From 3310337471fc46880c245508af6a23adcb192cee Mon Sep 17 00:00:00 2001
From: Jaebeom Kim <68528165+jaebeom-kim@users.noreply.github.com>
Date: Mon, 2 Dec 2024 16:27:35 +0900
Subject: [PATCH] Add parameter for taxonomy report in easy-search (#389)

---
 data/easystructuresearch.sh          |  6 ++++++
 src/commons/LocalParameters.cpp      |  1 +
 src/workflow/EasyStructureSearch.cpp | 10 ++++++++++
 3 files changed, 17 insertions(+)

diff --git a/data/easystructuresearch.sh b/data/easystructuresearch.sh
index ba482f01..30a06e7a 100644
--- a/data/easystructuresearch.sh
+++ b/data/easystructuresearch.sh
@@ -51,6 +51,12 @@ if notExists "${TMP_PATH}/alis.dbtype"; then
         || fail "Convert Alignments died"
 fi
 
+if [ -n "${TAXONOMY}" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" taxonomyreport "${TARGET}${INDEXEXT}" "${INTERMEDIATE}" "${RESULTS}_report" ${TAXONOMYREPORT_PAR} \
+        || fail "taxonomyreport died"
+fi
+
 if [ -n "${REMOVE_TMP}" ]; then
     if [ -n "${GREEDY_BEST_HITS}" ]; then
         # shellcheck disable=SC2086
diff --git a/src/commons/LocalParameters.cpp b/src/commons/LocalParameters.cpp
index 645ae0b2..cf85e356 100644
--- a/src/commons/LocalParameters.cpp
+++ b/src/commons/LocalParameters.cpp
@@ -158,6 +158,7 @@ LocalParameters::LocalParameters() :
 
     easystructuresearchworkflow = combineList(structuresearchworkflow, structurecreatedb);
     easystructuresearchworkflow = combineList(easystructuresearchworkflow, convertalignments);
+    easystructuresearchworkflow = combineList(easystructuresearchworkflow, taxonomyreport);
     easystructuresearchworkflow.push_back(&PARAM_GREEDY_BEST_HITS);
 
     structureclusterworkflow = combineList(prefilter, structurealign);
diff --git a/src/workflow/EasyStructureSearch.cpp b/src/workflow/EasyStructureSearch.cpp
index c59cc578..49b69b19 100644
--- a/src/workflow/EasyStructureSearch.cpp
+++ b/src/workflow/EasyStructureSearch.cpp
@@ -20,6 +20,7 @@ void setEasyStructureSearchDefaults(Parameters *p) {
     p->gapExtend = 1;
     p->alignmentMode = Parameters::ALIGNMENT_MODE_SCORE_COV_SEQID;
     p->removeTmpFiles = true;
+    p->reportMode = 2;
 }
 void setEasyStructureSearchMustPassAlong(Parameters *p) {
     p->PARAM_K.wasSet = true;
@@ -50,6 +51,8 @@ int easystructuresearch(int argc, const char **argv, const Command &command) {
     par.PARAM_THREADS.removeCategory(MMseqsParameter::COMMAND_EXPERT);
     par.PARAM_V.removeCategory(MMseqsParameter::COMMAND_EXPERT);
 
+    par.overrideParameterDescription(par.PARAM_REPORT_MODE, "Taxonomy report mode 0: Kraken 1: Krona 2: Skip taxonomy report", "^[0-2]{1}$", 0);
+
     setEasyStructureSearchDefaults(&par);
     par.parseParameters(argc, argv, command, true, Parameters::PARSE_VARIADIC, 0);
     setEasyStructureSearchMustPassAlong(&par);
@@ -69,6 +72,10 @@ int easystructuresearch(int argc, const char **argv, const Command &command) {
         bool needLDDT = false;
         LocalParameters::getOutputFormat(par.formatAlignmentMode, par.outfmt, needSequenceDB, needBacktrace, needFullHeaders,
                                     needLookup, needSource, needTaxonomyMapping, needTaxonomy, needQCA, needTCA, needTMalign, needLDDT);
+        if (par.reportMode != 2) {
+            needTaxonomy = true;
+            needTaxonomyMapping = true;
+        }
     }
 
     if (par.formatAlignmentMode == Parameters::FORMAT_ALIGNMENT_SAM ||
@@ -129,6 +136,9 @@ int easystructuresearch(int argc, const char **argv, const Command &command) {
     cmd.addVariable("CONVERT_PAR", par.createParameterString(par.convertalignments).c_str());
     cmd.addVariable("SUMMARIZE_PAR", par.createParameterString(par.summarizeresult).c_str());
 
+    cmd.addVariable("TAXONOMY", needTaxonomy && needTaxonomyMapping && par.reportMode != 2 ? "TRUE" : NULL);
+    cmd.addVariable("TAXONOMYREPORT_PAR", par.createParameterString(par.taxonomyreport).c_str());
+
     std::string program = tmpDir + "/easystructuresearch.sh";
     FileUtil::writeFile(program, easystructuresearch_sh, easystructuresearch_sh_len);
     cmd.execProgram(program.c_str(), par.filenames);