From 36f754352a6e35378a516649aa307ae509ad8f91 Mon Sep 17 00:00:00 2001
From: mourisl <mourisl@hotmail.com>
Date: Wed, 7 Aug 2024 13:11:21 -0400
Subject: [PATCH 1/3] The SSE macro for the popcount is unnecessary, and it
 should be changed to GNUC. Add the functionality to promote taxonomy ID to
 canonical ranks.

---
 Classifier.hpp      |  3 +++
 Makefile            |  2 +-
 Taxonomy.hpp        | 47 +++++++++++++++++++++++++++++++++++++++++++++
 compactds/Utils.hpp |  6 +-----
 4 files changed, 52 insertions(+), 6 deletions(-)

diff --git a/Classifier.hpp b/Classifier.hpp
index 7854ab5..f12830a 100644
--- a/Classifier.hpp
+++ b/Classifier.hpp
@@ -527,6 +527,9 @@ class Classifier
 
       SimpleVector<size_t> taxIds ;
       _taxonomy.ReduceTaxIds(bestSeqTaxIds, taxIds, _param.maxResult) ;
+      // Centrifuge will promote to canonical tax levels here. 
+      //   Maybe we will do the same in some future version.
+      //_taxonomy.PromoteToCanonicalTaxRank(taxIds, /*dedup=*/true) ;
 
       size = taxIds.Size() ;
       for (i = 0 ; i < size ; ++i)
diff --git a/Makefile b/Makefile
index 54282ef..4ea751b 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 CXX = g++
-CXXFLAGS= -Wall -g -O3 -msse4.2 #-pg -g #-Wall #-O3
+CXXFLAGS= -Wall -g -O3 -march=native #-pg -g #-Wall #-O3
 LINKPATH=
 LINKFLAGS = -lpthread -lz 
 DEBUG=
diff --git a/Taxonomy.hpp b/Taxonomy.hpp
index e219522..bc22676 100644
--- a/Taxonomy.hpp
+++ b/Taxonomy.hpp
@@ -370,6 +370,16 @@ class Taxonomy
         return i ;
     return _nodeCnt ;
   }
+
+  bool IsCanonicalRankNum(uint8_t r) // The taxonomy ranks defined in the TaxonomyPathTable in centrifuge, except for subspecies
+  {
+    if (r == RANK_STRAIN //|| r == RANK_SUB_SPECIES
+        || r == RANK_SPECIES || r == RANK_GENUS || r == RANK_FAMILY || r == RANK_ORDER
+        || r == RANK_CLASS || r == RANK_PHYLUM || r == RANK_KINGDOM || r == RANK_SUPER_KINGDOM
+        || r == RANK_DOMAIN)
+      return true ;
+    return false ;
+  }
 public:
   Taxonomy() 
   {
@@ -692,6 +702,43 @@ class Taxonomy
       promotedTaxIds.PushBack(_rootCTaxId) ;
   }
 
+  // Promote the taxIds to the ranks defined in the "IsCanonicalRankNum" function
+  // dedup: true: remove the duplicated item in the taxIds
+  void PromoteToCanonicalTaxRank(SimpleVector<size_t> &taxIds, bool dedup)
+  {
+    size_t i ;
+    size_t taxCnt = taxIds.Size() ;
+    for (i = 0 ; i < taxCnt ; ++i)
+    {
+      size_t p = taxIds[i] ;
+      uint8_t rank = _taxonomyTree[p].rank ;
+      while ( !IsCanonicalRankNum(rank) )
+      {
+        if (p == _taxonomyTree[p].parentTid)
+          break ;
+        p = _taxonomyTree[p].parentTid ;
+        rank = _taxonomyTree[p].rank ;
+      }
+      taxIds[i] = p ;
+    }
+
+    if (dedup)
+    {
+      std::map<size_t, int> used ;
+      size_t k = 0 ;
+      for (i = 0 ; i < taxCnt ; ++i)
+      {
+        if (used.find(taxIds[i]) == used.end())
+        {
+          taxIds[k] = taxIds[i] ;
+          ++k ;
+          used[taxIds[i]] = 1 ;
+        } 
+      }
+      taxIds.Resize(k) ;
+    }
+  }
+
   // @return: Number of children tax ids. childrenTax: compact tax ids below or equal to ctid.
   size_t GetChildrenTax(size_t ctid, std::map<size_t, int> &childrenTax)
   {
diff --git a/compactds/Utils.hpp b/compactds/Utils.hpp
index e97c051..97a09e8 100644
--- a/compactds/Utils.hpp
+++ b/compactds/Utils.hpp
@@ -10,10 +10,6 @@
 #include <math.h>
 #include <string.h>
 
-#ifdef __SSE4_2__
-#include <xmmintrin.h>
-#endif
-
 namespace compactds {
 #define WORD_64 // comment this out if word size is 32
 
@@ -80,7 +76,7 @@ class Utils
   // Count the number of 1's in x.
   static int Popcount(WORD x)
   {
-#ifdef __SSE4_2__
+#ifdef __GNUC__
       return __builtin_popcountll(x);
 #else
 #ifdef WORD_64

From e73972263b19a4c837a90bfe7e83824ac197de47 Mon Sep 17 00:00:00 2001
From: mourisl <mourisl@hotmail.com>
Date: Thu, 8 Aug 2024 16:06:33 -0400
Subject: [PATCH 2/3] Bump up version number

---
 defs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/defs.h b/defs.h
index f9ff27c..b002bda 100644
--- a/defs.h
+++ b/defs.h
@@ -5,7 +5,7 @@
 
 //#define DEBUG
 
-#define CENTRIFUGER_VERSION "1.0.5-r159"
+#define CENTRIFUGER_VERSION "1.0.5-r161"
 
 extern char nucToNum[26] ; 
 extern char numToNuc[26] ;

From 3de22d0b4055e4f82bea59328b88012697d6ffd9 Mon Sep 17 00:00:00 2001
From: mourisl <mourisl@hotmail.com>
Date: Thu, 8 Aug 2024 16:12:50 -0400
Subject: [PATCH 3/3] Correct the version number

---
 defs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/defs.h b/defs.h
index b002bda..362bc57 100644
--- a/defs.h
+++ b/defs.h
@@ -5,7 +5,7 @@
 
 //#define DEBUG
 
-#define CENTRIFUGER_VERSION "1.0.5-r161"
+#define CENTRIFUGER_VERSION "1.0.5-r163"
 
 extern char nucToNum[26] ; 
 extern char numToNuc[26] ;