Merge branch 'linclust_integration' of https://github.com/leejoey0921…

…/MMseqs2
martin-steinegger · Dec 14, 2024 · 738ca2f · 738ca2f
2 parents 7f5c000 + ecea89f
commit 738ca2f
Show file tree

Hide file tree

Showing 4 changed files with 434 additions and 132 deletions.
diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
@@ -157,7 +157,7 @@ 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_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_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_RESULT_DIRECTION(PARAM_RESULT_DIRECTION_ID, "--result-direction", "Result direction", "result is 0: query, 1: target centric", typeid(int), (void *) &resultDirection, "^[0-1]{1}$", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT),
         PARAM_WEIGHT_FILE(PARAM_WEIGHT_FILE_ID, "--weights", "Weight file name", "Weights used for cluster priorization", typeid(std::string), (void*) &weightFile, "", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT ),
         PARAM_WEIGHT_THR(PARAM_WEIGHT_THR_ID, "--cluster-weight-threshold", "Cluster Weight threshold", "Weight threshold used for cluster priorization", typeid(float), (void*) &weightThr, "^[0-9]*(\\.[0-9]+)?$", MMseqsParameter::COMMAND_CLUSTLINEAR | MMseqsParameter::COMMAND_EXPERT ),
@@ -811,7 +811,6 @@ Parameters::Parameters():
     kmerindexdb.push_back(&PARAM_KMER_PER_SEQ);
     kmerindexdb.push_back(&PARAM_KMER_PER_SEQ_SCALE);
     kmerindexdb.push_back(&PARAM_MIN_SEQ_ID);
-    kmerindexdb.push_back(&PARAM_ADJUST_KMER_LEN);
     kmerindexdb.push_back(&PARAM_SPLIT_MEMORY_LIMIT);
     kmerindexdb.push_back(&PARAM_IGNORE_MULTI_KMER);
     kmerindexdb.push_back(&PARAM_ALPH_SIZE);
@@ -993,7 +992,6 @@ Parameters::Parameters():
     kmermatcher.push_back(&PARAM_SPACED_KMER_MODE);
     kmermatcher.push_back(&PARAM_SPACED_KMER_PATTERN);
     kmermatcher.push_back(&PARAM_KMER_PER_SEQ_SCALE);
-    kmermatcher.push_back(&PARAM_ADJUST_KMER_LEN);
     kmermatcher.push_back(&PARAM_MASK_RESIDUES);
     kmermatcher.push_back(&PARAM_MASK_PROBABILTY);
     kmermatcher.push_back(&PARAM_MASK_LOWER_CASE);
@@ -2580,10 +2578,12 @@ void Parameters::setDefaults() {
     ignoreMultiKmer = false;
     hashShift = 67;
     pickNbest = 1;
-    adjustKmerLength = false;
     resultDirection = Parameters::PARAM_RESULT_DIRECTION_TARGET;
     weightThr = 0.9;
     weightFile = "";
+    // TODO: change to true after fixing regression tests
+    matchAdjacentSeq = false;
+    hashSeqBuffer = 1.05;
 
     // result2stats
     stat = "";

diff --git a/src/commons/Parameters.h b/src/commons/Parameters.h
@@ -559,10 +559,11 @@ class Parameters {
     bool ignoreMultiKmer;
     int hashShift;
     int pickNbest;
-    int adjustKmerLength;
     int resultDirection;
     float weightThr;
     std::string weightFile;
+    bool matchAdjacentSeq;
+    float hashSeqBuffer;
 
     // indexdb
     int checkCompatible;
@@ -880,7 +881,7 @@ class Parameters {
     PARAMETER(PARAM_IGNORE_MULTI_KMER)
     PARAMETER(PARAM_HASH_SHIFT)
     PARAMETER(PARAM_PICK_N_SIMILAR)
-    PARAMETER(PARAM_ADJUST_KMER_LEN)
+    PARAMETER(PARAM_MATCH_ADJACENT_SEQ)
     PARAMETER(PARAM_RESULT_DIRECTION)
     PARAMETER(PARAM_WEIGHT_FILE)
     PARAMETER(PARAM_WEIGHT_THR)