createsubdb and view can now return results from identifiers in .look…

…up with --id-mode 1
soedinglab · Apr 1, 2021 · 14a3dce · 14a3dce
1 parent 6622c9f
commit 14a3dce
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 4 deletions.
diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
@@ -272,6 +272,7 @@ Parameters::Parameters():
         PARAM_TAX_OUTPUT_MODE(PARAM_TAX_OUTPUT_MODE_ID, "--tax-output-mode", "Taxonomy output mode", "0: output LCA, 1: output alignment 2: output both", typeid(int), (void *) &taxonomyOutputMode, "^[0-2]{1}$"),
         // createsubdb, filtertaxseqdb
         PARAM_SUBDB_MODE(PARAM_SUBDB_MODE_ID, "--subdb-mode", "Subdb mode", "Subdb mode 0: copy data 1: soft link data and write index", typeid(int), (void *) &subDbMode, "^[0-1]{1}$"),
+        PARAM_ID_MODE(PARAM_ID_MODE_ID, "--id-mode", "Database ID mode", "Select DB entries based on 0: database keys, 1: FASTA identifiers (.lookup)", typeid(int), (void *) &dbIdMode, "^[0-1]{1}$"),
         PARAM_TAR_INCLUDE(PARAM_TAR_INCLUDE_ID, "--tar-include", "Tar Inclusion Regex", "Include file names based on this regex", typeid(std::string), (void *) &tarInclude, "^.*$"),
         PARAM_TAR_EXCLUDE(PARAM_TAR_EXCLUDE_ID, "--tar-exclude", "Tar Exclusion Regex", "Exclude file names based on this regex", typeid(std::string), (void *) &tarExclude, "^.*$"),
         // unpackdb
@@ -1030,6 +1031,7 @@ Parameters::Parameters():
 
     // createsubdb
     createsubdb.push_back(&PARAM_SUBDB_MODE);
+    createsubdb.push_back(&PARAM_ID_MODE);
     createsubdb.push_back(&PARAM_V);
 
     // renamedbkeys
@@ -1060,6 +1062,7 @@ Parameters::Parameters():
 
     // view
     view.push_back(&PARAM_ID_LIST);
+    view.push_back(&PARAM_ID_MODE);
     view.push_back(&PARAM_IDX_ENTRY_TYPE);
     view.push_back(&PARAM_V);
 
@@ -2344,6 +2347,7 @@ void Parameters::setDefaults() {
 
     // createsubdb
     subDbMode = Parameters::SUBDB_MODE_HARD;
+    dbIdMode = Parameters::ID_MODE_KEYS;
 
     // tar2db
     tarInclude = ".*";

diff --git a/src/commons/Parameters.h b/src/commons/Parameters.h
@@ -280,6 +280,9 @@ class Parameters {
     static const int SUBDB_MODE_HARD = 0;
     static const int SUBDB_MODE_SOFT = 1;
 
+    static const int ID_MODE_KEYS = 0;
+    static const int ID_MODE_LOOKUP = 1;
+
     // unpackdb
     static const int UNPACK_NAME_KEY = 0;
     static const int UNPACK_NAME_ACCESSION = 1;
@@ -635,6 +638,7 @@ class Parameters {
 
     // createsubdb
     int subDbMode;
+    int dbIdMode;
 
     // tar2db
     std::string tarInclude;
@@ -959,6 +963,7 @@ class Parameters {
 
     // createsubdb
     PARAMETER(PARAM_SUBDB_MODE)
+    PARAMETER(PARAM_ID_MODE)
 
     // tar2db
     PARAMETER(PARAM_TAR_INCLUDE)

diff --git a/src/util/createsubdb.cpp b/src/util/createsubdb.cpp
@@ -23,7 +23,12 @@ int createsubdb(int argc, const char **argv, const Command& command) {
         }
     }
 
-    DBReader<unsigned int> reader(par.db2.c_str(), par.db2Index.c_str(), 1, DBReader<unsigned int>::USE_INDEX|DBReader<unsigned int>::USE_DATA);
+    const bool lookupMode = par.dbIdMode == Parameters::ID_MODE_LOOKUP;
+    int dbMode = DBReader<unsigned int>::USE_INDEX|DBReader<unsigned int>::USE_DATA;
+    if (lookupMode) {
+        dbMode |= DBReader<unsigned int>::USE_LOOKUP_REV;
+    }
+    DBReader<unsigned int> reader(par.db2.c_str(), par.db2Index.c_str(), 1, dbMode);
     reader.open(DBReader<unsigned int>::NOSORT);
     const bool isCompressed = reader.isCompressed();
 
@@ -37,7 +42,18 @@ int createsubdb(int argc, const char **argv, const Command& command) {
     bool isOrdered = true;
     while (getline(&line, &len, orderFile) != -1) {
         Util::parseKey(line, dbKey);
-        const unsigned int key = Util::fast_atoi<unsigned int>(dbKey);
+        unsigned int key;
+        if (lookupMode) {
+            size_t lookupId = reader.getLookupIdByAccession(dbKey);
+            if (lookupId == SIZE_MAX) {
+                Debug(Debug::WARNING) << "Could not find name " << dbKey << " in lookup\n";
+                continue;
+            }
+            key = reader.getLookupKey(lookupId);
+        } else {
+            key = Util::fast_atoi<unsigned int>(dbKey);
+        }
+
         isOrdered &= (prevKey <= key);
         prevKey = key;
         const size_t id = reader.getId(key);

diff --git a/src/util/view.cpp b/src/util/view.cpp
@@ -22,9 +22,26 @@ int view(int argc, const char **argv, const Command& command) {
             indexSrcType = IndexReader::SRC_HEADERS;
             break;
     }
-    IndexReader reader(par.db1, par.threads, indexSrcType, 0);
+    const bool lookupMode = par.dbIdMode == Parameters::ID_MODE_LOOKUP;
+    int dbMode = DBReader<unsigned int>::USE_INDEX|DBReader<unsigned int>::USE_DATA;
+    if (lookupMode) {
+        dbMode |= DBReader<unsigned int>::USE_LOOKUP_REV;
+    }
+    IndexReader reader(par.db1, par.threads, indexSrcType, false, dbMode);
     for (size_t i = 0; i < ids.size(); ++i) {
-        const unsigned int key = Util::fast_atoi<unsigned int>(ids[i].c_str());
+        unsigned int key;
+        std::string& ref = ids[i];
+        if (lookupMode) {
+            size_t lookupId = reader.sequenceReader->getLookupIdByAccession(ref);
+            if (lookupId == SIZE_MAX) {
+                Debug(Debug::WARNING) << "Could not find " << ref << " in lookup\n";
+                continue;
+            }
+            key = reader.sequenceReader->getLookupKey(lookupId);
+        } else {
+            key = Util::fast_atoi<unsigned int>(ref.c_str());
+        }
+
         const size_t id = reader.sequenceReader->getId(key);
         if (id >= UINT_MAX) {
             Debug(Debug::ERROR) << "Key " << ids[i] << " not found in database\n";