delete tax test files

Signed-off-by: Radu Muntean <[email protected]>

.gitignore

@@ -2,7 +2,6 @@
 *.fai
 !metagraph/tests/data/*.fa
 !metagraph/tests/data/*.fai
-!metagraph/tests/data/taxonomic_data/*.fa
 metagraph/tests/data/*dump_test*
 projects/*/temp
 visualization/geolocation/data/*

metagraph/src/annotation/taxonomy/label_to_taxid.cpp

@@ -10,12 +10,11 @@ namespace annot {
 using mtg::common::logger;
 
 void TaxonomyBase::assign_label_type(const std::string &label, bool *require_accversion_to_taxid_map) {
-    if (utils::starts_with(label, ">gi|")) {
+    if (utils::starts_with(label, "gi|")) {
         // e.g.   >gi|1070643132|ref|NC_031224.1| Arthrobacter phage Mudcat, complete genome
         label_type = GEN_BANK;
         *require_accversion_to_taxid_map = true;
-    } else if (utils::starts_with(label, ">") &&
-                utils::starts_with(utils::split_string(label, ":")[1], "taxid|")) {
+    } else if (utils::starts_with(utils::split_string(label, ":")[1], "taxid|")) {
         // e.g.   >kraken:taxid|2016032|NC_047834.1 Alteromonas virus vB_AspP-H4/4, complete genome
         label_type = TAXID;
         *require_accversion_to_taxid_map = false;
@@ -71,8 +70,8 @@ void TaxonomyBase::read_accversion_to_taxid_map(const std::string &filepath,
 
     tsl::hopscotch_set<std::string> input_accessions;
     if (anno_matrix != NULL) {
-        for (const std::string &accversion : anno_matrix->get_annotation().get_all_labels()) {
-            input_accessions.insert(accversion);
+        for (const std::string &label : anno_matrix->get_annotation().get_all_labels()) {
+            input_accessions.insert(get_accession_version_from_label(label));
         }
     }
 

metagraph/src/annotation/taxonomy/tax_classifier.cpp

@@ -107,7 +107,7 @@ void TaxonomyClsAnno::read_tree(const std::string &tax_tree_filepath, ChildrenLi
             continue;
         }
 
-        if (considered_relevant_taxids.find(full_parents_list[taxid]) == considered_relevant_taxids.end()) {
+        if (not considered_relevant_taxids.count(full_parents_list[taxid])) {
             relevant_taxids.push_back(full_parents_list[taxid]);
             considered_relevant_taxids.insert(full_parents_list[taxid]);
         }
@@ -137,11 +137,13 @@ void TaxonomyClsAnno::dfs_statistics(const TaxId node,
     node_to_linearization_idx[node] = tree_linearization->size();
     tree_linearization->push_back(node);
     uint32_t depth = 0;
-    for (const TaxId &child : tree.at(node)) {
-        dfs_statistics(child, tree, tree_linearization);
-        tree_linearization->push_back(node);
-        if (node_depth[child] > depth) {
-            depth = node_depth[child];
+    if (tree.count(node)) {
+        for (const TaxId &child : tree.at(node)) {
+            dfs_statistics(child, tree, tree_linearization);
+            tree_linearization->push_back(node);
+            if (node_depth[child] > depth) {
+                depth = node_depth[child];
+            }
         }
     }
     node_depth[node] = depth + 1;
@@ -177,8 +179,26 @@ void TaxonomyClsAnno::rmq_preprocessing(const std::vector<TaxId> &tree_lineariza
     }
 }
 
-TaxId TaxonomyClsAnno::assign_class(const std::string &sequence) const {
-    throw std::runtime_error("Assign class not implemented. Received " + sequence);
+std::vector<TaxId> TaxonomyClsAnno::get_lca_taxids_for_seq(const std::string_view &sequence, bool reversed) const {
+    cerr << "Assign class not implemented reversed = " << reversed << "\n";
+    throw std::runtime_error("get_lca_taxids_for_seq TaxonomyClsAnno not implemented. Received seq size" + to_string(sequence.size()));
+    exit(0);
+}
+
+std::vector<TaxId> TaxonomyClsImportDB::get_lca_taxids_for_seq(const std::string_view &sequence, bool reversed) const {
+    cerr << "Assign class not implemented reversed = " << reversed << "\n";
+    throw std::runtime_error("get_lca_taxids_for_seq TaxonomyClsImportDB not implemented. Received seq size" + to_string(sequence.size()));
+    exit(0);
+}
+
+TaxId TaxonomyClsAnno::find_lca(const std::vector<TaxId> &taxids) const {
+    throw std::runtime_error("find_lca TaxonomyClsAnno not implemented. Received taxids size" + to_string(taxids.size()));
+    exit(0);
+}
+
+TaxId TaxonomyClsImportDB::find_lca(const std::vector<TaxId> &taxids) const {
+    throw std::runtime_error("find_lca TaxonomyClsImportDB not implemented. Received taxids size" + to_string(taxids.size()));
+    exit(0);
 }
 
 } // namespace annot

metagraph/src/annotation/taxonomy/tax_classifier.hpp

@@ -37,14 +37,12 @@ class TaxonomyBase {
 
     virtual ~TaxonomyBase() {};
 
-    // TODO implement
-    virtual TaxId assign_class(const std::string &sequence) const = 0;
+    TaxId assign_class(const std::string &sequence) const;
 
 PROTECTED_TESTABLE:
     void assign_label_type(const std::string &label, bool *require_accversion_to_taxid_map);
 
-    // TODO implement.
-    TaxId find_lca(const std::vector<TaxId> &taxids) const;
+    virtual TaxId find_lca(const std::vector<TaxId> &taxids) const = 0;
 
     std::string get_accession_version_from_label(const std::string &label) const;
 
@@ -57,7 +55,6 @@ class TaxonomyBase {
     */
     void read_accversion_to_taxid_map(const std::string &filepath, const graph::AnnotatedDBG *anno_matrix);
 
-    // TODO implement.
     /**
      * Update the current node_scores and best_lca by taking into account the weight of the start_node and all its ancestors.
      *
@@ -75,7 +72,13 @@ class TaxonomyBase {
                                tsl::hopscotch_map<TaxId, uint64_t> *node_scores,
                                tsl::hopscotch_set<TaxId> *nodes_already_propagated,
                                TaxId *best_lca,
-                               uint32_t *best_lca_dist_to_root);
+                               uint32_t *best_lca_dist_to_root) const;
+
+    /**
+     * Get the list of LCA taxid for each kmer in a given sequences.
+     * The sequence can be given in forward or in reversed orientation.
+     */
+    virtual std::vector<TaxId> get_lca_taxids_for_seq(const std::string_view &sequence, bool reversed) const = 0;
 
     LabelType label_type;
 
@@ -104,7 +107,10 @@ class TaxonomyClsImportDB : public TaxonomyBase {
     TaxonomyClsImportDB(const std::string &taxdb_filepath,
                         const double lca_coverage_rate,
                         const double kmers_discovery_rate);
-    TaxId assign_class(const std::string &sequence) const;
+
+PRIVATE_TESTABLE:
+    std::vector<TaxId> get_lca_taxids_for_seq(const std::string_view &sequence, bool reversed) const;
+    TaxId find_lca(const std::vector<TaxId> &taxids) const;
 };
 
 class TaxonomyClsAnno : public TaxonomyBase {
@@ -128,8 +134,7 @@ class TaxonomyClsAnno : public TaxonomyBase {
     // todo implement
     void export_taxdb(const std::string &filepath) const;
 
-    // todo implement
-    TaxId assign_class(const std::string &sequence) const;
+    TaxId assign_class_toplabels(const std::string &sequence, const double label_fraction) const;
 
 PRIVATE_TESTABLE:
     /**
@@ -162,6 +167,10 @@ class TaxonomyClsAnno : public TaxonomyBase {
                         const ChildrenList &tree,
                         std::vector<TaxId> *tree_linearization);
 
+    TaxId find_lca(const std::vector<TaxId> &taxids) const;
+
+    std::vector<TaxId> get_lca_taxids_for_seq(const std::string_view &sequence, bool reversed) const;
+
     /**
      * rmq_data[0] contains the taxonomic tree linearization
      *          (e.g. for root 1 and edges={1-2; 1-3}, the linearization is "1 2 1 3 1").

metagraph/tests/annotation/taxonomy/test_taxonomy.cpp

@@ -4,15 +4,13 @@
 
 #include <tsl/hopscotch_map.h>
 #include <tsl/hopscotch_set.h>
-#include <stdio.h>
-#include <stdlib.h>
 
 #include "annotation/taxonomy/tax_classifier.hpp"
 
 namespace mtg {
 namespace test {
 
-TEST (TaxonomyTest, DfsStatistics) {
+TEST (TaxonomyTest, ClsAnno_DfsStatistics) {
     mtg::annot::TaxonomyClsAnno *tax = new mtg::annot::TaxonomyClsAnno();
     tsl::hopscotch_map<uint32_t, std::vector<uint32_t>> tree {
                 {0, {1, 2, 3}},      // node 0 -> root
@@ -59,7 +57,7 @@ TEST (TaxonomyTest, DfsStatistics) {
     EXPECT_EQ(expected_node_to_linearization_idx, tax->node_to_linearization_idx);
 }
 
-TEST (TaxonomyTest, RmqPreprocessing) {
+TEST (TaxonomyTest, ClsAnno_RmqPreprocessing) {
     mtg::annot::TaxonomyClsAnno *tax = new mtg::annot::TaxonomyClsAnno();
 
     tax->node_depth = {