From 1f6509e160a9039fe43237920742eca1314aec75 Mon Sep 17 00:00:00 2001
From: Evelin Aasna <evelin.aasna@fu-berlin.de>
Date: Wed, 10 Apr 2024 17:08:46 +0200
Subject: [PATCH] Find IBF bin size for a genome minimiser set

---
 include/utilities/threshold/basics.hpp |  2 +-
 include/valik/search/search_local.hpp  |  3 +-
 src/argument_parsing/build.cpp         | 52 +++++++++++++-------------
 src/argument_parsing/search.cpp        |  3 ++
 src/prepare/compute_bin_size.cpp       |  1 +
 5 files changed, 33 insertions(+), 28 deletions(-)

diff --git a/include/utilities/threshold/basics.hpp b/include/utilities/threshold/basics.hpp
index 8969e3bb..98682b22 100644
--- a/include/utilities/threshold/basics.hpp
+++ b/include/utilities/threshold/basics.hpp
@@ -107,6 +107,6 @@ inline uint64_t combinations(size_t const k, size_t const n)
         return 0;
 }
 
-enum class search_kind {LEMMA, HEURISTIC, STELLAR};
+enum class search_kind {LEMMA, HEURISTIC, MINIMISER, STELLAR};
 
 }   //namespace valik
diff --git a/include/valik/search/search_local.hpp b/include/valik/search/search_local.hpp
index 39c3c1bf..00677207 100644
--- a/include/valik/search/search_local.hpp
+++ b/include/valik/search/search_local.hpp
@@ -100,10 +100,11 @@ bool search_local(search_arguments & arguments, search_time_statistics & time_st
 
                 std::cout << "\n-----------Search parameters-----------\n";
                 std::cout << "kmer size " << std::to_string(arguments.shape_size) << '\n';
+                std::cout << "window size " << std::to_string(arguments.window_size) << '\n';
                 switch (arguments.search_type)
                 {
                     case search_kind::LEMMA: std::cout << "k-mer lemma "; break;
-                    //case search_kind::MINIMISER: std::cout << "minimiser "; break;
+                    case search_kind::MINIMISER: std::cout << "minimiser "; break;
                     case search_kind::HEURISTIC: std::cout << "heuristic "; break;
                     default: break;
                 }
diff --git a/src/argument_parsing/build.cpp b/src/argument_parsing/build.cpp
index 656d8139..d199c38f 100644
--- a/src/argument_parsing/build.cpp
+++ b/src/argument_parsing/build.cpp
@@ -112,32 +112,6 @@ void run_build(sharg::parser & parser)
         arguments.window_size = search_profile.k;
     }
 
-    // ==========================================
-    // Various checks.
-    // ==========================================
-    if (parser.is_option_set("kmer-count-cutoff") && parser.is_option_set("use-filesize-dependent-cutoff"))
-        throw sharg::parser_error{"You cannot use both --kmer-count-cutoff and --use-filesize-dependent-cutoff."};
-
-    if (parser.is_option_set("window"))
-    {
-        if (arguments.kmer_size > arguments.window_size)
-            throw sharg::parser_error{"The k-mer size cannot be bigger than the window size."};
-    }
-    else
-    {
-        if (arguments.fast)
-        {
-            arguments.window_size = arguments.kmer_size + 2;
-            raptor::compute_minimiser(arguments);
-            arguments.input_is_minimiser = true;
-        }
-        else
-            arguments.window_size = arguments.kmer_size;
-    }
-
-    arguments.shape = seqan3::shape{seqan3::ungapped{arguments.kmer_size}};
-    arguments.shape_weight = arguments.shape.count();
-
     // ==========================================
     // Process bin_path:
     // if building from clustered sequences each line in input corresponds to a bin
@@ -167,6 +141,32 @@ void run_build(sharg::parser & parser)
         }
     }
 
+    // ==========================================
+    // Process minimiser parameters for IBF size calculation.
+    // ==========================================
+    if (parser.is_option_set("kmer-count-cutoff") && parser.is_option_set("use-filesize-dependent-cutoff"))
+        throw sharg::parser_error{"You cannot use both --kmer-count-cutoff and --use-filesize-dependent-cutoff."};
+
+    arguments.shape = seqan3::shape{seqan3::ungapped{arguments.kmer_size}};
+    arguments.shape_weight = arguments.shape.count();
+
+    if (parser.is_option_set("window"))
+    {
+        if (arguments.kmer_size > arguments.window_size)
+            throw sharg::parser_error{"The k-mer size cannot be bigger than the window size."};
+    }
+    else
+    {
+        if (arguments.fast)
+        {
+            arguments.window_size = arguments.kmer_size + 2;
+            raptor::compute_minimiser(arguments);
+            arguments.input_is_minimiser = true;
+        }
+        else
+            arguments.window_size = arguments.kmer_size;
+    }
+
     try
     {
         sharg::output_file_validator{sharg::output_file_open_options::open_or_create}(arguments.out_path);
diff --git a/src/argument_parsing/search.cpp b/src/argument_parsing/search.cpp
index 7be85dc1..6a56d5c5 100644
--- a/src/argument_parsing/search.cpp
+++ b/src/argument_parsing/search.cpp
@@ -341,6 +341,9 @@ void run_search(sharg::parser & parser)
             arguments.threshold_percentage = arguments.threshold / (double) (arguments.pattern_size - arguments.shape.size() + 1);
             arguments.fnr = error_profile.fnr;
         }
+
+        if (arguments.window_size > arguments.shape_size)
+            arguments.search_type = search_kind::MINIMISER;
     }
 
     // ==========================================
diff --git a/src/prepare/compute_bin_size.cpp b/src/prepare/compute_bin_size.cpp
index 598802e8..404b7de5 100644
--- a/src/prepare/compute_bin_size.cpp
+++ b/src/prepare/compute_bin_size.cpp
@@ -8,6 +8,7 @@
  */
 
 #include <utilities/prepare/compute_bin_size.hpp>
+#include <seqan3/core/debug_stream.hpp>
 
 namespace raptor
 {