From 277cc1215f9eab4c5bd452de98907e8281b64a69 Mon Sep 17 00:00:00 2001
From: Evelin Aasna <evelin.aasna@fu-berlin.de>
Date: Mon, 7 Oct 2024 12:44:34 +0200
Subject: [PATCH] Keep all or best repeats

---
 include/valik/search/producer_threads_parallel.hpp |  9 ++++++++-
 include/valik/shared.hpp                           |  3 ++-
 src/argument_parsing/search.cpp                    | 11 ++++++++---
 3 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/include/valik/search/producer_threads_parallel.hpp b/include/valik/search/producer_threads_parallel.hpp
index 067d4e29..261ec6c2 100644
--- a/include/valik/search/producer_threads_parallel.hpp
+++ b/include/valik/search/producer_threads_parallel.hpp
@@ -51,7 +51,7 @@ inline void prefilter_queries_parallel(seqan3::interleaved_bloom_filter<ibf_data
             {
                 if (arguments.verbose)
                     verbose_out.write_warning(record, bin_hits.size());
-                if (arguments.keep_repeats)    // keep bin hits that are supported by the most patterns per query segment
+                if (arguments.keep_best_repeats)    // keep bin hits that are supported by the most patterns per query segment
                 {
                     size_t mean_bin_support = std::max((size_t) 2, (size_t) std::round((double) total_pattern_hits / (double) bin_hits.size()));
                     for (auto const [bin, count] : bin_hits)
@@ -60,6 +60,13 @@ inline void prefilter_queries_parallel(seqan3::interleaved_bloom_filter<ibf_data
                             queue.insert(bin, record);
                     }
                 }
+                else if (arguments.keep_all_repeats)
+                {
+                    for (auto const [bin, count] : bin_hits)
+                    {
+                        queue.insert(bin, record);
+                    }
+                }
                 return;
             }
             
diff --git a/include/valik/shared.hpp b/include/valik/shared.hpp
index f0f97d92..61ab7b8f 100644
--- a/include/valik/shared.hpp
+++ b/include/valik/shared.hpp
@@ -181,7 +181,8 @@ struct search_arguments final : public minimiser_threshold_arguments, search_pro
     bool write_time{false};
     bool fast{false};
     bool verbose{false};
-    bool keep_repeats{false};
+    bool keep_best_repeats{false};
+    bool keep_all_repeats{false};
 
     size_t cart_max_capacity{1000};
     size_t max_queued_carts{std::numeric_limits<size_t>::max()};
diff --git a/src/argument_parsing/search.cpp b/src/argument_parsing/search.cpp
index 15efb778..7258cc6c 100644
--- a/src/argument_parsing/search.cpp
+++ b/src/argument_parsing/search.cpp
@@ -79,10 +79,15 @@ void init_search_parser(sharg::parser & parser, search_arguments & arguments)
                       .long_id = "without-parameter-tuning",
                       .description = "Preprocess database without setting default parameters.",
                       .advanced = true});
-    parser.add_flag(arguments.keep_repeats,
+    parser.add_flag(arguments.keep_best_repeats,
                     sharg::config{.short_id = '\0',
-                    .long_id = "keep-repeats",
-                    .description = "Do not filter out query matches from repeat regions.",
+                    .long_id = "keep-best-repeats",
+                    .description = "Find only highest similarity matches for repeat regions.",
+                    .advanced = true});
+    parser.add_flag(arguments.keep_all_repeats,
+                    sharg::config{.short_id = '\0',
+                    .long_id = "keep-all-repeats",
+                    .description = "Do not filter out query matches from repeat regions. This may significantly increase the runtime.",
                     .advanced = true});
     parser.add_option(arguments.seg_count_in,
                       sharg::config{.short_id = 'n',