Nov 5, 2024: Reserve memory and cleanup

vgl-hub · Nov 11, 2024 · 8e94d76 · 8e94d76
1 parent f3eb7c6
commit 8e94d76
Show file tree

Hide file tree

Showing 3 changed files with 64 additions and 29 deletions.
diff --git a/include/teloscope.h b/include/teloscope.h
@@ -104,9 +104,33 @@ class Teloscope {
     std::vector<float> entropyValues; // Total entropy values
     std::vector<float> gcContentValues; // Total GC content values
 
-    float getShannonEntropy(const uint32_t nucleotideCounts[4], uint32_t windowSize);
-    float getGCContent(const uint32_t nucleotideCounts[4], uint32_t windowSize);
-    void getPatternDensities(WindowData& windowData, uint32_t windowSize);
+
+    inline float getShannonEntropy(const uint32_t nucleotideCounts[4], uint32_t windowSize) {
+        float entropy = 0.0;
+        for (int i = 0; i < 4; ++i) {
+            if (nucleotideCounts[i] > 0) {
+                float probability = static_cast<float>(nucleotideCounts[i]) / windowSize;
+                entropy -= probability * std::log2(probability);
+            }
+        }
+        return entropy;
+    }
+
+
+    inline float getGCContent(const uint32_t nucleotideCounts[4], uint32_t windowSize) {
+        uint32_t gcCount = nucleotideCounts[1] + nucleotideCounts[2]; // Indices: 1 = C, 2 = G
+        return static_cast<float>(gcCount) / windowSize * 100.0;
+    }
+
+
+    inline void getPatternDensities(WindowData& windowData, uint32_t windowSize) {
+        for (auto &entry : windowData.patternMap) {
+            auto &pattern = entry.first;
+            auto &data = entry.second;
+            data.density = static_cast<float>(data.count * pattern.size()) / windowSize;
+        }
+    }
+
 
     float getMean(const std::vector<float>& values);
     float getMedian(std::vector<float> values);

diff --git a/include/tools.h b/include/tools.h
@@ -8,6 +8,7 @@
 #include <iostream>
 #include <algorithm>
 #include <stdexcept>
+// #include <numeric>
 
 void getCombinations(const std::string &pattern, std::string &current, size_t index, std::vector<std::string> &combinations);
 
@@ -16,4 +17,32 @@ std::unordered_map<std::string, uint8_t> getHammingDistances(
     const std::pair<std::string, std::string> &canonicalPatterns
 );
 
+// inline float getMean(const std::vector<float>& values) {
+//     if (values.empty()) return 0.0;
+//     float sum = std::accumulate(values.begin(), values.end(), 0.0);
+//     return sum / values.size();
+// }
+
+// inline float getMedian(std::vector<float> values) {
+//     if (values.empty()) return 0.0;
+//     std::sort(values.begin(), values.end());
+//     size_t size = values.size();
+//     if (size % 2 == 0) {
+//         return (values[size / 2 - 1] + values[size / 2]) / 2;
+//     } else {
+//         return values[size / 2];
+//     }
+// }
+
+// inline float getMin(const std::vector<float>& values) {
+//     if (values.empty()) return 0.0;
+//     return *std::min_element(values.begin(), values.end());
+// }
+
+// inline float getMax(const std::vector<float>& values) {
+//     if (values.empty()) return 0.0;
+//     return *std::max_element(values.begin(), values.end());
+// }
+
+
 #endif // TOOLS_H
diff --git a/src/teloscope.cpp b/src/teloscope.cpp
@@ -41,32 +41,6 @@ void Trie::insertPattern(const std::string& pattern) {
 }
 
 
-float Teloscope::getShannonEntropy(const uint32_t nucleotideCounts[4], uint32_t windowSize) {
-    float entropy = 0.0;
-    for (int i = 0; i < 4; ++i) {
-        if (nucleotideCounts[i] > 0) {
-            float probability = static_cast<float>(nucleotideCounts[i]) / windowSize;
-            entropy -= probability * std::log2(probability);
-        }
-    }
-    return entropy;
-}
-
-float Teloscope::getGCContent(const uint32_t nucleotideCounts[4], uint32_t windowSize) {
-    uint32_t gcCount = nucleotideCounts[1] + nucleotideCounts[2]; // Index 1 = C, Index 2 = G
-    return float(gcCount) / windowSize * 100.0;
-}
-
-
-void Teloscope::getPatternDensities(WindowData& windowData, uint32_t windowSize) {
-    for (auto &entry : windowData.patternMap) {
-        auto &pattern = entry.first;
-        auto &data = entry.second;
-        data.density = static_cast<float>(data.count * pattern.size()) / windowSize;
-    }
-}
-
-
 float Teloscope::getMean(const std::vector<float>& values) {
     if (values.empty()) return 0.0;
     float sum = std::accumulate(values.begin(), values.end(), 0.0);
@@ -344,6 +318,14 @@ SegmentData Teloscope::analyzeSegment(std::string &sequence, UserInputTeloscope
 
         // Prepare and analyze current window
         WindowData windowData = prevOverlapData;
+
+        // Reserve space for vectors
+        windowData.winBlocks.reserve(windowSize / 13 + 1);
+        windowData.hDistances.reserve(windowSize / 6);
+        windowData.canonicalMatches.reserve(windowSize / 6);
+        windowData.nonCanonicalMatches.reserve(windowSize / 6);
+        windowData.windowMatches.reserve(windowSize / 6);
+
         analyzeWindow(window, windowStart, windowData, nextOverlapData);
 
         if (userInput.modeGC) { windowData.gcContent = getGCContent(windowData.nucleotideCounts, window.size()); }