From da25260aa01343730158bbd1d39fc4dc1a2b94f9 Mon Sep 17 00:00:00 2001
From: Evelin Aasna <evelin.aasna@fu-berlin.de>
Date: Thu, 12 Dec 2024 14:01:13 +0100
Subject: [PATCH] Always distribute metagenome search (#135)

---
 include/valik/search/search_local.hpp |  8 +-------
 include/valik/shared.hpp              |  1 -
 src/argument_parsing/search.cpp       |  2 ++
 test/cli/valik_options_test.cpp       | 15 ---------------
 test/cli/valik_test.cpp               |  1 -
 5 files changed, 3 insertions(+), 24 deletions(-)

diff --git a/include/valik/search/search_local.hpp b/include/valik/search/search_local.hpp
index 49d308e7..f065d9a8 100644
--- a/include/valik/search/search_local.hpp
+++ b/include/valik/search/search_local.hpp
@@ -34,7 +34,6 @@ static inline dream_stellar::StellarOptions make_thread_options(search_arguments
     threadOptions.referenceLength = refLen;
     threadOptions.searchSegment = true;
     auto seg = ref_meta.segment_from_bin(bin_id);
-    //!TODO: deal with metagenome database
     threadOptions.binSequences.emplace_back(seg.seq_vec[0]);
     threadOptions.segmentBegin = seg.start;
     threadOptions.segmentEnd = seg.start + seg.len;
@@ -66,10 +65,6 @@ static inline dream_stellar::StellarOptions make_thread_options(search_arguments
 template <bool is_split, bool stellar_only>
 bool search_local(search_arguments & arguments, search_time_statistics & time_statistics)
 {
-    if (arguments.bin_path.size() > 1 || (arguments.bin_path.size() > 0 && arguments.bin_path[0].size() > 1))
-        throw std::runtime_error("Multiple reference files can not be searched in shared memory mode. "
-                                 "Add --distribute argument to launch multiple distributed instances of DREAM-Stellar search.");
-
     using index_structure_t = index_structure::ibf;
     auto index = valik_index<index_structure_t>{};
 
@@ -90,7 +85,7 @@ bool search_local(search_arguments & arguments, search_time_statistics & time_st
 
         auto prefilter_bin_count = ref_meta.seg_count;
         split_arguments stellar_dist_arguments;
-        // distribute stellar search
+        // stellar search without prefiltering
         
         // for some number of reference sequences split sequences into as many segments as is the next multiple of thread count
         if (ref_meta.seq_count % arguments.threads > 0)
@@ -188,7 +183,6 @@ bool search_local(search_arguments & arguments, search_time_statistics & time_st
     bool const databasesSuccess = input_databases_time.measure_time([&]()
     {
         std::cout << "Launching stellar search on a shared memory machine...\n";
-        //!TODO: allow metagenome database
         return dream_stellar::_importAllSequences(arguments.bin_path[0][0].c_str(), "database", databases, databaseIDs, refLen, std::cout, std::cerr);
     });
     if (!databasesSuccess)
diff --git a/include/valik/shared.hpp b/include/valik/shared.hpp
index 4cb54666..8d0613b3 100644
--- a/include/valik/shared.hpp
+++ b/include/valik/shared.hpp
@@ -136,7 +136,6 @@ struct search_profile_arguments
     //!TODO: deduce this automatically
     bool split_query{false}; 
     bool manual_parameters{false};
-    //!TODO: make fourth option: MINIMISER
     search_kind search_type{search_kind::LEMMA}; 
     double fnr;
 
diff --git a/src/argument_parsing/search.cpp b/src/argument_parsing/search.cpp
index 0eb58ba4..fdeaa469 100644
--- a/src/argument_parsing/search.cpp
+++ b/src/argument_parsing/search.cpp
@@ -275,6 +275,8 @@ void run_search(sharg::parser & parser)
         arguments.shape_weight = arguments.shape.count();
         arguments.window_size = tmp.window_size();
         arguments.bin_path = tmp.bin_path();
+        if (arguments.bin_path.size() > 1)
+            arguments.distribute = true;
     }
 
     // ==========================================
diff --git a/test/cli/valik_options_test.cpp b/test/cli/valik_options_test.cpp
index 3652395f..64ca43f8 100644
--- a/test/cli/valik_options_test.cpp
+++ b/test/cli/valik_options_test.cpp
@@ -307,18 +307,3 @@ TEST_F(argparse_search, not_manual_no_meta)
     EXPECT_EQ(result.out, std::string{});
     EXPECT_EQ(result.err, std::string{"[Error] Provide --ref-meta to deduce suitable search parameters or set --without-parameter-tuning and --pattern size.\n"});
 }
-
-TEST_F(argparse_search, shared_mem_metagenome)
-{
-    cli_test_result const result = execute_app("valik", "search",
-                                                         "--query ", data("query.fq"),
-                                                         "--index ", data("8bins19window.ibf"),
-                                                         "--output search.gff",
-                                                         "--ref-meta ", data("150overlap4bins.bin"),
-                                                         "--pattern 100",
-                                                         "--without-parameter-tuning");
-    EXPECT_NE(result.exit_code, 0);
-    EXPECT_EQ(result.out, std::string{});
-    EXPECT_EQ(result.err, std::string{"[Error] Multiple reference files can not be searched in shared memory mode. "
-                                      "Add --distribute argument to launch multiple distributed instances of DREAM-Stellar search.\n"});
-}
diff --git a/test/cli/valik_test.cpp b/test/cli/valik_test.cpp
index 2de1f137..c090673d 100644
--- a/test/cli/valik_test.cpp
+++ b/test/cli/valik_test.cpp
@@ -345,7 +345,6 @@ TEST_P(valik_search_clusters, search)
 
     cli_test_result const result = execute_app("valik", "search",
                                                         "--output search.gff",
-                                                        "--distribute",
                                                         "--pattern", std::to_string(pattern_size),
                                                         "--query-every", std::to_string(query_every),
                                                         "--error-rate ", std::to_string(error_rate),