pass guide tree and command string to msa2lddt visualisation

steineggerlab · Jan 10, 2024 · 32eb03b · 32eb03b
1 parent 015f426
commit 32eb03b
Show file tree

Hide file tree

Showing 8 changed files with 2,832 additions and 1,655 deletions.
diff --git a/data/easymsa.sh b/data/easymsa.sh
@@ -50,7 +50,11 @@ else
     fi   
 fi
 
-"$MMSEQS" msa2lddt "${STRUCTUREDB}" "${RESULTS}" "--lddt-html" "${RESULTS%%.*}.html" \
+# shellcheck disable=SC2086
+"$MMSEQS" msa2lddt "${STRUCTUREDB}" "${RESULTS}" \
+    "--lddt-html" "${RESULTS}.html" \
+    "--guide-tree" "${RESULTS}.nw" \
+    ${MSA2LDDT_PAR} \
     || fail "msa2lddt died"
 
 if [ -n "${REMOVE_TMP}" ]; then

diff --git a/data/main_foldmason.js b/data/main_foldmason.js
diff --git a/data/vendor_foldmason.js.zst b/data/vendor_foldmason.js.zst
diff --git a/src/commons/FoldmasonParameters.cpp b/src/commons/FoldmasonParameters.cpp
@@ -19,7 +19,8 @@ FoldmasonParameters::FoldmasonParameters() :
         PARAM_BITFACTOR_3DI(PARAM_BITFACTOR_3DI_ID, "--bitfactor-3di", "3Di matrix bit factor", "3Di matrix bit factor", typeid(float), (void *) &bitFactor3Di, "^([0-9]*\\.[0-9]*)$"),
         PARAM_OUTPUT_MODE(PARAM_OUTPUT_MODE_ID, "--output-mode", "Alignment output mode", "Output file mode: \n0: Amino acid\n1: 3Di alphabet", typeid(int), (void *) &outputmode, "[0-1]{1}$"),
         PARAM_LDDT_HTML(PARAM_LDDT_HTML_ID, "--lddt-html", "LDDT HTML file", "File to write LDDT MSA HTML visualisation to", typeid(std::string), (void *) &lddtHtml, ""),
-        PARAM_PAIR_THRESHOLD(PARAM_PAIR_THRESHOLD_ID,"--pair-threshold", "LDDT pair threshold", "% of pair subalignments with LDDT information [0.0,1.0]",typeid(float), (void *) &pairThreshold, "^0(\\.[0-9]+)?|1(\\.0+)?$")
+        PARAM_PAIR_THRESHOLD(PARAM_PAIR_THRESHOLD_ID, "--pair-threshold", "LDDT pair threshold", "% of pair subalignments with LDDT information [0.0,1.0]",typeid(float), (void *) &pairThreshold, "^0(\\.[0-9]+)?|1(\\.0+)?$"),
+        PARAM_REPORT_COMMAND(PARAM_REPORT_COMMAND_ID, "--report-command", "", "", typeid(std::string), (void *) &reportCommand, "")
 {
     // structuremsa
     structuremsa.push_back(&PARAM_WG);
@@ -53,16 +54,19 @@ FoldmasonParameters::FoldmasonParameters() :
     structuremsacluster = combineList(structuremsacluster, structuremsa);
 
     // msa2lddt
-    msa2lddt.push_back(&PARAM_HELP);
     msa2lddt.push_back(&PARAM_LDDT_HTML);
-    msa2lddt.push_back(&PARAM_THREADS);
     msa2lddt.push_back(&PARAM_PAIR_THRESHOLD);
-    
+    msa2lddt.push_back(&PARAM_THREADS);
+    msa2lddt.push_back(&PARAM_GUIDE_TREE);
+    msa2lddt.push_back(&PARAM_V);
+    msa2lddt.push_back(&PARAM_REPORT_COMMAND);
+
     // refinemsa
     refinemsa = combineList(refinemsa, structuremsa);
     
     easymsaworkflow = combineList(easymsaworkflow, structurecreatedb);
     easymsaworkflow = combineList(easymsaworkflow, structuremsa);
+    easymsaworkflow = combineList(easymsaworkflow, msa2lddt);
     easymsaworkflow.push_back(&PARAM_PRECLUSTER);
 
     pcaAa = 1.1;
@@ -73,6 +77,7 @@ FoldmasonParameters::FoldmasonParameters() :
     scoreBias3di = 0.6;
     matchRatio = 0.51;
     guideTree = "";
+    reportCommand = "";
     recomputeScores = false;
     regressive = false;
     precluster = false;

diff --git a/src/commons/FoldmasonParameters.h b/src/commons/FoldmasonParameters.h
@@ -50,7 +50,7 @@ class FoldmasonParameters : public LocalParameters {
     PARAMETER(PARAM_OUTPUT_MODE)
     PARAMETER(PARAM_LDDT_HTML)
     PARAMETER(PARAM_PAIR_THRESHOLD)
-    // PARAMETER(PARAM_NEWICK_OUTPUT)
+    PARAMETER(PARAM_REPORT_COMMAND)
 
     MultiParam<PseudoCounts> pcaAa;
     MultiParam<PseudoCounts> pcbAa;
@@ -67,6 +67,7 @@ class FoldmasonParameters : public LocalParameters {
     float bitFactor3Di;
     int outputmode;
     std::string lddtHtml;
+    std::string reportCommand;
     float pairThreshold;
 };
 #endif
diff --git a/src/strucclustutils/msa2lddt.cpp b/src/strucclustutils/msa2lddt.cpp
@@ -6,6 +6,7 @@
 #include "Parameters.h"
 #include "StructureUtil.h"
 #include "Util.h"
+#include <fstream>
 #include <cassert>
 
 #define ZSTD_STATIC_LINKING_ONLY
@@ -470,6 +471,7 @@ R"html(<!DOCTYPE html>
 
         free(dst);
 
+        // tree: string (optional, from --guide-tree)
         // entries: [ { name, aa, ss, ca }, ... ]
         // scores: [ float ]
         // statistics: { db, msaFile, msaLDDT }
@@ -504,13 +506,33 @@ R"html(<!DOCTYPE html>
                 entry.append(",");
             resultWriter.writeData(entry.c_str(), entry.length(), 0, 0, false, false);
         }
-        std::string end = "],\"statistics\": {";
-        end.append("\"db\":\"");
+        std::string end = "],";
+
+        if (par.guideTree != "") {
+            std::string tree;
+            std::string line;
+            std::ifstream newick(par.guideTree);
+            if (newick.is_open()) {
+                while (std::getline(newick, line))
+                    tree += line;
+                newick.close();
+            }
+            end.append("\"tree\": \"");
+            end.append(tree);            
+            end.append("\",");
+        }
+        end.append("\"statistics\": {\"db\":\"");
         end.append(par.db1);
         end.append("\",\"msaFile\":\"");
         end.append(par.db2);
         end.append("\",\"msaLDDT\":");
         end.append(std::to_string(lddtScore));
+
+        if (par.reportCommand != "") {
+            end.append(",\"cmdString\":\"");
+            end.append(par.reportCommand);
+            end.append("\"");
+        }
         end.append("}}</script>");
 
         resultWriter.writeData(end.c_str(), end.length(), 0, 0, false, false);
@@ -519,6 +541,10 @@ R"html(<!DOCTYPE html>
         FileUtil::remove(lddtHtmlIdx.c_str());
     }
 
+    if (par.reportCommand != "") {
+        std::cout << "Report command: " << par.reportCommand << '\n';
+    }
+
     seqDbrAA.close();
     seqDbrCA.close();
     seqDbr3Di.close();

diff --git a/src/strucclustutils/structuremsa.cpp b/src/strucclustutils/structuremsa.cpp
@@ -1451,21 +1451,27 @@ int structuremsa(int argc, const char **argv, const Command& command, bool preCl
     } else {
         memset(alreadyMerged, 0, sizeof(bool) * sequenceCnt);
     }       
-
+
+    // Check if guide tree argument given
+    // Try parse --> read if non-empty, otherwise generate one and write
+    std::string tree;
     std::vector<AlnSimple> hits;
+
     if (par.guideTree != "") {
-        std::cout << "Loading guide tree: " << par.guideTree << "\n";
-        std::string tree;
         std::string line;
         std::ifstream newick(par.guideTree);
         if (newick.is_open()) {
+            Debug(Debug::INFO) << "Writing guide tree to: " << par.guideTree << '\n';
             while (std::getline(newick, line))
                 tree += line;
             newick.close();
         }
+    }
+    if (tree != "") {
         hits = parseNewick(tree, headers_rev);
-        if (par.regressive)
+        if (par.regressive) {
             std::reverse(hits.begin(), hits.end());
+        }
     } else {
         hits = updateAllScores(
             tinySubMatAA,
@@ -1505,7 +1511,6 @@ int structuremsa(int argc, const char **argv, const Command& command, bool preCl
         hits = mst(hits, sequenceCnt);
         std::cout << "Generated guide tree\n";
     }
-
 
     std::cout << "Optimising merge order\n";
     std::vector<size_t> merges;
@@ -1521,9 +1526,13 @@ int structuremsa(int argc, const char **argv, const Command& command, bool preCl
     }
 
     std::string nw = orderToTree(hits, headers, sequenceCnt);
-    std::cout << "Tree: " << nw << ";\n";
+    std::string treeFile = par.filenames[par.filenames.size()-1] + ".nw";
+    Debug(Debug::INFO) << "Writing guide tree to: " << treeFile << '\n';
+    std::ofstream guideTree(treeFile, std::ofstream::out);        
+    guideTree << nw;
+    guideTree.close();
 
-    std::cout << "Merging:\n";
+    Debug(Debug::INFO) << "Merging:\n";
 
     size_t finalMSAId = 0;
 
@@ -1693,9 +1702,9 @@ int structuremsa(int argc, const char **argv, const Command& command, bool preCl
             groups[targetId].clear();
             mappings[targetId].clear();
 
-            testSeqLens(groups[mergedId], cigars_aa, seqLens);
+            // testSeqLens(groups[mergedId], cigars_aa, seqLens);
 
-if (true) {
+/* if (false) {
             // calculate LDDT of merged alignment
             float lddtScore = std::get<2>(calculate_lddt(cigars_aa, groups[mergedId], dbKeys, seqLens, &seqDbrCA, par.pairThreshold));
             std::cout << std::fixed << std::setprecision(4)
@@ -1705,7 +1714,14 @@ if (true) {
                 std::cout << "\t(TM-align)";
             }
             std::cout << '\n';
-}
+} */
+            std::cout << std::fixed << std::setprecision(4)
+                << queryIsProfile << "\t" << targetIsProfile << '\t' << headers[mergedId] << "\t" << headers[targetId]
+                << '\t' << res.score;
+            if (tmaligned){
+                std::cout << "\t(TM-align)";
+            }
+            std::cout << '\n';
 
             mappings[mergedId] = computeProfileMask(
                 groups[mergedId],
@@ -1802,7 +1818,7 @@ if (true) {
 
     // Write final MSA to file with correct headers
     DBWriter resultWriter(
-        par.filenames[par.filenames.size()-1].c_str(),
+        (par.filenames[par.filenames.size()-1] + ".fa").c_str(),
         (par.filenames[par.filenames.size()-1] + ".index").c_str(),
         static_cast<unsigned int>(par.threads), par.compressed, Parameters::DBTYPE_OMIT_FILE
     );

diff --git a/src/workflow/EasyMSA.cpp b/src/workflow/EasyMSA.cpp
@@ -89,11 +89,26 @@ int easymsa(int argc, const char **argv, const Command &command) {
     cmd.addVariable("REMOVE_TMP", par.removeTmpFiles ? "TRUE" : NULL);
 
     cmd.addVariable("RUNNER", par.runner.c_str());
+    cmd.addVariable("THREADS_PAR", par.createParameterString(par.onlythreads).c_str());
     cmd.addVariable("VERBOSITY", par.createParameterString(par.onlyverbosity).c_str());
-    cmd.addVariable("PRECLUSTER", par.precluster ? "TRUE" : NULL);
-
+    cmd.addVariable("PRECLUSTER", par.precluster ? "TRUE" : NULL); 
     cmd.addVariable("CREATEDB_PAR", par.createParameterString(par.structurecreatedb).c_str());
 
+    // needs to be last
+    std::vector<MMseqsParameter *> msa2lddtWithoutHtml;
+    par.PARAM_GAP_OPEN.wasSet = true;
+    par.PARAM_GAP_EXTEND.wasSet = true;
+    par.PARAM_MATCH_RATIO.wasSet = true;
+    par.PARAM_FILTER_MSA.wasSet = true;
+    par.reportCommand = par.createParameterString(par.easymsaworkflow, true);
+    for (size_t i = 0; i < par.msa2lddt.size(); i++) {
+        if (par.msa2lddt[i]->uniqid != par.PARAM_LDDT_HTML.uniqid &&
+            par.msa2lddt[i]->uniqid != par.PARAM_GUIDE_TREE.uniqid) {
+            msa2lddtWithoutHtml.push_back(par.msa2lddt[i]);
+        }
+    }
+    cmd.addVariable("MSA2LDDT_PAR", par.createParameterString(msa2lddtWithoutHtml).c_str());
+
     std::string program = tmpDir + "/easymsa.sh";
     FileUtil::writeFile(program, easymsa_sh, easymsa_sh_len);
     cmd.execProgram(program.c_str(), par.filenames);