Get foldseek path from parameter, defaults to same place as spacedust binary

Author: milot-mirdita

Spacedust.ipynb

@@ -354,6 +354,8 @@
         "  wget -q wget https://mmseqs.com/foldseek/foldseek-linux-avx2.tar.gz\n",
         "  tar -xzf foldseek-linux-avx2.tar.gz\n",
         "  rm -f foldseek-linux-avx2.tar.gz\n",
+        "  mv foldseek/bin/foldseek spacedust/bin\n",
+        "  rm -rf foldseek\n",
         "  touch FOLDSEEK_READY\n",
         "fi\n",
         "\n",

data/clusterdb.sh

@@ -9,7 +9,6 @@ notExists() {
 
 IN="$1"
 TMP_PATH="$2"
-FOLDSEEK="$(pwd)"/foldseek/bin/foldseek
 
 [ ! -f "${IN}.dbtype" ] && echo "${IN}.dbtype not found!" && exit 1;
 if [ -n "${USE_FOLDSEEK}" ]; then 

data/clustersearch.sh

@@ -21,7 +21,6 @@ QUERY="$1"
 TARGET="$2"
 OUTPUT="$3"
 TMP_PATH="$4"
-FOLDSEEK="$(pwd)"/foldseek/bin/foldseek
 
 if [ -n "${USE_FOLDSEEK}" ]; then 
     [ -n "${USE_PROFILE}" ] && [ ! -f "${TARGET}_foldseek_clu_seq.dbtype" ] && echo "${TARGET}_foldseek_clu_seq.dbtype not found! Please make sure the ${TARGET}_foldseek is clustered with clusterdb ${TARGET}_foldseek tmp --search-mode 1" && exit 1;

examples/Spacedust.ipynb

@@ -132,6 +132,8 @@
         "  wget -q wget https://mmseqs.com/foldseek/foldseek-linux-avx2.tar.gz\n",
         "  tar -xzf foldseek-linux-avx2.tar.gz\n",
         "  rm -f foldseek-linux-avx2.tar.gz\n",
+        "  mv foldseek/bin/foldseek spacedust/bin\n",
+        "  rm -rf foldseek\n",
         "  touch FOLDSEEK_READY\n",
         "fi\n",
         "\n",

src/commons/LocalParameters.h

@@ -40,6 +40,7 @@ class LocalParameters : public Parameters {
     PARAMETER(PARAM_FILE_INCLUDE)
     PARAMETER(PARAM_FILE_EXCLUDE)
     PARAMETER(PARAM_GFF_DIR)
+    PARAMETER(PARAM_FOLDSEEK_PATH)
 
     int clusterSearchMode;
     float pMHThr;
@@ -53,6 +54,7 @@ class LocalParameters : public Parameters {
     int suboptHitsFactor;
     std::string fileInclude;
     std::string fileExclude;
+    std::string foldseekPath;
 
 private:
     LocalParameters() : 
@@ -68,7 +70,8 @@ class LocalParameters : public Parameters {
         PARAM_PROFILE_CLUSTER_SEARCH(PARAM_PROFILE_CLUSTER_SEARCH_ID, "--profile-cluster-search", "Cluster search against profiles", "Perform profile(target)-sequence searches in clustersearch", typeid(bool), (void *) &profileClusterSearch, ""),
         PARAM_FILE_INCLUDE(PARAM_FILE_INCLUDE_ID, "--file-include", "File Inclusion Regex", "Include file names based on this regex", typeid(std::string), (void *) &fileInclude, "^.*$"),
         PARAM_FILE_EXCLUDE(PARAM_FILE_EXCLUDE_ID, "--file-exclude", "File Exclusion Regex", "Exclude file names based on this regex", typeid(std::string), (void *) &fileExclude, "^.*$"),
-        PARAM_GFF_DIR(PARAM_GFF_DIR_ID, "--gff-dir", "gff dir file", "Path to gff dir file", typeid(std::string), (void *) &gffDir, "")
+        PARAM_GFF_DIR(PARAM_GFF_DIR_ID, "--gff-dir", "gff dir file", "Path to gff dir file", typeid(std::string), (void *) &gffDir, ""),
+        PARAM_FOLDSEEK_PATH(PARAM_FOLDSEEK_PATH_ID, "--foldseek-path", "Path to Foldseek", "Path to Foldseek binary", typeid(std::string), (void *) &foldseekPath, "")
     {
 
         // clusterhits
@@ -132,6 +135,7 @@ class LocalParameters : public Parameters {
         clustersearchworkflow = combineList(clustersearchworkflow, clusterhits);
         clustersearchworkflow.push_back(&PARAM_PROFILE_CLUSTER_SEARCH);
         clustersearchworkflow.push_back(&PARAM_CLUSTERSEARCH_MODE);
+        clustersearchworkflow.push_back(&PARAM_FOLDSEEK_PATH);
 
         //aa2foldseek
         aa2foldseek = combineList(prefilter, align);
@@ -141,6 +145,7 @@ class LocalParameters : public Parameters {
         //clusterdb
         clusterdb = combineList(clusterworkflow, profile2seq);
         clusterdb.push_back(&PARAM_CLUSTERSEARCH_MODE);
+        clusterdb.push_back(&PARAM_FOLDSEEK_PATH);
 
         clusterSearchMode = 0;
         suboptHitsFactor = 0;
@@ -154,6 +159,7 @@ class LocalParameters : public Parameters {
         fileInclude = ".*";
         fileExclude = "^$";
         gffDir = "";
+        foldseekPath = "foldseek";
 
         //TODO: add citations (foldseek & mmseqs & clustersearch)
         citations.emplace(CITATION_SPACEDUST, "");

src/workflow/clusterdb.cpp

@@ -15,6 +15,7 @@ void setclusterDbDefaults(LocalParameters *p) {
 int clusterdb(int argc, const char **argv, const Command &command) {
     LocalParameters &par = LocalParameters::getLocalInstance();
     setclusterDbDefaults(&par);
+    par.foldseekPath = FileUtil::dirName(*(argv - 2)) + "/foldseek";
     par.parseParameters(argc, argv, command, true, 0, 0);
 
     if (FileUtil::directoryExists(par.db2.c_str()) == false) {
@@ -42,7 +43,23 @@ int clusterdb(int argc, const char **argv, const Command &command) {
     if (par.removeTmpFiles) {
         cmd.addVariable("REMOVE_TMP", "TRUE");
     }
-    cmd.addVariable("USE_FOLDSEEK", par.clusterSearchMode == 1 ? "TRUE" : NULL);
+
+    bool useFoldseek = false;
+    if (par.clusterSearchMode == 1) {
+        useFoldseek = true;
+        struct stat st;
+        if (stat(par.foldseekPath.c_str(), &st) != 0) {
+            Debug(Debug::ERROR) << "Cannot find foldseek binary " << par.foldseekPath << ".\n";
+            EXIT(EXIT_FAILURE);
+        }
+        bool isExecutable = (st.st_mode & S_IXUSR) || (st.st_mode & S_IXGRP) || (st.st_mode & S_IXOTH);
+        if (isExecutable == false) {
+            Debug(Debug::ERROR) << "Cannot execute foldseek binary " << par.foldseekPath << ".\n";
+            EXIT(EXIT_FAILURE);
+        }
+    }
+    cmd.addVariable("FOLDSEEK", par.foldseekPath.c_str());
+    cmd.addVariable("USE_FOLDSEEK", useFoldseek ? "TRUE" : NULL);
     cmd.addVariable("CLUSTER_PAR", par.createParameterString(par.clusterworkflow).c_str());
     cmd.addVariable("CONSENSUS_PAR", par.createParameterString(par.profile2seq).c_str());
     cmd.addVariable("THREADS_PAR", par.createParameterString(par.onlythreads).c_str());

src/workflow/clustersearch.cpp

@@ -39,6 +39,7 @@ void setClusterSearchWorkflowDefaults(LocalParameters *p) {
 int clustersearch(int argc, const char **argv, const Command &command) {
     LocalParameters &par = LocalParameters::getLocalInstance();
     setClusterSearchWorkflowDefaults(&par);
+    par.foldseekPath = FileUtil::dirName(*(argv - 2)) + "/foldseek";
 
     par.PARAM_MAX_REJECTED.addCategory(MMseqsParameter::COMMAND_EXPERT);
     par.PARAM_DB_OUTPUT.addCategory(MMseqsParameter::COMMAND_EXPERT);
@@ -86,8 +87,24 @@ int clustersearch(int argc, const char **argv, const Command &command) {
     if (par.removeTmpFiles) {
         cmd.addVariable("REMOVE_TMP", "TRUE");
     }
+
+    bool useFoldseek = false;
+    if (par.clusterSearchMode == 1) {
+        useFoldseek = true;
+        struct stat st;
+        if (stat(par.foldseekPath.c_str(), &st) != 0) {
+            Debug(Debug::ERROR) << "Cannot find foldseek binary " << par.foldseekPath << ".\n";
+            EXIT(EXIT_FAILURE);
+        }
+        bool isExecutable = (st.st_mode & S_IXUSR) || (st.st_mode & S_IXGRP) || (st.st_mode & S_IXOTH);
+        if (isExecutable == false) {
+            Debug(Debug::ERROR) << "Cannot execute foldseek binary " << par.foldseekPath << ".\n";
+            EXIT(EXIT_FAILURE);
+        }
+    }
     cmd.addVariable("USE_PROFILE", par.profileClusterSearch == 1 ? "TRUE" : NULL);
-    cmd.addVariable("USE_FOLDSEEK", par.clusterSearchMode == 1 ? "TRUE" : NULL);
+    cmd.addVariable("FOLDSEEK", par.foldseekPath.c_str());
+    cmd.addVariable("USE_FOLDSEEK", useFoldseek ? "TRUE" : NULL);
     cmd.addVariable("CLUSTER_PAR", par.createParameterString(par.clusterworkflow).c_str());
     if(par.numIterations <= 1){
         std::vector<MMseqsParameter*> searchwithoutnumiter;