Squashed commit of the following:

commit f2488f431acbc45cd04cc1705fb0e0480f9b2d3f
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Tue Jul 22 18:50:02 2025 +0200

    Updated version

commit 4f3650ed6690bdfd4d4322f80ceb7755159e943e
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Tue Jul 22 18:32:18 2025 +0200

    Fixed target-indexed bug

commit e8cc97fd0df0356074f1e6c802c13f062657b46b
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Sun Jul 20 11:31:54 2025 +0200

    Updated changelog

commit a8b454b4422acf7cc3d917e6e1e9428cdbb5c868
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Thu Jul 10 23:33:05 2025 +0200

    Added bidirectional coverage

commit 07d2b898299c98f90e5d174567a199db66c222f7
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Fri Jul 4 22:55:44 2025 +0200

    Added multiround logic

commit 2e3c6cafe5d4700b23f7dcc2bfcdc9a1d0850503
Merge: d832aa36 8ec818c
Author: Benjamin Buchfink <buchfink@gmail.com>
Date:   Mon Jun 30 17:19:04 2025 +0200

    Merged w/master

Author: bbuchfink

CMakeLists.txt

@@ -5,25 +5,28 @@ include(CheckSymbolExists)
 include(CheckTypeSize)
 
 option(BUILD_STATIC "BUILD_STATIC" OFF)
-option(EXTRA "EXTRA" OFF)
 option(STATIC_LIBGCC "STATIC_LIBGCC" OFF)
 option(STATIC_LIBSTDC++ "STATIC_LIBSTDC++" OFF)
 option(X86 "X86" ON)
 option(ARM "ARM" OFF)
 option(AARCH64 "AARCH64" OFF)
+option(WITH_ZSTD "WITH_ZSTD" OFF)
+
+# experimental options not suitable for deployment purposes
+option(EXTRA "EXTRA" OFF)
 option(STRICT_BAND "STRICT_BAND" ON)
 option(LEFTMOST_SEED_FILTER "LEFTMOST_SEED_FILTER" ON)
 option(SEQ_MASK "SEQ_MASK" ON)
 option(DP_STAT "DP_STAT" OFF)
 option(SINGLE_THREADED "SINGLE_THREADED" OFF)
-option(WITH_ZSTD "WITH_ZSTD" OFF)
 option(HIT_KEEP_TARGET_ID "HIT_KEEP_TARGET_ID" OFF)
 option(LONG_SEEDS "LONG_SEEDS" OFF)
 option(WITH_AVX512 "WITH_AVX512" OFF)
 option(WITH_DNA "WITH_DNA" OFF)
 option(WITH_MCL "WITH_MCL" OFF)
 option(WITH_MIMALLOC "WITH_MIMALLOC" OFF)
 option(USE_TLS "USE_TLS" OFF)
+
 set(MAX_SHAPE_LEN 19)
 set(BLAST_INCLUDE_DIR "" CACHE STRING "BLAST_INCLUDE_DIR")
 set(BLAST_LIBRARY_DIR "" CACHE STRING "BLAST_LIBRARY_DIR")
@@ -382,6 +385,7 @@ set(OBJECTS
         src/cluster/external/external.cpp
         src/cluster/external/align.cpp
         src/cluster/external/cluster.cpp
+        src/cluster/external/output.cpp
         )
 
 if(WITH_DNA)
@@ -503,7 +507,8 @@ endif()
 
 if (${CMAKE_CXX_COMPILER_ID} STREQUAL MSVC)
   find_library(BCRYPT_LIBRARY BCrypt)
-  target_link_libraries(diamond ${BCRYPT_LIBRARY})
+  find_library(WS2_32_LIBRARY Ws2_32)
+  target_link_libraries(diamond ${BCRYPT_LIBRARY} ${WS2_32_LIBRARY})
 endif()
 
 target_link_libraries(diamond ${ZLIB_LIBRARY} ${CMAKE_THREAD_LIBS_INIT})

src/ChangeLog

@@ -1,6 +1,19 @@
+[2.1.13]
+- Fixed an invalid error message for the `cluster`, `deepclust` and `linclust`
+  workflows.
+- Added the option `--oid-output` to output OIDs instead of accessions for
+  the clustering workflows, reducing its memory use.
+- Added support for using the `--multiprocessing` feature on Windows.
+- Using `--multiprocessing` requires explicitly setting `--parallel-tmpdir`.
+- Fixed a bug that could cause a crash when the `--target-indexed` option was
+  used.
+- As of now, a macOS binary is available as a GitHub release, supporting both
+  x86 and Apple silicon CPUs. Using BLAST databases is also supported.
+
 [2.1.12]
 - Added support for the new NCBI taxonomic ranks "cellular root",
   "acellular root", "domain" and "realm".
+- Added support for using BLAST databases to the Bioconda release.
 - Fixed compiler errors for Clang 20.
 - Enabled transitive closure computation in earlier clustering rounds and
   for bi-directional coverage clustering.

src/basic/basic.cpp

@@ -29,7 +29,7 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #include "stats/standard_matrix.h"
 #include "util/log_stream.h"
 
-const char* Const::version_string = "2.1.12";
+const char* Const::version_string = "2.1.13";
 using std::string;
 using std::vector;
 using std::count;

src/basic/config.cpp

@@ -37,9 +37,6 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #include "search/search.h"
 #include "stats/cbs.h"
 #include "basic/shape.h"
-#ifndef _MSC_VER
-#include <sys/stat.h>
-#endif
 
 using std::thread;
 using std::stringstream;
@@ -930,26 +927,15 @@ Config::Config(int argc, const char **argv, bool check_io, CommandLineParser& pa
 	if (query_range_culling && taxon_k != 0)
 		throw std::runtime_error("--taxon-k is not supported for --range-culling mode.");
 
-	if (parallel_tmpdir == "") {
-		parallel_tmpdir = tmpdir;
-	} else {
-#ifndef WIN32
-		if (multiprocessing) {
-			// char * env_str = std::getenv("SLURM_JOBID");
-			// if (env_str) {
-			// 	parallel_tmpdir = join_path(parallel_tmpdir, "diamond_job_"+string(env_str));
-			// }
-			errno = 0;
-			int s = mkdir(parallel_tmpdir.c_str(), 00770);
-			if (s != 0) {
-				if (errno == EEXIST) {
-					// directory did already exist
-				} else {
-					throw(std::runtime_error("could not create parallel temporary directory " + parallel_tmpdir));
-				}
-			}
-		}
-#endif
+	if (multiprocessing && parallel_tmpdir.empty())
+		throw std::runtime_error("--multiprocessing requires setting --parallel-tmpdir");
+	
+	if (multiprocessing) {
+		// char * env_str = std::getenv("SLURM_JOBID");
+		// if (env_str) {
+		// 	parallel_tmpdir = join_path(parallel_tmpdir, "diamond_job_"+string(env_str));
+		// }
+		mkdir(parallel_tmpdir);
 	}
 
 	log_stream << "MAX_SHAPE_LEN=" << MAX_SHAPE_LEN;

src/basic/const.h

@@ -25,7 +25,7 @@ struct Const
 {
 
 	enum {
-		build_version = 166,
+		build_version = 167,
 		max_seed_weight = 32,
 		max_shapes = 64
 	};

src/cluster/cascaded/wrapper.cpp

@@ -160,6 +160,7 @@ static vector<SuperBlockId> search_vs_centroids(shared_ptr<FastaFile>& super_blo
 
 void Cascaded::run() {
 	config.database.require();
+	init_thresholds();
 	if (!config.parallel_tmpdir.empty()) {
 #ifdef EXTRA
 		external();
@@ -168,7 +169,6 @@ void Cascaded::run() {
 #endif
 		return;
 	}
-	init_thresholds();
 	config.hamming_ext = config.approx_min_id >= 50.0;
 
 	TaskTimer total_time;

src/cluster/external/align.cpp

@@ -36,15 +36,29 @@ static void align_rep(ThreadPool& tp, const ChunkSeqs& chunk_seqs, vector<PairEn
 	}
 	DP::Params params{ rep, nullptr, Frame(0), rep.length(), nullptr, DP::Flags::FULL_MATRIX, false, max_len, HspValues::COORDS, stats, &tp };
 	const list<Hsp> hsps = DP::BandedSwipe::swipe(targets, params);
+	const bool unid = !config.mutual_cover.present();
 	for (const Hsp& h : hsps) {
 		const int64_t member_oid = begin[h.swipe_target].member_oid;
 		const Sequence member = chunk_seqs[member_oid];
 		if (h.approx_id_percent(rep, member) < config.approx_min_id.get(0))
 			continue;
-		if (h.subject_cover_percent(member.length()) >= config.member_cover.get(80))
-			out.write(MurmurHash()(member_oid) & (RADIX_COUNT - 1), Edge(rep_oid, member_oid, rep.length(), member.length()));
-		if (h.query_cover_percent(rep.length()) >= config.member_cover.get(80))
-			out.write(MurmurHash()(rep_oid) & (RADIX_COUNT - 1), Edge(member_oid, rep_oid, member.length(), rep.length()));
+		if (unid) {
+			if (h.subject_cover_percent(member.length()) >= config.member_cover.get(80))
+				out.write(MurmurHash()(member_oid) & (RADIX_COUNT - 1), Edge(rep_oid, member_oid, rep.length(), member.length()));
+			if (h.query_cover_percent(rep.length()) >= config.member_cover.get(80))
+				out.write(MurmurHash()(rep_oid) & (RADIX_COUNT - 1), Edge(member_oid, rep_oid, member.length(), rep.length()));
+		}
+		else if (h.subject_cover_percent(member.length()) >= config.mutual_cover.get_present() && h.query_cover_percent(rep.length()) >= config.mutual_cover.get_present()) {
+			int64_t oid1 = rep_oid;
+			int64_t oid2 = member_oid;
+			int64_t len1 = rep.length();
+			int64_t len2 = member.length();
+			if (oid1 > oid2) {
+				std::swap(oid1, oid2);
+				std::swap(len1, len2);
+			}
+			out.write(MurmurHash()(oid1) & (RADIX_COUNT - 1), Edge(oid1, oid2, len1, len2));
+		}
 	}
 }
 
@@ -95,6 +109,7 @@ vector<string> align(Job& job, int chunk_count, int64_t db_size) {
 		log_stream << "pairs=" << pairs_processed << endl;
 		timer.go("Deallocating memory");
 		pairs_file.close();
+		remove(pairs_file.file_name.c_str());
 		chunk_seqs.reset();
 		++chunks_processed;
 		timer.finish();

src/cluster/external/build_pair_table.h

@@ -0,0 +1,77 @@
+#pragma once
+#include "util/util.h"
+#include "util/hash_function.h"
+#include "external.h"
+
+namespace Cluster {
+
+struct SeedEntry {
+	SeedEntry() :
+		seed(),
+		oid(),
+		len()
+	{}
+	SeedEntry(uint64_t seed, int64_t oid, int32_t len):
+		seed(seed),
+		oid(oid),
+		len(len)
+	{}
+	uint64_t key() const {
+		return seed;
+	}
+	bool operator<(const SeedEntry& e) const {
+		return seed < e.seed || (seed == e.seed && (len > e.len || (len == e.len && oid < e.oid)));
+	}
+	struct Key {
+		uint64_t operator()(const SeedEntry& e) const {
+			return e.seed;
+		}
+	};
+	uint64_t seed;
+	int64_t oid;
+	int32_t len;
+};
+
+void get_pairs_uni_cov(KeyMergeIterator<const SeedEntry*, SeedEntry::Key>& it, BufferArray& buffers) {
+	const int64_t rep_oid = it.begin()->oid, radix = MurmurHash()(rep_oid) & (RADIX_COUNT - 1);
+	const int32_t rep_len = it.begin()->len;
+	for (auto j = it.begin() + 1; j < it.end(); ++j) {
+		if (rep_oid == j->oid)
+			continue;
+		buffers.write(radix, PairEntry(rep_oid, j->oid, rep_len, j->len));
+	}
+}
+
+void get_pairs_mutual_cov(KeyMergeIterator<const SeedEntry*, SeedEntry::Key>& it, BufferArray& buffers) {
+	const double mlr = config.min_length_ratio;
+	const int32_t s = (int)it.count();
+	int32_t j = 0;
+	const auto begin = it.begin();
+	for (int32_t i = 0; i < s;) {
+		const Loc qlen = begin[i].len;
+		int32_t j1 = j;
+		for (; j1 < s; ++j1) {
+			const Loc tlen = begin[j1].len;
+			if ((double)tlen / qlen < mlr)
+				break;
+		}
+		const int32_t qpos = i + (j1 - j) / 2;
+		const int64_t rep_oid = begin[qpos].oid;
+		const int32_t rep_len = begin[qpos].len;
+		const int64_t radix = MurmurHash()(rep_oid) & (RADIX_COUNT - 1);
+		for (int k = j; k < j1; ++k)
+			if (rep_oid != begin[k].oid)
+				buffers.write(radix, PairEntry(rep_oid, begin[k].oid, rep_len, begin[k].len));
+		j = j1;
+		if (j == s)
+			break;
+		const Loc tlen = begin[j].len;
+		for (; i < s; ++i) {
+			const Loc qlen = begin[i].len;
+			if ((double)tlen / qlen >= mlr)
+				break;
+		}
+	}
+}
+
+}