MPI export fixes

veg · May 30, 2024 · f3cb549 · f3cb549
1 parent 7009f2f
commit f3cb549
Show file tree

Hide file tree

Showing 9 changed files with 36 additions and 7 deletions.
diff --git a/res/TemplateBatchFiles/GARD.bf b/res/TemplateBatchFiles/GARD.bf
@@ -313,7 +313,7 @@ if (fastRunMode) {
 gard.queue = mpi.CreateQueue (
                             {
                             "LikelihoodFunctions" : {{"gard.exportedModel"}},
-                            "Headers" : {{"libv3/all-terms.bf"}},
+                            "Headers" : {{"libv3/all-terms.bf","libv3/tasks/estimators.bf"}},
                             "Variables" : {{"gard.globalParameterCount", "gard.numSites", "gard.alignment", "gard.variableSiteMap", "gard.dataType", "terms.gard.codon","OPTIMIZE_SUMMATION_ORDER_PARTITION", "OPTIMIZATION_PRECISION"}}
                             }
                         );

diff --git a/res/TemplateBatchFiles/MSS-joint-fitter.bf b/res/TemplateBatchFiles/MSS-joint-fitter.bf
@@ -239,8 +239,8 @@ function mss.MSS_generator (type) {
     return model;
 }
 
-mss.tree_objects = {};
-mss.fit_objects = {};
+mss.tree_objects  = {};
+mss.fit_objects   = {};
 mss.lf_components = {   
                             2*mss_selector.file_count, 
                             1

diff --git a/res/TemplateBatchFiles/SelectionAnalyses/BUSTED.bf b/res/TemplateBatchFiles/SelectionAnalyses/BUSTED.bf
@@ -1362,7 +1362,7 @@ while (!busted.converged) {
         (busted.json [busted.json.evidence_ratios])[busted.constrained] = busted.EvidenceRatios ( (busted.json [busted.json.site_logl])[busted.unconstrained],  (busted.json [busted.json.site_logl])[busted.constrained]);
         (busted.json [busted.json.evidence_ratios ])[busted.optimized_null] = busted.EvidenceRatios ( (busted.json [busted.json.site_logl])[busted.unconstrained],  (busted.json [busted.json.site_logl])[busted.optimized_null]);
 
-        if (busted.LRT[terms.LRT] < 0) {
+        if (busted.LRT[terms.LRT] < -0.01) {
             parameters.RemoveConstraint (model.generic.GetGlobalParameter (busted.test.bsrel_model , terms.AddCategory (terms.parameters.omega_ratio,busted.rate_classes)));
             console.log ("----\n## Negative test LRT (convergence problem). Refitting the alternative model using the null model as a start.\n----\n");
             busted.restart_optimization = busted.null_results;

diff --git a/res/TemplateBatchFiles/SelectionAnalyses/aBSREL.bf b/res/TemplateBatchFiles/SelectionAnalyses/aBSREL.bf
@@ -95,6 +95,8 @@ absrel.json    = {
 
 selection.io.startTimer (absrel.json [terms.json.timers], "Overall", 0);
 
+KeywordArgument ("blb", "[Advanced option] Bag of little bootstrap alignment resampling rate", 1.0);
+absrel.blb = io.PromptUser ("[Advanced option] Bag of little bootstrap alignment resampling rate", 1.0, 0.01, 1, FALSE);
 
 /*------------------------------------------------------------------------------
     Key word arguments
@@ -110,9 +112,14 @@ KeywordArgument ("branches",  "Branches to test", "All");
     Continued Analysis Setup
 */
 
+
 namespace absrel {
     LoadFunctionLibrary ("modules/shared-load-file.bf");
-    load_file ("absrel");
+
+    load_file ({
+                utility.getGlobalValue("terms.prefix"): "absrel", 
+                utility.getGlobalValue("terms.data.blb_subsample") : blb
+               });
 }
 
 KeywordArgument ("multiple-hits",  "Include support for multiple nucleotide substitutions", "None");
@@ -147,6 +154,7 @@ if (absrel.do_srv) {
     absrel.synonymous_rate_classes = io.PromptUser ("The number omega rate classes to include in the model", absrel.synonymous_rate_classes, 1, 10, TRUE);
 }  
 
+
 selection.io.json_store_setting  (absrel.json, "srv", absrel.do_srv);
 
 

diff --git a/res/TemplateBatchFiles/SelectionAnalyses/modules/shared-load-file.bf b/res/TemplateBatchFiles/SelectionAnalyses/modules/shared-load-file.bf
@@ -154,11 +154,12 @@ function load_file (prefix) {
 
     settings = None;
     multiple_files = FALSE;
-
+    blb = 1.0;
 
     if (Type (prefix) == "AssociativeList") {
         multiple_files  = prefix [utility.getGlobalValue("terms.multiple_files")];
-        settings        = prefix[utility.getGlobalValue("terms.settings")];
+        settings        = prefix [utility.getGlobalValue("terms.settings")];
+        blb             = prefix [utility.getGlobalValue("terms.data.blb_subsample")];
         prefix          = prefix [utility.getGlobalValue("terms.prefix")];
     }
 
@@ -199,6 +200,10 @@ function load_file (prefix) {
     } else {
         datasets = prefix+".codon_data";
         codon_data_info = alignments.PromptForGeneticCodeAndAlignment(datasets, prefix+".codon_filter");
+        if (blb < 1 && blb > 0) {
+            console.log (blb);
+            codon_data_info [utility.getGlobalValue("terms.data.blb_subsample")] = blb;
+        }
         /** example output
         {
             "sequences": 13,

diff --git a/res/TemplateBatchFiles/libv3/all-terms.bf b/res/TemplateBatchFiles/libv3/all-terms.bf
@@ -211,6 +211,7 @@ namespace terms{
         characters     = "characters";
         pattern_id     = "pattern id";
         filename_to_index = "filename-to-index";
+        blb_subsample  = "blb-subsample"
     }
 
 

diff --git a/res/TemplateBatchFiles/libv3/tasks/alignments.bf b/res/TemplateBatchFiles/libv3/tasks/alignments.bf
@@ -418,6 +418,8 @@ function alignments.LoadCodonDataFile(dataset_name, datafilter_name, data_info)
         }
         io.CheckAssertion("`datafilter_name`.sites*3==`dataset_name`.sites", "The input alignment must have the number of sites that is divisible by 3 and must not contain stop codons");
     }
+
+
     data_info[terms.data.sites] = ^ "`datafilter_name`.sites";
     data_info[terms.data.dataset] = dataset_name;
     data_info[terms.data.datafilter] = datafilter_name;
@@ -563,6 +565,17 @@ lfunction alignments.DefineFiltersForPartitions(partitions, source_data, prefix,
             diff = test.sites - 3 * ^ (this_filter[utility.getGlobalValue("terms.data.name")] + ".sites");
 
             io.CheckAssertion("`&diff` == 0", "Partition " + this_filter[utility.getGlobalValue("terms.data.name")] + " is either has stop codons or is not in frame");
+
+            if (Type (data_info[^"terms.data.blb_subsample"]) == "Number") {
+                datafilter_name = (this_filter[utility.getGlobalValue("terms.data.name")]);
+                i = (^ "`datafilter_name`.sites") ^ (data_info[^"terms.data.blb_subsample"]) $ 1;
+                if (i > 1 && i < ^"`datafilter_name`.sites") {
+                    console.log (">Subsampling/upsampling using bag of little bootstraps with sample size `i`.");
+                    DataSetFilter test = Bootstrap (^datafilter_name, {"BLB_size": i, "BLB_sampler" : ""});
+                    DataSetFilter ^datafilter_name = CreateFilter (test,3,"","", data_info[utility.getGlobalValue("terms.stop_codons")]);
+                }
+            }
+
 
             this_filter[utility.getGlobalValue("terms.data.coverage")] = utility.DictToArray(utility.Map(utility.Filter( ^ (this_filter[utility.getGlobalValue("terms.data.name")] + ".site_map"), "_value_", "_value_%3==0"), "_value_", "_value_$3"));
             filters + this_filter;

diff --git a/res/TemplateBatchFiles/libv3/tasks/mpi.bf b/res/TemplateBatchFiles/libv3/tasks/mpi.bf
@@ -337,6 +337,7 @@ namespace mpi {
         //#profile START;
 
         LFCompute (^lf_id, LF_START_COMPUTE);
+
 
         results = {};
         task_ids = utility.Keys (tasks);

diff --git a/src/core/batchlan.cpp b/src/core/batchlan.cpp
@@ -4770,6 +4770,7 @@ bool    _ElementaryCommand::ConstructDataSetFilter (_StringBuffer&source, _Execu
         } else if (operation_type == kBootstrap) {
             datafilter_command = new _ElementaryCommand(28);
         } else {
+
             throw _String ("Expected: DataSetFilter	  dataSetFilterid = CreateFilter (datasetid,unit,vertical partition,horizontal partition,alphabet exclusions); or Permute/Bootstrap (dataset/filter,<atom>,<column partition>)");
         }