All combining various index subset modes for createindex

soedinglab · Aug 2, 2023 · 8fe3bf9 · 8fe3bf9
1 parent 91f2a6a
commit 8fe3bf9
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 5 deletions.
diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
@@ -188,7 +188,7 @@ Parameters::Parameters():
         // indexdb
         PARAM_CHECK_COMPATIBLE(PARAM_CHECK_COMPATIBLE_ID, "--check-compatible", "Check compatible", "0: Always recreate index, 1: Check if recreating index is needed, 2: Fail if index is incompatible", typeid(int), (void *) &checkCompatible, "^[0-2]{1}$", MMseqsParameter::COMMAND_MISC),
         PARAM_SEARCH_TYPE(PARAM_SEARCH_TYPE_ID, "--search-type", "Search type", "Search type 0: auto 1: amino acid, 2: translated, 3: nucleotide, 4: translated nucleotide alignment", typeid(int), (void *) &searchType, "^[0-4]{1}"),
-        PARAM_INDEX_SUBSET(PARAM_INDEX_SUBSET_ID, "--index-subset", "Index subset", "Create specialized index with subset of entries 0: normal index 1: index without headers 1: index without prefiltering data", typeid(int), (void *) &indexSubset, "^[0-2]{1}", MMseqsParameter::COMMAND_EXPERT),
+        PARAM_INDEX_SUBSET(PARAM_INDEX_SUBSET_ID, "--index-subset", "Index subset", "Create specialized index with subset of entries\n0: normal index\n1: index without headers\n2: index without prefiltering data\nFlags can be combined bit wise", typeid(int), (void *) &indexSubset, "^[0-3]{1}", MMseqsParameter::COMMAND_EXPERT),
         // createdb
         PARAM_USE_HEADER(PARAM_USE_HEADER_ID, "--use-fasta-header", "Use fasta header", "Use the id parsed from the fasta header as the index key instead of using incrementing numeric identifiers", typeid(bool), (void *) &useHeader, ""),
         PARAM_ID_OFFSET(PARAM_ID_OFFSET_ID, "--id-offset", "Offset of numeric ids", "Numeric ids in index file are offset by this value", typeid(int), (void *) &identifierOffset, "^(0|[1-9]{1}[0-9]*)$"),

diff --git a/src/prefiltering/PrefilteringIndexReader.cpp b/src/prefiltering/PrefilteringIndexReader.cpp
@@ -190,13 +190,14 @@ void PrefilteringIndexReader::createIndexFile(const std::string &outDB,
             (Parameters::isEqualDbtype(seqType, Parameters::DBTYPE_NUCLEOTIDES) || Parameters::isEqualDbtype(seqType, Parameters::DBTYPE_AMINO_ACIDS))
                 ? alphabetSize -1: alphabetSize;
 
-    if (indexSubset == Parameters::INDEX_SUBSET_NO_PREFILTER) {
+    const bool noPrefilter = (indexSubset & Parameters::INDEX_SUBSET_NO_PREFILTER) != 0;
+    if (noPrefilter) {
         splits = 0;
     }
 
     ScoreMatrix s3;
     ScoreMatrix s2;
-    if (Parameters::isEqualDbtype(seqType, Parameters::DBTYPE_HMM_PROFILE) == false && indexSubset != Parameters::INDEX_SUBSET_NO_PREFILTER) {
+    if (Parameters::isEqualDbtype(seqType, Parameters::DBTYPE_HMM_PROFILE) == false && noPrefilter == false) {
         int alphabetSize = subMat->alphabetSize;
         subMat->alphabetSize = subMat->alphabetSize-1;
         s3 = ExtendedSubstitutionMatrix::calcScoreMatrix(*subMat, 3);

diff --git a/src/util/indexdb.cpp b/src/util/indexdb.cpp
@@ -135,15 +135,16 @@ int indexdb(int argc, const char **argv, const Command &command) {
         }
     }
 
+    const bool noHeaders = (par.indexSubset & Parameters::INDEX_SUBSET_NO_HEADERS) != 0;
     if (recreate) {
         DBReader<unsigned int> *hdbr1 = NULL;
-        if (par.indexSubset != Parameters::INDEX_SUBSET_NO_HEADERS) {
+        if (noHeaders == false) {
             hdbr1 = new DBReader<unsigned int>(hdr1.c_str(), hdr1Index.c_str(), par.threads, DBReader<unsigned int>::USE_INDEX | DBReader<unsigned int>::USE_DATA);
             hdbr1->open(DBReader<unsigned int>::NOSORT);
         }
 
         DBReader<unsigned int> *hdbr2 = NULL;
-        if (sameDB == false && ppDB == false && par.indexSubset != Parameters::INDEX_SUBSET_NO_HEADERS) {
+        if (sameDB == false && ppDB == false && noHeaders == false) {
             hdbr2 = new DBReader<unsigned int>(par.hdr2.c_str(), par.hdr2Index.c_str(), par.threads, DBReader<unsigned int>::USE_INDEX | DBReader<unsigned int>::USE_DATA);
             hdbr2->open(DBReader<unsigned int>::NOSORT);
         }