Orthomcl split peripheral

Main/lib/perl/WorkflowSteps/CopyPeripheralGroupsResultsFromCache.pm

@@ -9,17 +9,43 @@ sub run {
 
   my $preprocessedDataCache = $self->getSharedConfig('preprocessedDataCache');
   my $workflowDataDir = $self->getWorkflowDataDir();
-  my $resultsDir = join("/", $workflowDataDir, $self->getParamValue("resultsDir"));
+  my $analysisDir = join("/", $workflowDataDir, $self->getParamValue("analysisDir"));
 
   if ($undo) {
-      $self->runCmd(0, "rm -rf $resultsDir/*");
+      $self->runCmd(0, "echo 'undo'");
   }
   elsif ($test) {
       $self->runCmd(0, "echo 'test'");
   }
   else {
 
-      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/* $resultsDir/");
+      # peripheralEntryResults
+      $self->runCmd(0, "mkdir -p $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/peripherals.fasta $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/intraGroupBlastFile.tsv $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/coreBestReps.txt $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/GroupsFile.txt $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/peripheralCacheDir.tar.gz $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/residuals.fasta $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/groupStats $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/residuals.fasta $analysisDir/peripheralEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/groupFastas $analysisDir/peripheralEntryResults/");
+      # postResidualEntryResults
+      $self->runCmd(0, "mkdir -p $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/groupStats $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/residualBestReps.txt $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/residualGroupFastas $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/residualBuildVersion.txt $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/intraResidualGroupBlastFile.tsv $analysisDir/postResidualEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/reformattedGroups.txt $analysisDir/postResidualEntryResults");
+      # postProcessingEntryResults
+      $self->runCmd(0, "mkdir -p $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/bestRepsFull.fasta $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/fullGroupFile.txt $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/similar_groups.tsv $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/ortho*db.dmnd $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/previousGroups.txt $analysisDir/postProcessingEntryResults/");
+      $self->runCmd(0, "cp -r ${preprocessedDataCache}/OrthoMCL/OrthoMCL_peripheralGroups/officialDiamondCache/geneTrees $analysisDir/postProcessingEntryResults/");
 
   }
 }

Main/lib/perl/WorkflowSteps/InsertOrthoMCLDerivedECGenomic.pm

@@ -8,11 +8,10 @@ use ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep;
 sub run {
   my ($self, $test, $undo) = @_;
 
-  my $inputFile = $self->getWorkflowDataDir() . "/" . $self->getParamValue('inputFile');
+  my $inputFile    = $self->getWorkflowDataDir() . "/" . $self->getParamValue('inputFile');
   my $evidenceCode = $self->getParamValue('evidenceCode');
-  my $idSql = $self->getParamValue('idSql');
 
-  my $args = "--ECMappingFile $inputFile --evidenceCode '$evidenceCode' --aaSeqLocusTagMappingSql '$idSql'";
+  my $args = "--ECMappingFile $inputFile --evidenceCode '$evidenceCode'";
 
   $self->testInputFile('inputFile', "$inputFile") unless $undo;
 

Main/lib/perl/WorkflowSteps/MakeOrthoFinderPeripheralNextflowConfig.pm → Main/lib/perl/WorkflowSteps/MakeOrthoFinderPeripheralEntryNextflowConfig.pm

@@ -1,4 +1,4 @@
-package ApiCommonWorkflow::Main::WorkflowSteps::MakeOrthoFinderPeripheralNextflowConfig;
+package ApiCommonWorkflow::Main::WorkflowSteps::MakeOrthoFinderPeripheralEntryNextflowConfig;
 
 @ISA = (ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep);
 
@@ -16,11 +16,10 @@ sub run {
   my $coreGroupsFile = $self->getParamValue("coreGroupsFile");
   my $coreGroupSimilarities = $self->getParamValue("coreGroupSimilarities");
   my $coreTranslateSequenceFile = $self->getParamValue("coreTranslateSequenceFile");
+  my $outdatedOrganisms = $self->getParamValue("outdatedOrganisms");  
+  my $peripheralDiamondCache = $self->getParamValue("peripheralDiamondCache");
+
   my $buildVersion = $self->getSharedConfig("buildVersion");
-  my $residualBuildVersion = $self->getSharedConfig("residualBuildVersion");
-  my $peripheralCacheDir = $self->getParamValue("peripheralCacheDir");
-  my $outdatedOrganisms = $self->getParamValue("outdated");
-  my $oldGroupsFile = $self->getParamValue("oldGroupsFile");
 
   my $resultsDirectory = $self->getParamValue("clusterResultDir");
   my $configPath = join("/", $self->getWorkflowDataDir(),  $self->getParamValue("analysisDir"), $self->getParamValue("configFileName"));
@@ -32,9 +31,8 @@ sub run {
   my $coreGroupsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreGroupsFile);
   my $coreGroupSimilaritiesInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreGroupSimilarities);
   my $coreTranslateSequenceFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreTranslateSequenceFile);
-  my $peripheralCacheDirInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $peripheralCacheDir);
-  my $outdatedFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $outdatedOrganisms);
-  my $oldGroupsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $oldGroupsFile);
+  my $peripheralDiamondCacheInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $peripheralDiamondCache);
+  my $outdatedOrganismsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $outdatedOrganisms);
   my $resultsDirectoryInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $resultsDirectory);
 
   my $executor = $self->getClusterExecutor();
@@ -54,15 +52,10 @@ params {
     coreGroupsFile = \"$coreGroupsFileInNextflowWorkingDirOnCluster\"
     coreGroupSimilarities = \"$coreGroupSimilaritiesInNextflowWorkingDirOnCluster\"
     coreTranslateSequenceFile = \"$coreTranslateSequenceFileInNextflowWorkingDirOnCluster\"
-    outdatedOrganisms = \"$outdatedFileInNextflowWorkingDirOnCluster\"
-    oldGroupsFile = \"$oldGroupsFileInNextflowWorkingDirOnCluster\"
-    peripheralDiamondCache = \"$peripheralCacheDirInNextflowWorkingDirOnCluster\"
-    blastArgs = \"\"
+    outdatedOrganisms = \"$outdatedOrganismsFileInNextflowWorkingDirOnCluster\"
+    peripheralDiamondCache = \"$peripheralDiamondCacheInNextflowWorkingDirOnCluster\"
     buildVersion = $buildVersion
-    residualBuildVersion = $residualBuildVersion
-    bestRepDiamondOutputFields = \"qseqid sseqid evalue\"
     orthoFinderDiamondOutputFields = \"qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore\"
-
 }
 
 process {

Main/lib/perl/WorkflowSteps/MakeOrthoFinderPostProcessingEntryNextflowConfig.pm

@@ -0,0 +1,88 @@
+package ApiCommonWorkflow::Main::WorkflowSteps::MakeOrthoFinderPostProcessingEntryNextflowConfig;
+
+@ISA = (ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep);
+
+use strict;
+use warnings;
+use ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep;
+
+sub run {
+  my ($self, $test, $undo) = @_;
+
+  my $clusterWorkflowDataDir = $self->getClusterWorkflowDataDir();
+  my $analysisDir = $self->getParamValue("analysisDir");
+  my $residualFasta = $self->getParamValue("residualFasta");
+  my $residualBestRepsFasta = $self->getParamValue("residualBestRepsFasta");
+  my $coreBestRepsFasta = $self->getParamValue("coreBestRepsFasta");
+  my $coreAndPeripheralGroups = $self->getParamValue("coreAndPeripheralGroups");
+  my $coreAndPeripheralProteome = $self->getParamValue("coreAndPeripheralProteome");
+  my $residualGroups = $self->getParamValue("residualGroups");
+  my $coreGroupFastas = $self->getParamValue("coreGroupFastas");
+  my $residualGroupFastas = $self->getParamValue("residualGroupFastas");
+  my $buildVersion = $self->getSharedConfig("buildVersion");
+  my $oldGroupsFile = $self->getParamValue("oldGroupsFile");
+  my $bestRepDiamondOutputFields = $self->getParamValue("bestRepDiamondOutputFields");  
+
+  my $resultsDirectory = $self->getParamValue("clusterResultDir");
+  my $configPath = join("/", $self->getWorkflowDataDir(),  $self->getParamValue("analysisDir"), $self->getParamValue("configFileName"));
+
+  my $workingDirRelativePath = $self->getParamValue("workingDirRelativePath");
+
+  my $residualFastaInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualFasta);
+  my $residualBestRepsFastaInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualBestRepsFasta);
+  my $coreBestRepsFastaInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreBestRepsFasta);
+  my $coreAndPeripheralGroupsInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreAndPeripheralGroups);
+  my $coreAndPeripheralProteomeInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreAndPeripheralProteome);
+  my $residualGroupsInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualGroups);
+  my $coreGroupFastasInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $coreGroupFastas);
+  my $residualGroupFastasInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualGroupFastas);
+  my $oldGroupsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $oldGroupsFile);
+
+  my $resultsDirectoryInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $resultsDirectory);
+
+  my $executor = $self->getClusterExecutor();
+  my $queue = $self->getClusterQueue();
+
+  if ($undo) {
+    $self->runCmd(0,"rm -rf $configPath");
+  } else {
+    open(F, ">", $configPath) or die "$! :Can't open config file '$configPath' for writing";
+
+    print F
+"
+params {
+    outputDir = \"$resultsDirectoryInNextflowWorkingDirOnCluster\"
+    residualFasta  = \"$residualFastaInNextflowWorkingDirOnCluster\"
+    residualBestRepsFasta = \"$residualBestRepsFastaInNextflowWorkingDirOnCluster\"
+    coreBestRepsFasta = \"$coreBestRepsFastaInNextflowWorkingDirOnCluster\"
+    coreAndPeripheralGroups = \"$coreAndPeripheralGroupsInNextflowWorkingDirOnCluster\"
+    coreAndPeripheralProteome = \"$coreAndPeripheralProteomeInNextflowWorkingDirOnCluster\"
+    residualGroups = \"$residualGroupsInNextflowWorkingDirOnCluster\"
+    coreGroupFastas = \"$coreGroupFastasInNextflowWorkingDirOnCluster\"
+    residualGroupFastas = \"$residualGroupFastasInNextflowWorkingDirOnCluster\"
+    buildVersion = $buildVersion
+    oldGroupsFile = \"$oldGroupsFileInNextflowWorkingDirOnCluster\"
+    bestRepDiamondOutputFields = \"$bestRepDiamondOutputFields\"
+}
+
+process {
+  executor = \'$executor\'
+  queue = \'$queue\'
+}
+
+env {
+  _JAVA_OPTIONS=\"-Xmx8192M\"
+  NXF_OPTS=\"-Xmx8192M\"
+  NXF_JVM_ARGS=\"-Xmx8192M\"
+}
+
+singularity {
+  enabled = true
+  autoMounts = true
+}
+";
+  close(F);
+ }
+}
+
+1;

Main/lib/perl/WorkflowSteps/MakeOrthoFinderPostResidualEntryNextflowConfig.pm

@@ -0,0 +1,75 @@
+package ApiCommonWorkflow::Main::WorkflowSteps::MakeOrthoFinderPostResidualEntryNextflowConfig;
+
+@ISA = (ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep);
+
+use strict;
+use warnings;
+use ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep;
+
+sub run {
+  my ($self, $test, $undo) = @_;
+
+  my $clusterWorkflowDataDir = $self->getClusterWorkflowDataDir();
+  my $analysisDir = $self->getParamValue("analysisDir");
+  my $residualFasta = $self->getParamValue("residualFasta");
+  my $groupsFile = $self->getParamValue("groupsFile");
+  my $speciesMapping = $self->getParamValue("speciesMapping");
+  my $sequenceMapping = $self->getParamValue("sequenceMapping");
+  my $diamondResultsFile = $self->getParamValue("diamondResultsFile");    
+  my $buildVersion = $self->getSharedConfig("buildVersion");
+  my $residualBuildVersion = $self->getSharedConfig("residualBuildVersion");
+
+  my $resultsDirectory = $self->getParamValue("clusterResultDir");
+  my $configPath = join("/", $self->getWorkflowDataDir(),  $self->getParamValue("analysisDir"), $self->getParamValue("configFileName"));
+
+  my $workingDirRelativePath = $self->getParamValue("workingDirRelativePath");
+
+  my $residualFastaInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualFasta);
+  my $groupsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $groupsFile);
+  my $speciesMappingInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $speciesMapping);
+  my $sequenceMappingInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $sequenceMapping);  
+  my $diamondResultsFileInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $diamondResultsFile);
+  my $resultsDirectoryInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $resultsDirectory);
+
+  my $executor = $self->getClusterExecutor();
+  my $queue = $self->getClusterQueue();
+
+  if ($undo) {
+    $self->runCmd(0,"rm -rf $configPath");
+  } else {
+    open(F, ">", $configPath) or die "$! :Can't open config file '$configPath' for writing";
+
+    print F
+"
+params {
+    outputDir = \"$resultsDirectoryInNextflowWorkingDirOnCluster\"
+    residualFasta = \"$residualFastaInNextflowWorkingDirOnCluster\"
+    groupsFile = \"$groupsFileInNextflowWorkingDirOnCluster\"
+    speciesMapping = \"$speciesMappingInNextflowWorkingDirOnCluster\"
+    sequenceMapping = \"$sequenceMappingInNextflowWorkingDirOnCluster\"
+    diamondResultsFile = \"$diamondResultsFileInNextflowWorkingDirOnCluster\"
+    buildVersion = $buildVersion
+    residualBuildVersion = $residualBuildVersion
+}
+
+process {
+  executor = \'$executor\'
+  queue = \'$queue\'
+}
+
+env {
+  _JAVA_OPTIONS=\"-Xmx8192M\"
+  NXF_OPTS=\"-Xmx8192M\"
+  NXF_JVM_ARGS=\"-Xmx8192M\"
+}
+
+singularity {
+  enabled = true
+  autoMounts = true
+}
+";
+  close(F);
+ }
+}
+
+1;

Main/lib/perl/WorkflowSteps/MakeOrthoFinderResidualEntryNextflowConfig.pm

@@ -0,0 +1,61 @@
+package ApiCommonWorkflow::Main::WorkflowSteps::MakeOrthoFinderResidualEntryNextflowConfig;
+
+@ISA = (ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep);
+
+use strict;
+use warnings;
+use ApiCommonWorkflow::Main::WorkflowSteps::WorkflowStep;
+
+sub run {
+  my ($self, $test, $undo) = @_;
+
+  my $clusterWorkflowDataDir = $self->getClusterWorkflowDataDir();
+  my $analysisDir = $self->getParamValue("analysisDir");
+  my $residualFastaDir = $self->getParamValue("residualFastaDir");
+
+  my $resultsDirectory = $self->getParamValue("clusterResultDir");
+  my $configPath = join("/", $self->getWorkflowDataDir(),  $self->getParamValue("analysisDir"), $self->getParamValue("configFileName"));
+
+  my $workingDirRelativePath = $self->getParamValue("workingDirRelativePath");
+
+  my $residualFastaDirInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $residualFastaDir);
+  my $resultsDirectoryInNextflowWorkingDirOnCluster = $self->relativePathToNextflowClusterPath($workingDirRelativePath, $resultsDirectory);
+
+  my $executor = $self->getClusterExecutor();
+  my $queue = $self->getClusterQueue();
+
+  if ($undo) {
+    $self->runCmd(0,"rm -rf $configPath");
+  } else {
+    open(F, ">", $configPath) or die "$! :Can't open config file '$configPath' for writing";
+
+    print F
+"
+params {
+    outputDir = \"$resultsDirectoryInNextflowWorkingDirOnCluster\"
+    residualFastaDir = \"$residualFastaDirInNextflowWorkingDirOnCluster\"
+    orthoFinderDiamondOutputFields = \"qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore\"
+
+}
+
+process {
+  executor = \'$executor\'
+  queue = \'$queue\'
+}
+
+env {
+  _JAVA_OPTIONS=\"-Xmx8192M\"
+  NXF_OPTS=\"-Xmx8192M\"
+  NXF_JVM_ARGS=\"-Xmx8192M\"
+}
+
+singularity {
+  enabled = true
+  autoMounts = true
+}
+";
+  close(F);
+ }
+}
+
+1;

Main/lib/perl/WorkflowSteps/PredictEcNumbers.pm

@@ -9,15 +9,15 @@ sub run {
   my ($self, $test, $undo) = @_;
 
   my $workflowDataDir = $self->getWorkflowDataDir();
-  my $outputDir  = $self->getParamValue('outputDir');
-  my $outputFile  = $workflowDataDir."/".$self->getParamValue('outputFile');
+  my $outputFile    = $workflowDataDir . "/" . $self->getParamValue('outputFile');
   my $gusConfigFile = $workflowDataDir . "/" . $self->getParamValue('gusConfigFile');
+  my $threshold     = 0.4;
 
   if ($undo) {
-    my $cmd = "rm $outputFile";
+    my $cmd = "rm -f $outputFile";
     $self->runCmd($test, $cmd);
   } else {
-    my $cmd = "orthomclEcPrediction $outputDir $outputFile $gusConfigFile";
+    my $cmd = "assignEcByOrthologs.pl --output $outputFile --gusConfigFile $gusConfigFile --threshold $threshold";
     $self->runCmd($test, $cmd);
   }
 }