Merge pull request #1 from leejoey0921/linclust_integration

Make adjacent sequence matching configurable
ChunShow · Aug 30, 2024 · 5b68a8e · 5b68a8e
2 parents e12665b + ecea89f
commit 5b68a8e
Show file tree

Hide file tree

Showing 4 changed files with 332 additions and 139 deletions.
diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
@@ -152,6 +152,7 @@ Parameters::Parameters():
         PARAM_IGNORE_MULTI_KMER(PARAM_IGNORE_MULTI_KMER_ID, "--ignore-multi-kmer", "Skip repeating k-mers", "Skip k-mers occurring multiple times (>=2)", typeid(bool), (void *) &ignoreMultiKmer, "", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
         PARAM_HASH_SHIFT(PARAM_HASH_SHIFT_ID, "--hash-shift", "Shift hash", "Shift k-mer hash initialization", typeid(int), (void *) &hashShift, "^[1-9]{1}[0-9]*$", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
         PARAM_PICK_N_SIMILAR(PARAM_PICK_N_SIMILAR_ID, "--pick-n-sim-kmer", "Add N similar to search", "Add N similar k-mers to search", typeid(int), (void *) &pickNbest, "^[1-9]{1}[0-9]*$", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
+        PARAM_MATCH_ADJACENT_SEQ(PARAM_MATCH_ADJACENT_SEQ_ID, "--match-adjacent-seq", "Compare adjacent sequences to k-mers", "Compare sequence information adjacent to k-mers and elect multiple representative sequences per cluster", typeid(bool), (void *) &matchAdjacentSeq, "", MMseqsParameter::COMMAND_CLUSTLINEAR),
         PARAM_ADJUST_KMER_LEN(PARAM_ADJUST_KMER_LEN_ID, "--adjust-kmer-len", "Adjust k-mer length", "Adjust k-mer length based on specificity (only for nucleotides)", typeid(bool), (void *) &adjustKmerLength, "", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
         PARAM_RESULT_DIRECTION(PARAM_RESULT_DIRECTION_ID, "--result-direction", "Result direction", "result is 0: query, 1: target centric", typeid(int), (void *) &resultDirection, "^[0-1]{1}$", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
         PARAM_WEIGHT_FILE(PARAM_WEIGHT_FILE_ID, "--weights", "Weight file name", "Weights used for cluster priorization", typeid(std::string), (void*) &weightFile, "", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT ),
@@ -2513,6 +2514,8 @@ void Parameters::setDefaults() {
     resultDirection = Parameters::PARAM_RESULT_DIRECTION_TARGET;
     weightThr = 0.9;
     weightFile = "";
+    // TODO: change to true after fixing regression tests
+    matchAdjacentSeq = false;
     hashSeqBuffer = 1.05;
 
     // result2stats

diff --git a/src/commons/Parameters.h b/src/commons/Parameters.h
@@ -552,6 +552,7 @@ class Parameters {
     int resultDirection;
     float weightThr;
     std::string weightFile;
+    bool matchAdjacentSeq;
     float hashSeqBuffer;
 
     // indexdb
@@ -866,6 +867,7 @@ class Parameters {
     PARAMETER(PARAM_IGNORE_MULTI_KMER)
     PARAMETER(PARAM_HASH_SHIFT)
     PARAMETER(PARAM_PICK_N_SIMILAR)
+    PARAMETER(PARAM_MATCH_ADJACENT_SEQ)
     PARAMETER(PARAM_ADJUST_KMER_LEN)
     PARAMETER(PARAM_RESULT_DIRECTION)
     PARAMETER(PARAM_WEIGHT_FILE)