add a separate entry point for generating JSON in msa2lddt

steineggerlab · Jul 29, 2024 · e19eb74 · e19eb74
1 parent 4d67ddd
commit e19eb74
Show file tree

Hide file tree

Showing 5 changed files with 52 additions and 31 deletions.
diff --git a/src/LocalCommandDeclarations.h b/src/LocalCommandDeclarations.h
@@ -24,6 +24,7 @@ extern int structuremsa(int argc, const char** argv, const Command &command);
 extern int structuremsacluster(int argc, const char** argv, const Command &command);
 extern int msa2lddt(int argc, const char** argv, const Command &command);
 extern int msa2lddtreport(int argc, const char** argv, const Command &command);
+extern int msa2lddtjson(int argc, const char** argv, const Command &command);
 extern int refinemsa(int argc, const char** argv, const Command &command);
 extern int convert2pdb(int argc, const char** argv, const Command &command);
 extern int compressca(int argc, const char** argv, const Command &command);

diff --git a/src/commons/FoldmasonParameters.cpp b/src/commons/FoldmasonParameters.cpp
@@ -64,7 +64,8 @@ FoldmasonParameters::FoldmasonParameters() :
     refinemsa = combineList(refinemsa, structuremsa);
     
     // easymsa
-    PARAM_REPORT_MODE.description = "MSA report mode 0: AA/3Di FASTA files only, 1: Compute LDDT and generate HTML report";
+    PARAM_REPORT_MODE.description = "MSA report mode 0: AA/3Di FASTA files only, 1: Compute LDDT and generate HTML report, 2: Compute LDDT and generate JSON";
+    PARAM_REPORT_MODE.regex = "^[0-2]{1}$";
     easymsaworkflow = combineList(easymsaworkflow, structurecreatedb);
     easymsaworkflow = combineList(easymsaworkflow, structuremsa);
     easymsaworkflow = combineList(easymsaworkflow, msa2lddt);

diff --git a/src/foldmason.cpp b/src/foldmason.cpp
@@ -64,6 +64,14 @@ std::vector<struct Command> foldmasonCommands = {
                 CITATION_FOLDMASON, {{"queryDB", DbType::ACCESS_MODE_INPUT, DbType::NEED_DATA, &DbValidator::sequenceDb },
                                     {"msaFile", DbType::ACCESS_MODE_INPUT, DbType::NEED_DATA, &DbValidator::flatfileAndStdin },
                                     {"htmlFile", DbType::ACCESS_MODE_OUTPUT, DbType::NEED_DATA, &DbValidator::flatfile }}},
+        {"msa2lddtjson",      msa2lddtjson,      &foldmasonPar.msa2lddt,      COMMAND_ALIGNMENT,
+                "Calculate LDDT of a MSA and generate HTML report",
+                NULL,
+                "Cameron Gilchrist <[email protected]> & Martin Steinegger <[email protected]>",
+                "<i:queryDB> <i:msaFile> <o:jsonFile>",
+                CITATION_FOLDMASON, {{"queryDB", DbType::ACCESS_MODE_INPUT, DbType::NEED_DATA, &DbValidator::sequenceDb },
+                                    {"msaFile", DbType::ACCESS_MODE_INPUT, DbType::NEED_DATA, &DbValidator::flatfileAndStdin },
+                                    {"jsonFile", DbType::ACCESS_MODE_OUTPUT, DbType::NEED_DATA, &DbValidator::flatfile }}},
         {"refinemsa",     refinemsa,      &foldmasonPar.refinemsa,      COMMAND_ALIGNMENT,
                 "Iteratively refine a MSA using structural information",
                 NULL,

diff --git a/src/strucclustutils/msa2lddt.cpp b/src/strucclustutils/msa2lddt.cpp
@@ -360,7 +360,7 @@ float getLDDTScore(
     return lddtScore;
 }
 
-int msa2lddt(int argc, const char **argv, const Command& command, bool makeReport) {
+int msa2lddt(int argc, const char **argv, const Command& command, int makeReport) {
     FoldmasonParameters &par = FoldmasonParameters::getFoldmasonInstance();
 
     const bool touch = (par.preloadMode != Parameters::PRELOAD_MODE_MMAP);
@@ -419,12 +419,13 @@ int msa2lddt(int argc, const char **argv, const Command& command, bool makeRepor
         resultWriter.writeData(dst, realSize, 0, 0, false, false);
 */
 
-        size_t dstSize = ZSTD_findDecompressedSize(vendor_foldmason_js_zst, vendor_foldmason_js_zst_len);
-        char* dst = (char*)malloc(sizeof(char) * dstSize);
-        size_t realSize = ZSTD_decompress(dst, dstSize, vendor_foldmason_js_zst, vendor_foldmason_js_zst_len);
-
-        std::string mainJS(const_cast<char *>(reinterpret_cast<const char *>(main_foldmason_js)), main_foldmason_js_len);
-        std::string htmlTemplate(
+        if (makeReport == 1) {
+            size_t dstSize = ZSTD_findDecompressedSize(vendor_foldmason_js_zst, vendor_foldmason_js_zst_len);
+            char* dst = (char*)malloc(sizeof(char) * dstSize);
+            size_t realSize = ZSTD_decompress(dst, dstSize, vendor_foldmason_js_zst, vendor_foldmason_js_zst_len);
+
+            std::string mainJS(const_cast<char *>(reinterpret_cast<const char *>(main_foldmason_js)), main_foldmason_js_len);
+            std::string htmlTemplate(
 R"html(<!DOCTYPE html>
 <html lang="en">
 <head>
@@ -436,26 +437,30 @@ R"html(<!DOCTYPE html>
 <div id="app"></div>
 )html");
 
-        resultWriter.writeData(htmlTemplate.c_str(), htmlTemplate.size(), 0, 0, false, false);
-
-        std::string scriptStart = "<script>";
-        std::string scriptEnd   = "</script>";
+            resultWriter.writeData(htmlTemplate.c_str(), htmlTemplate.size(), 0, 0, false, false);
 
-        // vendor.js
-        resultWriter.writeData(scriptStart.c_str(), scriptStart.size(), 0, 0, false, false);
-        resultWriter.writeData(dst, realSize, 0, 0, false, false);
-        resultWriter.writeData(scriptEnd.c_str(), scriptEnd.size(), 0, 0, false, false);
-
-        // main.js
-        resultWriter.writeData(scriptStart.c_str(), scriptStart.size(), 0, 0, false, false);
-        resultWriter.writeData(mainJS.c_str(), mainJS.size(), 0, 0, false, false);
-        resultWriter.writeData(scriptEnd.c_str(), scriptEnd.size(), 0, 0, false, false);
-
-        // Data <div>
-        const char* dataStart = "<div id=\"data\" style=\"display: none;\">\n{\"entries\": [";
-        resultWriter.writeData(dataStart, strlen(dataStart), 0, 0, false, false);
+            std::string scriptStart = "<script>";
+            std::string scriptEnd   = "</script>";
 
-        free(dst);
+            // vendor.js
+            resultWriter.writeData(scriptStart.c_str(), scriptStart.size(), 0, 0, false, false);
+            resultWriter.writeData(dst, realSize, 0, 0, false, false);
+            resultWriter.writeData(scriptEnd.c_str(), scriptEnd.size(), 0, 0, false, false);
+
+            // main.js
+            resultWriter.writeData(scriptStart.c_str(), scriptStart.size(), 0, 0, false, false);
+            resultWriter.writeData(mainJS.c_str(), mainJS.size(), 0, 0, false, false);
+            resultWriter.writeData(scriptEnd.c_str(), scriptEnd.size(), 0, 0, false, false);
+
+            // Data <div>
+            const char* dataStart = "<div id=\"data\" style=\"display: none;\">\n{\"entries\": [";
+            resultWriter.writeData(dataStart, strlen(dataStart), 0, 0, false, false);
+
+            free(dst);
+        } else if (makeReport == 2) {
+            const char* dataStart = "{\"entries\": [";
+            resultWriter.writeData(dataStart, strlen(dataStart), 0, 0, false, false);
+        }
 
         // tree: string (optional, from --guide-tree)
         // entries: [ { name, aa, ss, ca }, ... ]
@@ -519,8 +524,11 @@ R"html(<!DOCTYPE html>
             end.append(par.reportCommand);
             end.append("\"");
         }
-        end.append("}}</script>");
-
+        end.append("}}");
+
+        if (makeReport == 1) {
+            end.append("</script>");
+        }
         resultWriter.writeData(end.c_str(), end.length(), 0, 0, false, false);
         resultWriter.writeEnd(0, 0, false, 0);
         resultWriter.close(true);
@@ -540,8 +548,11 @@ R"html(<!DOCTYPE html>
 
 
 int msa2lddt(int argc, const char **argv, const Command& command) {
-    return msa2lddt(argc, argv, command, false);
+    return msa2lddt(argc, argv, command, 0);
 }
 int msa2lddtreport(int argc, const char **argv, const Command& command) {
-    return msa2lddt(argc, argv, command, true);
+    return msa2lddt(argc, argv, command, 1);
+}
+int msa2lddtjson(int argc, const char **argv, const Command& command) {
+    return msa2lddt(argc, argv, command, 2);
 }
diff --git a/src/workflow/EasyMSA.cpp b/src/workflow/EasyMSA.cpp
@@ -92,7 +92,7 @@ int easymsa(int argc, const char **argv, const Command &command) {
     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("MAKE_REPORT", (par.reportMode == 1) ? "TRUE" : NULL);
+    cmd.addVariable("MAKE_REPORT", std::to_string(par.reportMode).c_str());
     cmd.addVariable("CREATEDB_PAR", par.createParameterString(par.structurecreatedb).c_str());
 
     // needs to be last