| 1 | // =============================================================== // |
|---|
| 2 | // // |
|---|
| 3 | // File : ARB_Tree.hxx // |
|---|
| 4 | // Purpose : Tree types with sequence knowledge // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Ralf Westram (coder@reallysoft.de) in October 2009 // |
|---|
| 7 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 8 | // http://www.arb-home.de/ // |
|---|
| 9 | // // |
|---|
| 10 | // =============================================================== // |
|---|
| 11 | |
|---|
| 12 | #ifndef ARB_TREE_HXX |
|---|
| 13 | #define ARB_TREE_HXX |
|---|
| 14 | |
|---|
| 15 | #ifndef ALIVIEW_HXX |
|---|
| 16 | #include <AliView.hxx> |
|---|
| 17 | #endif |
|---|
| 18 | #ifndef DOWNCAST_H |
|---|
| 19 | #include <downcast.h> |
|---|
| 20 | #endif |
|---|
| 21 | #ifndef ROOTEDTREE_H |
|---|
| 22 | #include <RootedTree.h> |
|---|
| 23 | #endif |
|---|
| 24 | |
|---|
| 25 | #define at_assert(cond) arb_assert(cond) |
|---|
| 26 | |
|---|
| 27 | typedef void (*ARB_tree_node_del_cb)(GBDATA*, class ARB_seqtree*); |
|---|
| 28 | |
|---|
| 29 | class ARB_seqtree_root; |
|---|
| 30 | class ARB_seqtree; |
|---|
| 31 | class ARB_edge; |
|---|
| 32 | class AP_weights; |
|---|
| 33 | class AP_sequence; |
|---|
| 34 | class AliView; |
|---|
| 35 | |
|---|
| 36 | class ARB_seqtree_root : public TreeRoot { // derived from Noncopyable |
|---|
| 37 | AliView *ali; |
|---|
| 38 | |
|---|
| 39 | AP_sequence *seqTemplate; |
|---|
| 40 | |
|---|
| 41 | // following variables are set, if the tree has been loaded from DB |
|---|
| 42 | char *tree_name; // name of the tree in DB |
|---|
| 43 | GBDATA *gb_tree; // tree container in DB |
|---|
| 44 | |
|---|
| 45 | bool isLinkedToDB; |
|---|
| 46 | bool addDeleteCallbacks; |
|---|
| 47 | |
|---|
| 48 | protected: |
|---|
| 49 | void set_gb_tree(GBDATA *gbTree) { |
|---|
| 50 | at_assert(implicated(gb_tree, gb_tree == gbTree)); |
|---|
| 51 | at_assert(tree_name); |
|---|
| 52 | gb_tree = gbTree; |
|---|
| 53 | } |
|---|
| 54 | void set_gb_tree_and_name(GBDATA *gbTree, const char *name) { |
|---|
| 55 | at_assert(!gb_tree); |
|---|
| 56 | at_assert(!tree_name); |
|---|
| 57 | gb_tree = gbTree; |
|---|
| 58 | tree_name = strdup(name); |
|---|
| 59 | } |
|---|
| 60 | |
|---|
| 61 | public: |
|---|
| 62 | ARB_seqtree_root(AliView *aliView, RootedTreeNodeFactory *nodeMaker_, AP_sequence *seqTempl, bool add_delete_callbacks); |
|---|
| 63 | ~ARB_seqtree_root() OVERRIDE; |
|---|
| 64 | |
|---|
| 65 | DEFINE_TREE_ROOT_ACCESSORS(ARB_seqtree_root, ARB_seqtree); |
|---|
| 66 | |
|---|
| 67 | const AliView *get_aliview() const { return ali; } |
|---|
| 68 | |
|---|
| 69 | const AP_filter *get_filter() const { return ali->get_filter(); } |
|---|
| 70 | const AP_weights *get_weights() const { return ali->get_weights(); } |
|---|
| 71 | |
|---|
| 72 | GBDATA *get_gb_main() const { return ali->get_gb_main(); } |
|---|
| 73 | |
|---|
| 74 | GBDATA *get_gb_tree() const { return gb_tree; } // NULL if no tree loaded (or tree disappeared from DB) |
|---|
| 75 | void tree_deleted_cb(GBDATA *gb_tree_del); // internal |
|---|
| 76 | |
|---|
| 77 | const char *get_tree_name() const { return tree_name; } // NULL if no tree loaded (or tree disappeared from DB) |
|---|
| 78 | |
|---|
| 79 | virtual GB_ERROR loadFromDB(const char *name) __ATTR__USERESULT; |
|---|
| 80 | virtual GB_ERROR saveToDB() __ATTR__USERESULT; |
|---|
| 81 | |
|---|
| 82 | const AP_sequence *get_seqTemplate() const { return seqTemplate; } |
|---|
| 83 | |
|---|
| 84 | GB_ERROR linkToDB(int *zombies, int *duplicates) __ATTR__USERESULT; |
|---|
| 85 | void unlinkFromDB(); // @@@ is (but should not be) unused |
|---|
| 86 | }; |
|---|
| 87 | |
|---|
| 88 | |
|---|
| 89 | struct ARB_tree_info { |
|---|
| 90 | size_t leafs; |
|---|
| 91 | size_t innerNodes; |
|---|
| 92 | size_t groups; |
|---|
| 93 | size_t unlinked; // unlinked leafs (same as 'leafs' if tree is unlinked) |
|---|
| 94 | size_t marked; // leafs linked to marked species |
|---|
| 95 | // size_t with_sequence; // @@@ add when AP_sequence is member of ARB_seqtree |
|---|
| 96 | |
|---|
| 97 | ARB_tree_info(); |
|---|
| 98 | |
|---|
| 99 | size_t nodes() const { return innerNodes+leafs; } |
|---|
| 100 | size_t linked() const { return leafs-unlinked; } |
|---|
| 101 | size_t unmarked() const { return linked()-marked; } |
|---|
| 102 | }; |
|---|
| 103 | |
|---|
| 104 | |
|---|
| 105 | class ARB_seqtree : public RootedTree { // derived from Noncopyable |
|---|
| 106 | friend GB_ERROR ARB_seqtree_root::loadFromDB(const char *name); |
|---|
| 107 | friend GB_ERROR ARB_seqtree_root::linkToDB(int *zombies, int *duplicates); |
|---|
| 108 | friend void ARB_seqtree_root::unlinkFromDB(); |
|---|
| 109 | |
|---|
| 110 | AP_sequence *seq; /* NULL if tree is unlinked |
|---|
| 111 | * otherwise automatically valid for leafs with gb_node! |
|---|
| 112 | * may be set manually for inner nodes |
|---|
| 113 | */ |
|---|
| 114 | |
|---|
| 115 | // ------------------ |
|---|
| 116 | // functions |
|---|
| 117 | |
|---|
| 118 | void unloadSequences(); |
|---|
| 119 | GB_ERROR preloadLeafSequences(); |
|---|
| 120 | |
|---|
| 121 | GB_ERROR add_delete_cb_rec(ARB_tree_node_del_cb cb) __ATTR__USERESULT; |
|---|
| 122 | void remove_delete_cb_rec(ARB_tree_node_del_cb cb); |
|---|
| 123 | |
|---|
| 124 | protected: |
|---|
| 125 | AP_sequence *take_seq() { // afterwards not has no seq and caller is responsible for sequence |
|---|
| 126 | AP_sequence *result = seq; |
|---|
| 127 | seq = NULL; |
|---|
| 128 | return result; |
|---|
| 129 | } |
|---|
| 130 | void replace_seq(AP_sequence *sequence); |
|---|
| 131 | |
|---|
| 132 | public: |
|---|
| 133 | ARB_seqtree(ARB_seqtree_root *root) |
|---|
| 134 | : RootedTree(root), |
|---|
| 135 | seq(NULL) |
|---|
| 136 | {} |
|---|
| 137 | ~ARB_seqtree() OVERRIDE; |
|---|
| 138 | |
|---|
| 139 | DEFINE_TREE_ACCESSORS(ARB_seqtree_root, ARB_seqtree); |
|---|
| 140 | |
|---|
| 141 | void calcTreeInfo(ARB_tree_info& info); |
|---|
| 142 | |
|---|
| 143 | // order in dendogram: |
|---|
| 144 | bool is_upper_son() const { return is_leftson(); } |
|---|
| 145 | bool is_lower_son() const { return is_rightson(); } |
|---|
| 146 | |
|---|
| 147 | AP_sequence *get_seq() { return seq; } |
|---|
| 148 | const AP_sequence *get_seq() const { return seq; } |
|---|
| 149 | AP_sequence * set_seq(AP_sequence *sequence) { |
|---|
| 150 | at_assert(!seq); // already set |
|---|
| 151 | #ifndef UNIT_TESTS // UT_DIFF |
|---|
| 152 | // unit tests are allowed to leave sequence undefined // @@@ better solution? |
|---|
| 153 | at_assert(sequence); |
|---|
| 154 | #endif |
|---|
| 155 | seq = sequence; |
|---|
| 156 | return seq; |
|---|
| 157 | } |
|---|
| 158 | |
|---|
| 159 | void mark_subtree(); |
|---|
| 160 | bool contains_marked_species(); |
|---|
| 161 | |
|---|
| 162 | }; |
|---|
| 163 | |
|---|
| 164 | struct ARB_tree_predicate { |
|---|
| 165 | virtual ~ARB_tree_predicate() {} |
|---|
| 166 | virtual bool selects(const ARB_seqtree& tree) const = 0; |
|---|
| 167 | }; |
|---|
| 168 | |
|---|
| 169 | // ------------------------ |
|---|
| 170 | // ARB_countedTree |
|---|
| 171 | // tree that knows its size |
|---|
| 172 | |
|---|
| 173 | struct ARB_countedTree : public ARB_seqtree { |
|---|
| 174 | explicit ARB_countedTree(ARB_seqtree_root *tree_root_) |
|---|
| 175 | : ARB_seqtree(tree_root_) |
|---|
| 176 | {} |
|---|
| 177 | DEFINE_TREE_ACCESSORS(ARB_seqtree_root, ARB_countedTree); |
|---|
| 178 | |
|---|
| 179 | // @@@ TODO: |
|---|
| 180 | // - add debug code (checking init_tree() is called exactly once) |
|---|
| 181 | // - init_tree() might be called automatically via announce_tree_constructed() |
|---|
| 182 | |
|---|
| 183 | virtual void init_tree() = 0; /* impl. shall initialize the tree |
|---|
| 184 | * (including some kind of leaf counter) |
|---|
| 185 | * needs to be called manually */ |
|---|
| 186 | |
|---|
| 187 | void compute_tree() OVERRIDE {} // ARB_countedTree always is informed about its subtrees |
|---|
| 188 | |
|---|
| 189 | size_t relative_position_in(const ARB_countedTree *upgroup) const; |
|---|
| 190 | }; |
|---|
| 191 | |
|---|
| 192 | |
|---|
| 193 | |
|---|
| 194 | #else |
|---|
| 195 | #error ARB_Tree.hxx included twice |
|---|
| 196 | #endif // ARB_TREE_HXX |
|---|