From 7f5c0003a9135add5a2b6f14ab822e527b3df547 Mon Sep 17 00:00:00 2001
From: Martin Steinegger <themartinsteinegger@gmail.com>
Date: Mon, 9 Dec 2024 01:45:10 +0700
Subject: [PATCH] Fix memory issue

---
 src/linclust/kmermatcher.cpp      | 12 ++++++++----
 src/prefiltering/IndexBuilder.cpp | 18 ++++++++++--------
 src/util/makepaddedseqdb.cpp      |  3 +--
 src/util/masksequence.cpp         |  2 +-
 4 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/src/linclust/kmermatcher.cpp b/src/linclust/kmermatcher.cpp
index 5d898826..8bdf737e 100644
--- a/src/linclust/kmermatcher.cpp
+++ b/src/linclust/kmermatcher.cpp
@@ -63,10 +63,7 @@ std::pair<size_t, size_t> fillKmerPositionArray(KmerPosition<T> * kmerArray, siz
     size_t offset = 0;
     int querySeqType  =  seqDbr.getDbtype();
     size_t longestKmer = par.kmerSize;
-    Masker *masker = NULL;
-    if (par.maskMode == 1) {
-        masker = new Masker(*subMat);
-    }
+
 
     ScoreMatrix two;
     ScoreMatrix three;
@@ -85,6 +82,10 @@ std::pair<size_t, size_t> fillKmerPositionArray(KmerPosition<T> * kmerArray, siz
         unsigned short * scoreDist= new unsigned short[65536];
         unsigned int * hierarchicalScoreDist= new unsigned int[128];
 
+        Masker *masker = NULL;
+        if (par.maskMode == 1) {
+            masker = new Masker(*subMat);
+        }
         const int adjustedKmerSize = (par.adjustKmerLength) ? std::min( par.kmerSize+5, 23) :   par.kmerSize;
         Sequence seq(par.maxSeqLen, querySeqType, subMat, adjustedKmerSize, par.spacedKmer, false, true, par.spacedKmerPattern);
         KmerGenerator* generator;
@@ -336,6 +337,9 @@ std::pair<size_t, size_t> fillKmerPositionArray(KmerPosition<T> * kmerArray, siz
             if (thread_idx == 0) {
                 seqDbr.remapData();
             }
+            if (masker != NULL) {
+                delete masker;
+            }
 #pragma omp barrier
         }
 
diff --git a/src/prefiltering/IndexBuilder.cpp b/src/prefiltering/IndexBuilder.cpp
index fa2ff7df..81965a0b 100644
--- a/src/prefiltering/IndexBuilder.cpp
+++ b/src/prefiltering/IndexBuilder.cpp
@@ -81,11 +81,6 @@ void IndexBuilder::fillDatabase(IndexTable *indexTable, SequenceLookup **maskedL
         EXIT(EXIT_FAILURE);
     }
 
-    // need to prune low scoring k-mers through masking
-    Masker *masker = NULL;
-    if (maskedLookup != NULL) {
-        masker = new Masker(subMat);
-    }
 
     // identical scores for memory reduction code
     char *idScoreLookup = getScoreLookup(subMat);
@@ -99,6 +94,12 @@ void IndexBuilder::fillDatabase(IndexTable *indexTable, SequenceLookup **maskedL
 #ifdef OPENMP
         thread_idx = static_cast<unsigned int>(omp_get_thread_num());
 #endif
+        // need to prune low scoring k-mers through masking
+        Masker *masker = NULL;
+        if (maskedLookup != NULL) {
+            masker = new Masker(subMat);
+        }
+
 
         Indexer idxer(static_cast<unsigned int>(indexTable->getAlphabetSize()), seq->getKmerSize());
         Sequence s(seq->getMaxLen(), seq->getSeqType(), &subMat, seq->getKmerSize(), seq->isSpaced(), false, true, seq->getUserSpacedKmerPattern());
@@ -159,11 +160,12 @@ void IndexBuilder::fillDatabase(IndexTable *indexTable, SequenceLookup **maskedL
         if (generator != NULL) {
             delete generator;
         }
+        if(masker != NULL) {
+            delete masker;
+        }
     }
 
-    if(masker != NULL) {
-        delete masker;
-    }
+
 
     Debug(Debug::INFO) << "Index table: Masked residues: " << maskedResidues << "\n";
     if(totalKmerCount == 0) {
diff --git a/src/util/makepaddedseqdb.cpp b/src/util/makepaddedseqdb.cpp
index 16b2f8fb..2fd4641d 100644
--- a/src/util/makepaddedseqdb.cpp
+++ b/src/util/makepaddedseqdb.cpp
@@ -31,7 +31,6 @@ int makepaddedseqdb(int argc, const char **argv, const Command &command) {
     dbhw.open();
 
     // need to prune low scoring k-mers through masking
-    Masker masker(subMat);
 
     Debug::Progress progress(dbr.getSize());
 #pragma omp parallel
@@ -40,7 +39,7 @@ int makepaddedseqdb(int argc, const char **argv, const Command &command) {
 #ifdef OPENMP
     thread_idx = static_cast<unsigned int>(omp_get_thread_num());
 #endif
-
+    Masker masker(subMat);
     std::string result;
     result.reserve(par.maxSeqLen);
 
diff --git a/src/util/masksequence.cpp b/src/util/masksequence.cpp
index 012568b2..b1c8a836 100644
--- a/src/util/masksequence.cpp
+++ b/src/util/masksequence.cpp
@@ -29,7 +29,6 @@ int masksequence(int argc, const char **argv, const Command& command) {
     }
 
     // need to prune low scoring k-mers through masking
-    Masker masker(*subMat);
 
     DBWriter writer(par.db2.c_str(), par.db2Index.c_str(), par.threads, par.compressed, reader.getDbtype());
     writer.open();
@@ -40,6 +39,7 @@ int masksequence(int argc, const char **argv, const Command& command) {
 #ifdef OPENMP
         thread_idx = (unsigned int) omp_get_thread_num();
 #endif
+        Masker masker(*subMat);
         unsigned char *charSequence = new unsigned char[reader.getMaxSeqLen() + 1];
         Sequence seq(reader.getMaxSeqLen(), reader.getDbtype(), subMat,  0, false, false);
 #pragma omp for schedule(dynamic, 1)