From 609958bc6910d01355eb14f3b948bc20ddc2860e Mon Sep 17 00:00:00 2001
From: Nikola Petrov <nikola@petrovv.com>
Date: Wed, 29 Jan 2025 23:08:47 +0100
Subject: [PATCH] add similarity functions

---
 shared/inc/values/Similarity.hpp |  16 +++++
 shared/src/values/Similarity.cpp | 114 +++++++++++++++++++++++++++++++
 2 files changed, 130 insertions(+)
 create mode 100644 shared/inc/values/Similarity.hpp
 create mode 100644 shared/src/values/Similarity.cpp

diff --git a/shared/inc/values/Similarity.hpp b/shared/inc/values/Similarity.hpp
new file mode 100644
index 0000000..f257082
--- /dev/null
+++ b/shared/inc/values/Similarity.hpp
@@ -0,0 +1,16 @@
+#include "Dna.hpp"
+#include <vector>
+
+namespace Similarity
+{
+  // float euclidean_distance(Dna *d1, Dna *d2); direct distance betwen vector. wont give 0 and 1
+  //  float dot_product(Dna *d1, Dna *d2); doent return betwen 0 to 1
+  float cosine_similarity(Dna *d1, Dna *d2);
+  float hamming_distance(Dna *d1, Dna *d2);
+  float jaccard_index(Dna *d1, Dna *d2);
+  float levenshtein_distance(Dna *d1, Dna *d2);
+  // float needleman_wunsch(Dna *d1, Dna *d2); used for bioinformatics and aligment. Dont need its aligned alredy
+
+  typedef float(simil_func)(Dna *d1, Dna *d2);
+  float calc_similarity(std::vector<Dna> vec, simil_func f);
+}
\ No newline at end of file
diff --git a/shared/src/values/Similarity.cpp b/shared/src/values/Similarity.cpp
new file mode 100644
index 0000000..197f6ea
--- /dev/null
+++ b/shared/src/values/Similarity.cpp
@@ -0,0 +1,114 @@
+#include "values/Similarity.hpp"
+#include <cmath>
+
+#define MATCH 1
+#define MISMATCH -1
+#define GAP -2
+
+namespace Similarity
+{
+
+  float cosine_similarity(Dna *d1, Dna *d2)
+  {
+    uint8_t *d1a = (uint8_t *)d1;
+    uint8_t *d2a = (uint8_t *)d2;
+
+    float mag1 = 0.0f;
+    float mag2 = 0.0f;
+    float dot_prod = 0.0f;
+    for (size_t i = 0; i < sizeof(Dna); i++)
+    {
+      dot_prod += d1a[i] * d2a[i];
+      mag1 += d1a[i] * d1a[i];
+      mag2 += d2a[i] * d2a[i];
+    }
+    mag1 = sqrt(mag1);
+    mag2 = sqrt(mag2);
+
+    return dot_prod / (mag1 * mag2);
+  }
+
+  float hamming_distance(Dna *d1, Dna *d2)
+  {
+    uint8_t *d1a = (uint8_t *)d1;
+    uint8_t *d2a = (uint8_t *)d2;
+    float distance = 0;
+    for (size_t i = 0; i < sizeof(Dna); i++)
+    {
+      if (d1a[i] != d2a[i])
+      {
+        distance++;
+      }
+    }
+    return 1 - (distance / sizeof(Dna));
+  }
+
+  float jaccard_index(Dna *d1, Dna *d2)
+  {
+    uint8_t *d1a = (uint8_t *)d1;
+    uint8_t *d2a = (uint8_t *)d2;
+    size_t intersection = 0;
+    size_t union_size = sizeof(Dna) + sizeof(Dna);
+
+    for (size_t i = 0; i < sizeof(Dna); i++)
+    {
+      for (size_t j = 0; j < sizeof(Dna); j++)
+      {
+        if (d1a[i] == d2a[j])
+        {
+          intersection++;
+          break;
+        }
+      }
+    }
+
+    union_size -= intersection;
+    return (float)intersection / union_size;
+  }
+
+  float levenshtein_distance(Dna *d1, Dna *d2)
+  {
+    auto min = [](uint8_t a, uint8_t b, uint8_t c) -> uint8_t
+    {
+      return (a < b ? (a < c ? a : c) : (b < c ? b : c));
+    };
+
+    uint8_t *d1a = (uint8_t *)d1;
+    uint8_t *d2a = (uint8_t *)d2;
+    float matrix[sizeof(Dna) + 1][sizeof(Dna) + 1];
+    for (size_t i = 0; i <= sizeof(Dna); i++)
+    {
+      matrix[i][0] = i;
+    }
+    for (size_t j = 0; j <= sizeof(Dna); j++)
+    {
+      matrix[0][j] = j;
+    }
+    for (size_t i = 1; i <= sizeof(Dna); i++)
+    {
+      for (size_t j = 1; j <= sizeof(Dna); j++)
+      {
+        uint8_t cost = (d1a[i - 1] == d2a[j - 1]) ? 0 : 1;
+        matrix[i][j] = min(matrix[i - 1][j] + 1, matrix[i][j - 1] + 1, matrix[i - 1][j - 1] + cost);
+      }
+    }
+    float ld = matrix[sizeof(Dna)][sizeof(Dna)];
+    return 1 - (ld / sizeof(Dna));
+  }
+
+  float calc_similarity(std::vector<Dna> vec, simil_func f)
+  {
+    size_t num_pairs = (vec.size() * (vec.size() - 1)) / 2;
+
+    float total_similarity = 0.0;
+    for (size_t i = 0; i < vec.size(); i++)
+    {
+      for (size_t j = i + 1; j < vec.size(); j++)
+      {
+        total_similarity += f(&vec[i], &vec[j]);
+      }
+    }
+    float average_similarity = total_similarity / num_pairs;
+    return average_similarity;
+  }
+}