From d99d79c15463abfe260d5f402493355248262222 Mon Sep 17 00:00:00 2001
From: Woosub-Kim <woosubgo@gmail.com>
Date: Tue, 10 Sep 2024 22:52:31 +0900
Subject: [PATCH 1/2] single chain allowing multimersearch

---
 src/strucclustutils/scoremultimer.cpp | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/src/strucclustutils/scoremultimer.cpp b/src/strucclustutils/scoremultimer.cpp
index 9679013e..03946144 100644
--- a/src/strucclustutils/scoremultimer.cpp
+++ b/src/strucclustutils/scoremultimer.cpp
@@ -181,7 +181,7 @@ class DBSCANCluster {
 public:
     DBSCANCluster(SearchResult &searchResult, std::set<cluster_t> &finalClusters, double minCov) : searchResult(searchResult), finalClusters(finalClusters) {
         cLabel = 0;
-        minimumClusterSize = std::max(MULTIPLE_CHAINED_COMPLEX, (unsigned int) ((double) searchResult.qChainKeys.size() * minCov));
+        minimumClusterSize = (unsigned int) ((double) searchResult.qChainKeys.size() * minCov);
         maximumClusterSize = std::min(searchResult.qChainKeys.size(), searchResult.dbChainKeys.size());
         maximumClusterNum = searchResult.alnVec.size() / maximumClusterSize;
         prevMaxClusterSize = 0;
@@ -191,6 +191,9 @@ class DBSCANCluster {
     }
 
     bool getAlnClusters() {
+        if (searchResult.qChainKeys.size() < MULTIPLE_CHAINED_COMPLEX || searchResult.dbChainKeys.size() < MULTIPLE_CHAINED_COMPLEX)
+            return earlyStopForSingleChainComplex();
+
         // rbh filter
         filterAlnsByRBH();
         fillDistMatrix();
@@ -225,6 +228,16 @@ class DBSCANCluster {
     std::map<unsigned int, float> qBestTmScore;
     std::map<unsigned int, float> dbBestTmScore;
 
+    bool earlyStopForSingleChainComplex() {
+        if (minimumClusterSize >= MULTIPLE_CHAINED_COMPLEX)
+            return finishDBSCAN();
+
+        for (size_t alnIdx = 0; alnIdx < searchResult.alnVec.size(); alnIdx++ ) {
+            finalClusters.insert(cluster_t(alnIdx));
+        }
+        return finishDBSCAN();
+    }
+
     bool runDBSCAN() {
         unsigned int neighborIdx;
         unsigned int neighborAlnIdx;
@@ -519,7 +532,7 @@ class ComplexScorer {
                 continue;
             }
             paredSearchResult.standardize();
-            if (!paredSearchResult.alnVec.empty() && currDbChainKeys.size() >= MULTIPLE_CHAINED_COMPLEX)
+            if (!paredSearchResult.alnVec.empty())
                 searchResults.emplace_back(paredSearchResult);
 
             paredSearchResult.alnVec.clear();
@@ -531,7 +544,7 @@ class ComplexScorer {
         }
         currAlns.clear();
         paredSearchResult.standardize();
-        if (!paredSearchResult.alnVec.empty() && currDbChainKeys.size() >= MULTIPLE_CHAINED_COMPLEX)
+        if (!paredSearchResult.alnVec.empty())
             searchResults.emplace_back(paredSearchResult);
 
         paredSearchResult.alnVec.clear();
@@ -714,8 +727,6 @@ int scoremultimer(int argc, const char **argv, const Command &command) {
         for (size_t qCompIdx = 0; qCompIdx < qComplexIndices.size(); qCompIdx++) {
             unsigned int qComplexId = qComplexIndices[qCompIdx];
             std::vector<unsigned int> &qChainKeys = qComplexIdToChainKeysMap.at(qComplexId);
-            if (qChainKeys.size() < MULTIPLE_CHAINED_COMPLEX)
-                continue;
             complexScorer.getSearchResults(qComplexId, qChainKeys, dbChainKeyToComplexIdMap, dbComplexIdToChainKeysMap, searchResults);
             // for each db complex
             for (size_t dbId = 0; dbId < searchResults.size(); dbId++) {

From 9a2b7da48eef75c56104293e68a754ac78a10cbb Mon Sep 17 00:00:00 2001
From: Woosub-Kim <woosubgo@gmail.com>
Date: Wed, 11 Sep 2024 03:04:56 +0900
Subject: [PATCH 2/2] bug fixed: single elemented vector for single chain
 alignment

---
 src/strucclustutils/scoremultimer.cpp | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/src/strucclustutils/scoremultimer.cpp b/src/strucclustutils/scoremultimer.cpp
index 03946144..7e9e514d 100644
--- a/src/strucclustutils/scoremultimer.cpp
+++ b/src/strucclustutils/scoremultimer.cpp
@@ -191,7 +191,8 @@ class DBSCANCluster {
     }
 
     bool getAlnClusters() {
-        if (searchResult.qChainKeys.size() < MULTIPLE_CHAINED_COMPLEX || searchResult.dbChainKeys.size() < MULTIPLE_CHAINED_COMPLEX)
+        // if Query or Target is a Single Chain Complex.
+        if (std::min(searchResult.qChainKeys.size(), searchResult.dbChainKeys.size()) < MULTIPLE_CHAINED_COMPLEX)
             return earlyStopForSingleChainComplex();
 
         // rbh filter
@@ -232,8 +233,9 @@ class DBSCANCluster {
         if (minimumClusterSize >= MULTIPLE_CHAINED_COMPLEX)
             return finishDBSCAN();
 
-        for (size_t alnIdx = 0; alnIdx < searchResult.alnVec.size(); alnIdx++ ) {
-            finalClusters.insert(cluster_t(alnIdx));
+        for (unsigned int alnIdx = 0; alnIdx < searchResult.alnVec.size(); alnIdx++ ) {
+            neighbors = {alnIdx};
+            finalClusters.insert(neighbors);
         }
         return finishDBSCAN();
     }
@@ -375,6 +377,7 @@ class DBSCANCluster {
         // Too few alns => do nothing and finish it
         if (searchResult.alnVec.size() < minimumClusterSize)
             return finishDBSCAN();
+        // All alns as a cluster
         for (size_t alnIdx=0; alnIdx<searchResult.alnVec.size(); alnIdx++) {
             neighbors.emplace_back(alnIdx);
         }
@@ -384,7 +387,6 @@ class DBSCANCluster {
             return runDBSCAN();
         }
         // Already good => finish it without clustering
-        prevMaxClusterSize = neighbors.size();
         finalClusters.insert(neighbors);
         return finishDBSCAN();
     }