feat: improve loadDB speeds

GenSpectrum · Aug 22, 2023 · 2b7cd7d · 2b7cd7d
1 parent d1e4b2b
commit 2b7cd7d
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 13 deletions.
diff --git a/include/silo/storage/sequence_store.h b/include/silo/storage/sequence_store.h
@@ -34,8 +34,8 @@ class NucPosition {
    template <class Archive>
    void serialize(Archive& archive, [[maybe_unused]] const uint32_t version) {
       // clang-format off
-      archive & symbol_whose_bitmap_is_flipped;
       archive & bitmaps;
+      archive & symbol_whose_bitmap_is_flipped;
       // clang-format on
    }
 
@@ -46,7 +46,7 @@ class NucPosition {
    explicit NucPosition(std::optional<NUCLEOTIDE_SYMBOL> symbol);
 
    SymbolMap<NUCLEOTIDE_SYMBOL, roaring::Roaring> bitmaps;
-   std::optional<NUCLEOTIDE_SYMBOL> symbol_whose_bitmap_is_flipped = std::nullopt;
+   std::optional<NUCLEOTIDE_SYMBOL> symbol_whose_bitmap_is_flipped;
 
    std::optional<silo::NUCLEOTIDE_SYMBOL> flipMostNumerousBitmap(uint32_t sequence_count);
 };
@@ -63,8 +63,8 @@ class SequenceStorePartition {
    template <class Archive>
    void serialize(Archive& archive, [[maybe_unused]] const uint32_t version) {
       // clang-format off
-      archive & positions;
       archive & indexing_differences_to_reference_genome;
+      archive & positions;
       archive & nucleotide_symbol_n_bitmaps;
       archive & sequence_count;
       // clang-format on

diff --git a/src/silo/storage/aa_store.cpp b/src/silo/storage/aa_store.cpp
@@ -56,15 +56,15 @@ std::optional<silo::AA_SYMBOL> silo::AAPosition::flipMostNumerousBitmap(uint32_t
 }
 
 silo::AAStorePartition::AAStorePartition(const std::vector<AA_SYMBOL>& reference_sequence)
-    : reference_sequence(reference_sequence) {
-   for (const AA_SYMBOL symbol : reference_sequence) {
-      positions.emplace_back(symbol);
-   }
-}
+    : reference_sequence(reference_sequence) {}
 
 size_t silo::AAStorePartition::fill(silo::ZstdFastaReader& input_file) {
    static constexpr size_t BUFFER_SIZE = 1024;
 
+   for (const AA_SYMBOL symbol : reference_sequence) {
+      positions.emplace_back(symbol);
+   }
+
    size_t read_sequences_count = 0;
 
    std::vector<std::string> sequence_buffer;

diff --git a/src/silo/storage/sequence_store.cpp b/src/silo/storage/sequence_store.cpp
@@ -61,15 +61,15 @@ std::optional<silo::NUCLEOTIDE_SYMBOL> silo::NucPosition::flipMostNumerousBitmap
 silo::SequenceStorePartition::SequenceStorePartition(
    const std::vector<NUCLEOTIDE_SYMBOL>& reference_genome
 )
-    : reference_genome(reference_genome) {
-   for (const NUCLEOTIDE_SYMBOL symbol : reference_genome) {
-      positions.emplace_back(symbol);
-   }
-}
+    : reference_genome(reference_genome) {}
 
 size_t silo::SequenceStorePartition::fill(silo::ZstdFastaReader& input_file) {
    static constexpr size_t BUFFER_SIZE = 1024;
 
+   for (const NUCLEOTIDE_SYMBOL symbol : reference_genome) {
+      positions.emplace_back(symbol);
+   }
+
    size_t read_sequences_count = 0;
 
    std::vector<std::string> genome_buffer;

diff --git a/src/silo_api/database_directory_watcher.cpp b/src/silo_api/database_directory_watcher.cpp
@@ -126,4 +126,8 @@ void silo_api::DatabaseDirectoryWatcher::checkDirectoryForData(Poco::Timer& /*ti
 
    SPDLOG_INFO("New data version detected: {}", most_recent_database_state->first.string());
    database_mutex.setDatabase(silo::Database::loadDatabaseState(most_recent_database_state->first));
+   SPDLOG_INFO(
+      "New database with version {} successfully loaded.",
+      most_recent_database_state->first.string()
+   );
 }