
|---|
| 20 | |
|---|
| 21 | #ifndef ARBDB_H |
|---|
| 22 | #include <arbdb.h> |
|---|
| 23 | #endif |
|---|
| 24 | |
|---|
| 25 | #if defined(DEBUG) |
|---|
| 26 | // # define DUMP_PRIMER |
|---|
| 27 | #endif |
|---|
| 28 | |
|---|
| 29 | class PrimerDesign : virtual Noncopyable { |
|---|
| 30 | // zu untersuchendes Genom |
|---|
| 31 | // sequence of bases/spaces to be examined |
|---|
| 32 | |
|---|
| 33 | const char* sequence; |
|---|
| 34 | long int seqLength; |
|---|
| 35 | |
|---|
| 36 | // Laenge der Primer (Anzahl der enthaltenen Basen) |
|---|
| 37 | // count of bases of primers (min/max) |
|---|
| 38 | Range primer_length; |
|---|
| 39 | |
|---|
| 40 | // Position der Primer ( [min,max) ) |
|---|
| 41 | // left/right position of primers |
|---|
| 42 | |
|---|
| 43 | Range primer1; |
|---|
| 44 | Range primer2; |
|---|
| 45 | |
|---|
| 46 | // Abstand der Primer (bzgl. der Basen dazwischen, <=0 = ignorieren) |
|---|
| 47 | // min/max distance of primers (regarding number of bases between, <=0 = ignore) |
|---|
| 48 | |
|---|
| 49 | Range primer_distance; |
|---|
| 50 | |
|---|
| 51 | // maximale Anzahl zurueckzuliefernder Primerpaare |
|---|
| 52 | // max. count of returned primer pairs |
|---|
| 53 | |
|---|
| 54 | int max_count_primerpairs; |
|---|
| 55 | |
|---|
| 56 | // GC-Verhaeltniss der Primer in Prozent <=0 = ignorieren |
|---|
| 57 | // GC-ratio of primers in percent <=0 = ignore |
|---|
| 58 | // [ ratio = (G+C)/Basen ] |
|---|
| 59 | |
|---|
| 60 | Range GC_ratio; |
|---|
| 61 | |
|---|
| 62 | // Temperatur der Primer <=0 = ignorieren |
|---|
| 63 | // temperature of primers <=0 = ignore |
|---|
| 64 | // [ temperature = 4*(G+C) + 2*(A+T) ] |
|---|
| 65 | |
|---|
| 66 | Range temperature; |
|---|
| 67 | |
|---|
| 68 | // Faktoren zur Gewichtung von CG-Verhaeltniss und Temperatur bei der Bewertung von Primerpaaren |
|---|
| 69 | // factors to assess the GC-ratio and temperature at the evaluation of primerpairs |
|---|
| 70 | // [ 0.0 .. 1.0 ] |
|---|
| 71 | |
|---|
| 72 | double GC_factor; |
|---|
| 73 | double temperature_factor; |
|---|
| 74 | |
|---|
| 75 | // wird ein Primer ausserhalb dieser Distanz nocheinmal gefunden wird das Vorkommen ignoriert ( <= 0 = eindeutig ) |
|---|
| 76 | // is a primer found again out of this range its occurrence is ignored ( <=0 = explicit match ) |
|---|
| 77 | |
|---|
| 78 | int min_distance_to_next_match; |
|---|
| 79 | |
|---|
| 80 | // erweitern der IUPAC-Codes beim matching der sequence gegen die primerbaeume ? |
|---|
| 81 | // expand IUPAC-Codes while matching sequence vs. primertrees ? |
|---|
| 82 | |
|---|
| 83 | bool expand_IUPAC_Codes; |
|---|
| 84 | |
|---|
| 85 | |
|---|
| 86 | // |
|---|
| 87 | // INTERNAL STUFF |
|---|
| 88 | // |
|---|
| 89 | static const int FORWARD = 1; |
|---|
| 90 | static const int BACKWARD = -1; |
|---|
| 91 | |
|---|
| 92 | // primertrees |
|---|
| 93 | Node* root1; |
|---|
| 94 | Node* root2; |
|---|
| 95 | |
|---|
| 96 | // primerlists |
|---|
| 97 | Item* list1; |
|---|
| 98 | Item* list2; |
|---|
| 99 | |
|---|
| 100 | // primerpairs |
|---|
| 101 | Pair* pairs; |
|---|
| 102 | |
|---|
| 103 | GB_ERROR error; |
|---|
| 104 | |
|---|
| 105 | unsigned long int total_node_counter_left; |
|---|
| 106 | unsigned long int total_node_counter_right; |
|---|
| 107 | unsigned long int primer_node_counter_left; |
|---|
| 108 | unsigned long int primer_node_counter_right; |
|---|
| 109 | |
|---|
| 110 | void reset_node_counters() { |
|---|
| 111 | total_node_counter_left = 0; |
|---|
| 112 | total_node_counter_right = 0; |
|---|
| 113 | primer_node_counter_left = 0; |
|---|
| 114 | primer_node_counter_right = 0; |
|---|
| 115 | } |
|---|
| 116 | |
|---|
| 117 | public: |
|---|
| 118 | PrimerDesign(const char *sequence_, long int seqLength_, |
|---|
| 119 | Range pos1_, Range pos2_, Range length_, Range distance_, |
|---|
| 120 | Range ratio_, Range temperature_, int min_dist_to_next_, bool expand_IUPAC_Codes_, |
|---|
| 121 | int max_count_primerpairs_, double GC_factor_, double temp_factor_); |
|---|
| 122 | PrimerDesign(const char *sequence_, long int seqLength_, |
|---|
| 123 | Range pos1_, Range pos2_, Range length_, Range distance_, |
|---|
| 124 | int max_count_primerpairs_, double GC_factor_, double temp_factor_); |
|---|
| 125 | |
|---|
| 126 | PrimerDesign(const char *sequence_, long int seqLength_); |
|---|
| 127 | ~PrimerDesign(); |
|---|
| 128 | |
|---|
| 129 | void setPositionalParameters (Range pos1_, Range pos2_, Range length_, Range distance_); |
|---|
| 130 | void setConditionalParameters(Range ratio_, Range temperature_, int min_dist_to_next_, bool expand_IUPAC_Codes_, int max_count_primerpairs_, double GC_factor_, double temp_factor_); |
|---|
| 131 | |
|---|
| 132 | void buildPrimerTrees (); |
|---|
| 133 | void matchSequenceAgainstPrimerTrees (); |
|---|
| 134 | void convertTreesToLists (); |
|---|
| 135 | void evaluatePrimerPairs (); |
|---|
| 136 | |
|---|
| 137 | #if defined(DUMP_PRIMER) |
|---|
| 138 | void printPrimerTrees (); |
|---|
| 139 | void printPrimerLists (); |
|---|
| 140 | void printPrimerPairs (); |
|---|
| 141 | #endif |
|---|
| 142 | |
|---|
| 143 | void run(); |
|---|
| 144 | |
|---|
| 145 | GB_ERROR get_error() const { return error; } |
|---|
| 146 | |
|---|
| 147 | PRD_Sequence_Pos get_max_primer_length() const { return primer_length.max(); } |
|---|
| 148 | PRD_Sequence_Pos get_max_primer_pos() const { return primer2.max(); } |
|---|
| 149 | const char *get_result(int num, const char *&primers, int max_primer_length, int max_position_length, int max_length_length) const; // return 0 if no more results (primers is set to "leftPrimer,rightPrimer") |
|---|
| 150 | |
|---|
| 151 | public: |
|---|
| 152 | static const int PRINT_RAW_TREES = 1; |
|---|
| 153 | static const int PRINT_MATCHED_TREES = 2; |
|---|
| 154 | static const int PRINT_PRIMER_LISTS = 4; |
|---|
| 155 | static const int PRINT_PRIMER_PAIRS = 8; |
|---|
| 156 | |
|---|
| 157 | private: |
|---|
| 158 | void init (const char *sequence_, long int seqLength_, Range pos1_, Range pos2_, Range length_, Range distance_, Range ratio_, Range temperature_, int min_dist_to_next_, bool expand_IUPAC_Codes_, int max_count_primerpairs_, double GC_factor_, double temp_factor_); |
|---|
| 159 | PRD_Sequence_Pos followUp (Node *node_, std::deque<char> *primer_, int direction_); |
|---|
| 160 | void findNextPrimer (Node *start_at_, int depth_, int *counter_, int delivered_); |
|---|
| 161 | int insertNode (Node *current_, unsigned char base_, PRD_Sequence_Pos pos_, int delivered_, int offset_, int left_, int right_); |
|---|
| 162 | void clearTree (Node *start, int left_, int right_); |
|---|
| 163 | bool treeContainsPrimer (Node *start); |
|---|
| 164 | void calcGCandAT (int &GC_, int &AT_, Node *start_at_); |
|---|
| 165 | double evaluatePair (Item *one_, Item *two_); |
|---|
| 166 | void insertPair (double rating_, Item *one_, Item *two_); |
|---|
| 167 | }; |
|---|
| 168 | |
|---|
| 169 | #else |
|---|
| 170 | #error PRD_Design.hxx included twice |
|---|
| 171 | #endif // PRD_DESIGN_HXX |
|---|
