Allow tsv2exprofiledb to not compress outputs

soedinglab · Jul 23, 2024 · a146887 · a146887
1 parent 62975ca
commit a146887
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 11 deletions.
diff --git a/data/workflow/tsv2exprofiledb.sh b/data/workflow/tsv2exprofiledb.sh
@@ -17,25 +17,37 @@ OUT="$2"
 [ -d "${OUT}.tsv" ] && echo "${OUT} is a directory!" && exit 1;
 
 if notExists "${OUT}_h.dbtype"; then
-  "$MMSEQS" tsv2db "${IN}_h.tsv" "${OUT}_h" --output-dbtype 12 ${VERBOSITY}
+  MMSEQS_FORCE_MERGE=1 "$MMSEQS" tsv2db "${IN}_h.tsv" "${OUT}_h" --output-dbtype 12 ${VERBOSITY}
 fi
 
 if notExists "${OUT}.dbtype"; then
-  "$MMSEQS" tsv2db "${IN}.tsv" "${OUT}_tmp" --output-dbtype 0 ${VERBOSITY}
-  MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_tmp" "${OUT}" ${VERBOSITY}
-  "$MMSEQS" rmdb "${OUT}_tmp" ${VERBOSITY}
+  if [ -n "${COMPRESSED}" ]; then
+    "$MMSEQS" tsv2db "${IN}.tsv" "${OUT}_tmp" --output-dbtype 0 ${VERBOSITY}
+    MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_tmp" "${OUT}" ${VERBOSITY}
+    "$MMSEQS" rmdb "${OUT}_tmp" ${VERBOSITY}
+  else
+    MMSEQS_FORCE_MERGE=1 "$MMSEQS" tsv2db "${IN}.tsv" "${OUT}" --output-dbtype 0 ${VERBOSITY}
+  fi
 fi
 
 if notExists "${OUT}_seq.dbtype"; then
-  "$MMSEQS" tsv2db "${IN}_seq.tsv" "${OUT}_seq_tmp" --output-dbtype 0 ${VERBOSITY}
-  MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_seq_tmp" "${OUT}_seq" ${VERBOSITY}
-  "$MMSEQS" rmdb "${OUT}_seq_tmp" ${VERBOSITY}
+  if [ -n "${COMPRESSED}" ]; then
+    "$MMSEQS" tsv2db "${IN}_seq.tsv" "${OUT}_seq_tmp" --output-dbtype 0 ${VERBOSITY}
+    MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_seq_tmp" "${OUT}_seq" ${VERBOSITY}
+    "$MMSEQS" rmdb "${OUT}_seq_tmp" ${VERBOSITY}
+  else
+    "$MMSEQS" tsv2db "${IN}_seq.tsv" "${OUT}_seq" --output-dbtype 0 ${VERBOSITY}
+  fi
 fi
 
 if notExists "${OUT}_aln.dbtype"; then
-  "$MMSEQS" tsv2db "${IN}_aln.tsv" "${OUT}_aln_tmp" --output-dbtype 5 ${VERBOSITY}
-  MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_aln_tmp" "${OUT}_aln" ${VERBOSITY}
-  "$MMSEQS" rmdb "${OUT}_aln_tmp" ${VERBOSITY}
+  if [ -n "${COMPRESSED}" ]; then
+    "$MMSEQS" tsv2db "${IN}_aln.tsv" "${OUT}_aln_tmp" --output-dbtype 5 ${VERBOSITY}
+    MMSEQS_FORCE_MERGE=1 "$MMSEQS" compress "${OUT}_aln_tmp" "${OUT}_aln" ${VERBOSITY}
+    "$MMSEQS" rmdb "${OUT}_aln_tmp" ${VERBOSITY}
+  else
+    MMSEQS_FORCE_MERGE=1 "$MMSEQS" tsv2db "${IN}_aln.tsv" "${OUT}_aln" --output-dbtype 5 ${VERBOSITY}
+  fi
 fi
 
 if notExists "${OUT}_seq_h.dbtype"; then

diff --git a/src/MMseqsBase.cpp b/src/MMseqsBase.cpp
@@ -1150,7 +1150,7 @@ std::vector<Command> baseCommands = {
                 "<i:hhsuiteHHMDB> <o:profileDB>",
                 CITATION_MMSEQS2,{{"",DbType::ACCESS_MODE_INPUT, DbType::NEED_DATA, NULL}}},
 
-        {"tsv2exprofiledb",      tsv2exprofiledb,      &par.onlyverbosity,        COMMAND_PROFILE_PROFILE,
+        {"tsv2exprofiledb",      tsv2exprofiledb,      &par.verbandcompression,   COMMAND_PROFILE_PROFILE,
                 "Create a expandable profile db from TSV files",
                 NULL,
                 "Milot Mirdita <[email protected]>",

diff --git a/src/util/tsv2exprofiledb.cpp b/src/util/tsv2exprofiledb.cpp
@@ -6,14 +6,20 @@
 
 #include "tsv2exprofiledb.sh.h"
 
+void setTsv2ExProfileDbDefaults(Parameters *p) {
+    p->compressed = true;
+}
+
 int tsv2exprofiledb(int argc, const char **argv, const Command &command) {
     Parameters &par = Parameters::getInstance();
+    setTsv2ExProfileDbDefaults(&par);
     par.parseParameters(argc, argv, command, true, 0, 0);
 
     std::string program = par.db2 + ".sh";
     FileUtil::writeFile(program, tsv2exprofiledb_sh, tsv2exprofiledb_sh_len);
 
     CommandCaller cmd;
+    cmd.addVariable("COMPRESSED", par.compressed ? "TRUE" : NULL);
     cmd.addVariable("VERBOSITY", par.createParameterString(par.onlyverbosity).c_str());
     cmd.execProgram(FileUtil::getRealPathFromSymLink(program).c_str(), par.filenames);