pskvins · antonvnv · Apr 10, 2026 · Apr 11, 2026 · Apr 11, 2026
diff --git a/data/workflow/databases.sh b/data/workflow/databases.sh
@@ -299,6 +299,28 @@ case "${SELECTION}" in
         push_back "${TMP_PATH}/kalamari.fasta"
         INPUT_TYPE="FASTA_LIST"
     ;;
+    "RNAcentral_current"|"RNAcentral_26_0")
+        case "${SELECTION}" in
+            "RNAcentral_current") RNACENTRAL_URL="https://ftp.ebi.ac.uk/pub/databases/RNAcentral/current_release/sequences/rnacentral_active.fasta.gz" ;;
+            "RNAcentral_26_0")    RNACENTRAL_URL="https://ftp.ebi.ac.uk/pub/databases/RNAcentral/releases/26.0/sequences/rnacentral_active.fasta.gz" ;;
+        esac
+        DL_DIR="${DOWNLOAD_DIR:-${TMP_PATH}}/${SELECTION}"
+        mkdir -p "${DL_DIR}"
+        if notExists "${DL_DIR}/rnacentral_active.fasta.gz"; then
+            downloadFile "${RNACENTRAL_URL}" "${DL_DIR}/rnacentral_active.fasta.gz"
+        fi
+        date "+%s" > "${TMP_PATH}/version"
+        push_back "${DL_DIR}/rnacentral_active.fasta.gz"
+        INPUT_TYPE="LOCAL_FASTA"
+    ;;
+    *)
+        if [ ! -f "${SELECTION}" ]; then
+            fail "Local file not found: ${SELECTION}"
+        fi
+        push_back "${SELECTION}"
+        date "+%s" > "${TMP_PATH}/version"
+        INPUT_TYPE="LOCAL_FASTA"
+    ;;
 esac
 
 if notExists "${OUTDB}.dbtype"; then
@@ -312,6 +334,42 @@ case "${INPUT_TYPE}" in
             rm -f -- "$i"
         done
     ;;
+    "LOCAL_FASTA")
+        eval "set -- $ARR"
+        export MMSEQS_FORCE_MERGE=1
+        # shellcheck disable=SC2086
+        set +e
+        CREATEDB_LOG=$("${MMSEQS}" createdb "${@}" "${TMP_PATH}/createdb" ${THREADS_COMP_PAR} 2>&1)
+        CREATEDB_RET=$?
+        set -e
+        printf "%s\n" "${CREATEDB_LOG}"
+        [ $CREATEDB_RET -ne 0 ] && fail "createdb died"
+        case "${CREATEDB_LOG}" in
+            *"Database type: Nucleotide"*) ;;
+            *) fail "Local file indexing only supports nucleotide databases" ;;
+        esac
+        # shellcheck disable=SC2086
+        "${MMSEQS}" splitsequence "${TMP_PATH}/createdb" "${TMP_PATH}/splitdb" \
+            --sequence-overlap 0 --sequence-split-mode 0 --headers-split-mode 1 \
+            ${THREADS_PAR} \
+            || fail "splitsequence died"
+        # shellcheck disable=SC2086
+        "${MMSEQS}" makepaddedseqdb "${TMP_PATH}/splitdb" "${OUTDB}" \
+            ${THREADS_PAR} \
+            || fail "makepaddedseqdb died"
+        mkdir -p "${TMP_PATH}/indexdb"
+        # shellcheck disable=SC2086
+        "${MMSEQS}" createindex "${OUTDB}" "${TMP_PATH}/indexdb" \
+            --split 1 --index-subset 2 ${THREADS_PAR} \
+            || fail "createindex died"
+        rm -rf -- "${TMP_PATH}/indexdb"
+        if [ -n "${REMOVE_TMP}" ]; then
+            # shellcheck disable=SC2086
+            "${MMSEQS}" rmdb "${TMP_PATH}/createdb" ${VERB_PAR}
+            # shellcheck disable=SC2086
+            "${MMSEQS}" rmdb "${TMP_PATH}/splitdb" ${VERB_PAR}
+        fi
+    ;;
     "FSA")
         # shellcheck disable=SC2086
         "${MMSEQS}" createdb "${TMP_PATH}/"*.fsa "${OUTDB}" ${COMP_PAR} --gpu 1 \

diff --git a/src/MMseqsBase.cpp b/src/MMseqsBase.cpp
@@ -129,9 +129,13 @@ std::vector<Command> baseCommands = {
                                    {"tmpDir", DbType::ACCESS_MODE_OUTPUT, DbType::NEED_DATA, &DbValidator::directory }}},
         {"databases",            databases,            &par.databases,            COMMAND_DATABASE_CREATION,
                 "List and download databases",
-                NULL,
+                "# Download a known database\n"
+                "mmseqs databases UniRef30_2302 db tmp\n\n"
+                "# Index a local FASTA file (use ./relative or /absolute paths)\n"
+                "mmseqs databases ./input.fasta.gz db tmp\n"
+                "mmseqs databases /data/rnacentral_active.fasta.gz db tmp\n\n",
                 "Milot Mirdita <milot@mirdita.de>",
-                "<name> <o:sequenceDB> <tmpDir>",
+                "<name|path> <o:sequenceDB> <tmpDir>",
                 CITATION_TAXONOMY|CITATION_MMSEQS2, {{"selection", 0, DbType::ZERO_OR_ALL, &DbValidator::empty },
                                                            {"sequenceDB", DbType::ACCESS_MODE_OUTPUT, DbType::NEED_DATA, &DbValidator::sequenceDb },
                                                            {"tmpDir",     DbType::ACCESS_MODE_OUTPUT, DbType::NEED_DATA, &DbValidator::directory }}},

diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
@@ -70,6 +70,7 @@ Parameters::Parameters():
         PARAM_PRELOAD_MODE(PARAM_PRELOAD_MODE_ID, "--db-load-mode", "Preload mode", "Database preload mode 0: auto, 1: fread, 2: mmap, 3: mmap+touch", typeid(int), (void *) &preloadMode, "[0-3]{1}", MMseqsParameter::COMMAND_COMMON | MMseqsParameter::COMMAND_EXPERT),
         PARAM_SPACED_KMER_PATTERN(PARAM_SPACED_KMER_PATTERN_ID, "--spaced-kmer-pattern", "Spaced k-mer pattern", "User-specified spaced k-mer pattern", typeid(std::string), (void *) &spacedKmerPattern, "^1[01]*1$", MMseqsParameter::COMMAND_PREFILTER | MMseqsParameter::COMMAND_EXPERT),
         PARAM_LOCAL_TMP(PARAM_LOCAL_TMP_ID, "--local-tmp", "Local temporary path", "Path where some of the temporary files will be created", typeid(std::string), (void *) &localTmp, "", MMseqsParameter::COMMAND_PREFILTER | MMseqsParameter::COMMAND_EXPERT),
+        PARAM_DOWNLOAD_DIR(PARAM_DOWNLOAD_DIR_ID, "--download-dir", "Download cache directory", "Persistent directory for downloaded files (avoids re-downloading)", typeid(std::string), (void *) &downloadDir, "", MMseqsParameter::COMMAND_MISC),
         // alignment
         PARAM_ALIGNMENT_MODE(PARAM_ALIGNMENT_MODE_ID, "--alignment-mode", "Alignment mode", "How to compute the alignment:\n0: automatic\n1: only score and end_pos\n2: also start_pos and cov\n3: also seq.id\n4: only ungapped alignment", typeid(int), (void *) &alignmentMode, "^[0-5]{1}$", MMseqsParameter::COMMAND_ALIGN),
         PARAM_ALIGNMENT_OUTPUT_MODE(PARAM_ALIGNMENT_OUTPUT_MODE_ID, "--alignment-output-mode", "Alignment mode", "How to compute the alignment:\n0: automatic\n1: only score and end_pos\n2: also start_pos and cov\n3: also seq.id\n4: only ungapped alignment\n5: score only (output) cluster format", typeid(int), (void *) &alignmentOutputMode, "^[0-1]{1}$", MMseqsParameter::COMMAND_ALIGN),
@@ -1633,6 +1634,7 @@ Parameters::Parameters():
     databases.push_back(&PARAM_COMPRESSED);
     databases.push_back(&PARAM_THREADS);
     databases.push_back(&PARAM_V);
+    databases.push_back(&PARAM_DOWNLOAD_DIR);
 
     // tar2db
     tar2db.push_back(&PARAM_OUTPUT_DBTYPE);
@@ -2527,6 +2529,7 @@ void Parameters::setDefaults() {
     splitAA = false;
     spacedKmerPattern = "";
     localTmp = "";
+    downloadDir = "downloads";
 
     // search workflow
     numIterations = 1;

diff --git a/src/commons/Parameters.h b/src/commons/Parameters.h
@@ -445,6 +445,7 @@ class Parameters {
     int    realignMaxSeqs;               // Max alignments to realign
     std::string spacedKmerPattern;       // User-specified kmer pattern
     std::string localTmp;                // Local temporary path
+    std::string downloadDir;             // Persistent download cache directory
 
 
     // ALIGNMENT
@@ -824,6 +825,7 @@ class Parameters {
     PARAMETER(PARAM_PRELOAD_MODE)
     PARAMETER(PARAM_SPACED_KMER_PATTERN)
     PARAMETER(PARAM_LOCAL_TMP)
+    PARAMETER(PARAM_DOWNLOAD_DIR)
     std::vector<MMseqsParameter*> prefilter;
     std::vector<MMseqsParameter*> ungappedprefilter;
     std::vector<MMseqsParameter*> gappedprefilter;

diff --git a/src/workflow/Databases.cpp b/src/workflow/Databases.cpp
@@ -157,6 +157,20 @@ std::vector<DatabaseDownload> downloads = {{
                                                    "https://github.com/lskatz/Kalamari",
                                                    true, Parameters::DBTYPE_NUCLEOTIDES, databases_sh, databases_sh_len,
                                                    { }
+                                           }, {
+                                                   "RNAcentral_current",
+                                                   "RNAcentral active sequences (latest release). Non-deterministic: contents change with each release.",
+                                                   "RNAcentral Consortium: RNAcentral 2021: secondary structure integration, improved sequence search and new member databases. Nucleic Acids Res 49(D1), D212-D220 (2021)",
+                                                   "https://rnacentral.org",
+                                                   false, Parameters::DBTYPE_NUCLEOTIDES, databases_sh, databases_sh_len,
+                                                   { }
+                                           }, {
+                                                   "RNAcentral_26_0",
+                                                   "RNAcentral active sequences (release 26.0, deterministic).",
+                                                   "RNAcentral Consortium: RNAcentral 2021: secondary structure integration, improved sequence search and new member databases. Nucleic Acids Res 49(D1), D212-D220 (2021)",
+                                                   "https://rnacentral.org",
+                                                   false, Parameters::DBTYPE_NUCLEOTIDES, databases_sh, databases_sh_len,
+                                                   { }
                                            },
 };
 
@@ -276,9 +290,12 @@ int databases(int argc, const char **argv, const Command &command) {
         }
     }
     if (downloadIdx == -1) {
-        par.printUsageMessage(command, par.help ? MMseqsParameter::COMMAND_EXPERT : 0, description.c_str());
-        Debug(Debug::ERROR) << "Selected database " << par.db1 << " was not found\n";
-        EXIT(EXIT_FAILURE);
+        if (par.db1.find('/') == std::string::npos) {
+            par.printUsageMessage(command, par.help ? MMseqsParameter::COMMAND_EXPERT : 0, description.c_str());
+            Debug(Debug::ERROR) << "Selected database " << par.db1 << " was not found\n";
+            EXIT(EXIT_FAILURE);
+        }
+        // Path contains '/' — treat as local file, shell script validates existence
     }
     par.printParameters(command.cmd, argc, argv, par.databases);
     std::string tmpDir = par.db3;
@@ -291,19 +308,41 @@ int databases(int argc, const char **argv, const Command &command) {
     par.filenames.push_back(tmpDir);
 
     CommandCaller cmd;
-    for (size_t i = 0; i < usedDownloads[downloadIdx].environment.size(); ++i) {
-        cmd.addVariable(usedDownloads[downloadIdx].environment[i].key, usedDownloads[downloadIdx].environment[i].value);
+    if (downloadIdx >= 0) {
+        for (size_t i = 0; i < usedDownloads[downloadIdx].environment.size(); ++i) {
+            cmd.addVariable(usedDownloads[downloadIdx].environment[i].key, usedDownloads[downloadIdx].environment[i].value);
+        }
+        cmd.addVariable("TAXONOMY", usedDownloads[downloadIdx].hasTaxonomy ? "TRUE" : NULL);
+    } else {
+        cmd.addVariable("TAXONOMY", NULL);
     }
-    cmd.addVariable("TAXONOMY", usedDownloads[downloadIdx].hasTaxonomy ? "TRUE" : NULL);
     cmd.addVariable("REMOVE_TMP", par.removeTmpFiles ? "TRUE" : NULL);
     cmd.addVariable("VERB_PAR", par.createParameterString(par.onlyverbosity).c_str());
     cmd.addVariable("COMP_PAR", par.createParameterString(par.verbandcompression).c_str());
     // aria2c gives an (undocumented error with more than 16 connections)
     cmd.addVariable("ARIA_NUM_CONN", SSTR(std::min(16, par.threads)).c_str());
     cmd.addVariable("THREADS_PAR", par.createParameterString(par.onlythreads).c_str());
     cmd.addVariable("THREADS_COMP_PAR", par.createParameterString(par.threadsandcompression).c_str());
+    if (downloadIdx >= 0 && par.downloadDir.empty() == false) {
+        std::string dlDir = par.downloadDir;
+        // Resolve relative paths against CWD
+        if (dlDir[0] != '/') {
+            char cwd[PATH_MAX];
+            if (getcwd(cwd, sizeof(cwd)) != NULL) {
+                dlDir = std::string(cwd) + "/" + dlDir;
+            }
+        }
+        if (FileUtil::directoryExists(dlDir.c_str()) == false) {
+            FileUtil::makeDir(dlDir.c_str());
+        }
+        cmd.addVariable("DOWNLOAD_DIR", dlDir.c_str());
+    }
     std::string program = tmpDir + "/download.sh";
-    FileUtil::writeFile(program, usedDownloads[downloadIdx].script, usedDownloads[downloadIdx].scriptLength);
+    if (downloadIdx >= 0) {
+        FileUtil::writeFile(program, usedDownloads[downloadIdx].script, usedDownloads[downloadIdx].scriptLength);
+    } else {
+        FileUtil::writeFile(program, databases_sh, databases_sh_len);
+    }
     cmd.execProgram(program.c_str(), par.filenames);
 
     // Should never get here