Rework masking parameter to solve #408

steineggerlab · Jan 13, 2025 · c438b91 · c438b91
1 parent 00adfe4
commit c438b91
Show file tree

Hide file tree

Showing 6 changed files with 24 additions and 40 deletions.
diff --git a/src/commons/LocalParameters.cpp b/src/commons/LocalParameters.cpp
@@ -286,36 +286,46 @@ LocalParameters::LocalParameters() :
     convert2pdb.push_back(&PARAM_THREADS);
     convert2pdb.push_back(&PARAM_V);
 
+    // set masking
+    maskMode = 0;
+    maskNrepeats = 10;
+    maskProb = 0.999995;
+    maskLowerCaseMode = 1;
+
+    // createdb
+    maskBfactorThreshold = 0;
+    chainNameMode = 0;
+    writeMapping = 0;
+    coordStoreMode = COORD_STORE_MODE_CA_DIFF;
+    inputFormat = 0; // auto detect
+    fileInclude = ".*";
+    fileExclude = "^$";
+    prostt5SplitLength = 1024;
+    prostt5Model = "";
+
+    // search parameter
     alignmentType = ALIGNMENT_TYPE_3DI_AA;
     tmScoreThr = 0.0;
     tmScoreThrMode = TMSCORE_THRESHOLD_MODE_ALIGNMENT;
     tmAlignHitOrder = TMALIGN_HIT_ORDER_AVG;
     lddtThr = 0.0;
     evalThr = 10;
     sortByStructureBits = 1;
+    clusterSearch = 0;
     minDiagScoreThr = 30;
-    maskBfactorThreshold = 0;
-    chainNameMode = 0;
     minAssignedChainsThreshold = 0.0;
     monomerIncludeMode = 0;
-    writeMapping = 0;
     tmAlignFast = 1;
     exactTMscore = 0;
     gapOpen = 10;
     gapExtend = 1;
     nsample = 5000;
-    maskLowerCaseMode = 1;
-    coordStoreMode = COORD_STORE_MODE_CA_DIFF;
-    clusterSearch = 0;
-    inputFormat = 0; // auto detect
-    fileInclude = ".*";
-    fileExclude = "^$";
     dbSuffixList = "_h,_ss,_ca";
     indexExclude = 0;
-    multimerReportMode = 1;
+
+    // multimer
     eValueThrExpandMultimer = 10000.0;
-    prostt5SplitLength = 1024;
-    prostt5Model = "";
+    multimerReportMode = 1;
     dbExtractionMode = DB_EXTRACT_MODE_CHAIN;
     distanceThreshold = 8.0;
     filtMultimerTmThr = 0.0;

diff --git a/src/workflow/EasyStructureSearch.cpp b/src/workflow/EasyStructureSearch.cpp
@@ -12,24 +12,15 @@
 void setEasyStructureSearchDefaults(Parameters *p) {
     // TODO: 7-mer sensitivity is not optimized yet
     p->kmerSize = 6;
-    p->maskMode = 0;
-    p->maskProb = 0.99995;
     p->sensitivity = 9.5;
     p->maxResListLen = 1000;
-    p->gapOpen = 10;
-    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;
-    p->PARAM_MASK_RESIDUES.wasSet = true;
-    p->PARAM_MASK_PROBABILTY.wasSet = true;
-    p->PARAM_NO_COMP_BIAS_CORR.wasSet = true;
     p->PARAM_S.wasSet = true;
-    p->PARAM_GAP_OPEN.wasSet = true;
-    p->PARAM_GAP_EXTEND.wasSet = true;
     p->PARAM_ALIGNMENT_MODE.wasSet = true;
     p->PARAM_REMOVE_TMP_FILES.wasSet = true;
 }

diff --git a/src/workflow/StructureCluster.cpp b/src/workflow/StructureCluster.cpp
@@ -16,10 +16,7 @@ void setStructureClusterWorkflowDefaults(LocalParameters *p) {
     p->maxResListLen = 1000;
     p->kmersPerSequence = 300;
     p->alignmentMode = Parameters::ALIGNMENT_MODE_SCORE_COV_SEQID;
-    p->maskMode = 0;
     p->compBiasCorrection = 0;
-    p->gapOpen = 10;
-    p->gapExtend = 1;
 }
 
 //TODO this makes no sense for structures

diff --git a/src/workflow/StructureIndex.cpp b/src/workflow/StructureIndex.cpp
@@ -35,7 +35,7 @@ int structureindex(int argc, const char **argv, const Command& command) {
 
     par.parseParameters(argc, argv, command, true, 0, 0);
     setStructureSearchMustPassAlong(&par);
-
+    par.PARAM_MASK_RESIDUES.wasSet = true;
     std::string tmpDir = par.db2;
     std::string hash = SSTR(par.hashParameter(command.databases, par.filenames, par.createindex));
     if (par.reuseLatest) {

diff --git a/src/workflow/StructureRbh.cpp b/src/workflow/StructureRbh.cpp
@@ -11,10 +11,7 @@ namespace structureRbh{
 
 
 void setStructureRbhDefaults(LocalParameters *p) {
-    p->maskMode = 0;
-    p->compBiasCorrection = 0;
     p->sortByStructureBits = 0;
-    p->maskProb = 0.99995;
     p->sensitivity = 9.5;
     p->maxResListLen = 1000;
     p->gapOpen = 10;

diff --git a/src/workflow/StructureSearch.cpp b/src/workflow/StructureSearch.cpp
@@ -11,27 +11,16 @@
 
 void setStructureSearchWorkflowDefaults(LocalParameters *p) {
     p->kmerSize = 0;
-    p->maskMode = 0;
-    p->maskNrepeats = 10;
-    p->maskProb = 0.999995;
     p->sensitivity = 9.5;
     p->maxResListLen = 1000;
-    p->gapOpen = 10;
-    p->gapExtend = 1;
     p->alignmentMode = Parameters::ALIGNMENT_MODE_SCORE_COV_SEQID;
     p->removeTmpFiles = true;
-
 }
 
 void setStructureSearchMustPassAlong(LocalParameters *p) {
     p->PARAM_K.wasSet = true;
-    p->PARAM_MASK_RESIDUES.wasSet = true;
-    p->PARAM_MASK_PROBABILTY.wasSet = true;
-    p->PARAM_MASK_N_REPEAT.wasSet = true;
-    p->PARAM_NO_COMP_BIAS_CORR.wasSet = true;
     p->PARAM_S.wasSet = true;
-    p->PARAM_GAP_OPEN.wasSet = true;
-    p->PARAM_GAP_EXTEND.wasSet = true;
+    p->PARAM_MAX_SEQS.wasSet = true;
     p->PARAM_ALIGNMENT_MODE.wasSet = true;
     p->PARAM_REMOVE_TMP_FILES.wasSet = true;
 }