Addidng parallel clustering to the density module

ding-lab · Mar 7, 2017 · 200fe3c · 200fe3c
1 parent b9d4835
commit 200fe3c
Show file tree

Hide file tree

Showing 6 changed files with 69 additions and 13 deletions.
diff --git a/HotSpot3D-1.5.0.tar.gz b/HotSpot3D-1.5.0.tar.gz
diff --git a/README.md b/README.md
@@ -10,7 +10,7 @@ Usage
 
          Stable:     v0.6.0 
 
-           Beta:     up to v1.4.1
+           Beta:     up to v1.5.0
         
          Author:     Beifang Niu, John Wallis, Adam D Scott, Sohini Sengupta, & Amila Weerasinghe
 

diff --git a/bin/hotspot3d b/bin/hotspot3d
@@ -8,7 +8,7 @@
 use strict;
 use warnings;
 
-our $VERSION = 'V1.4.1';
+our $VERSION = 'V1.5.0';
 
 use Carp;
 use FileHandle;

diff --git a/dist.ini b/dist.ini
@@ -1,6 +1,6 @@
 name    = HotSpot3D
 author  = Beifang Niu, John Wallis, Adam D Scott, Sohini Sengupta, Amila Weerasinghe, & Matthew H Bailey from McDonnell Genome Institute of Washington University at St. Louis
-version = 1.4.1
+version = 1.5.0
 license = Perl_5
 copyright_holder = McDonnell Genome Institute at Washington University
 copyright_year   = 2017

diff --git a/lib/TGI/Mutpro/Main/Cluster.pm b/lib/TGI/Mutpro/Main/Cluster.pm
@@ -1459,6 +1459,8 @@ Usage: hotspot3d cluster [options]
 --weight-header              .maf file column header for mutation weight, default: weight (used if vertex-type = weight)
 --parallel                   Parallelization for structure and subunit dependent runs (none or local), default: none
 --max-processes              Set if using parallel type local (CAUTION: make sure you know your max CPU processes)
+--gene-list-file             Choose mutations from the genes given in this list
+--structure-list-file        Choose mutations from the structures given in this list
 
 --help                       this message
 

diff --git a/lib/TGI/Mutpro/Main/Density.pm b/lib/TGI/Mutpro/Main/Density.pm
@@ -29,6 +29,8 @@ my $SHORTESTDISTANCE = "shortest";
 my $AVERAGEDISTANCE = "average";
 my $INDEPENDENT = "independent";
 my $DEPENDENT = "dependent";
+my $LOCAL = "local";
+my $NONE = "none";
 
 sub new {
     my $class = shift;
@@ -86,13 +88,45 @@ sub process {
 
     #####################################################
 
-    my $pairwiseFN = "$this->{'distance_measure'}.$this->{pairwise_file_name_only}"; 
-
     # print "distance matrix\n";
     # print Dumper $distance_matrix;
     # print "mutations\n";
     # print Dumper $mutations;
 
+    $this->densityClustering( $mutations, $distance_matrix );
+
+    my $numStructure = scalar keys %{ $distance_matrix };
+    print "Density-based clusters are being calculated for $numStructure structures.\n\n";
+
+}
+
+#####
+#   Functions
+#####
+
+sub densityClustering {
+    my ( $this, $mutations, $distance_matrix ) = @_;
+
+    print STDOUT "HotSpot3D::Cluster::Density::densityClustering\n";
+
+    if ( $this->{'parallel'} eq $LOCAL ) {
+        $this->localParallelDensityClustering( $mutations , $distance_matrix );
+    #} elsif ( $this->{'parallel'} eq $BSUB ) {
+    #   $this->bsubParallelNetworkClustering( $clusterings , $mutations , $distance_matrix );
+    } else {
+        $this->noParallelDensityClustering( $mutations , $distance_matrix );
+    }
+    return;
+}
+
+sub localParallelDensityClustering {
+    my ( $this , $mutations , $distance_matrix ) = @_;
+
+    my $pairwiseFN = "$this->{'distance_measure'}.$this->{pairwise_file_name_only}"; 
+    print STDOUT "\tParallel clustering over structures with up to ".$this->{'max_processes'}." processes\n";
+    my $pm = Parallel::ForkManager->new( $this->{'max_processes'} );
+
+    DATA_LOOP:
     foreach my $structure ( keys %{$distance_matrix} ) { # run the density calculation for each available structure
         #print "Structure= $structure\n";
         # name output files as *.$pdbID.structure.*
@@ -105,21 +139,41 @@ sub process {
         $this->MainOPTICS( $structure , $distance_matrix, $mutations ); # perform OPTICS for the first time
         $this->RunSuperClustersID(); # perform Clustering for the reference run
 
-        print "Reference run: Done.\nStart probability calculation\n";
+        print "Reference run: Done.\nStart probability calculation for $structure\n";
 
         $this->getClusterProbabilities( $structure , $distance_matrix, $mutations ); # perform cluster-membership probability calculation
 
-        print "\nProbability Calculation is Done.\n\n";
+        print "\nProbability Calculation is Done for $structure.\n\n";
     }
+    $pm->wait_all_children;
 
-    my $numStructure = scalar keys %{ $distance_matrix };
-    print "Density-based clusters are being calculated for $numStructure structures.\n\n";
-
+    return;
 }
 
-#####
-#   Functions
-#####
+sub noParallelDensityClustering {
+    my ( $this , $mutations , $distance_matrix ) = @_;
+    print "Serially clustering over structures\n";
+
+    my $pairwiseFN = "$this->{'distance_measure'}.$this->{pairwise_file_name_only}"; 
+    foreach my $structure ( keys %{$distance_matrix} ) { # run the density calculation for each available structure
+        #print "Structure= $structure\n";
+        # name output files as *.$pdbID.structure.*
+        $this->{pairwise_file_name_only} = "$structure.Structure.$pairwiseFN";
+
+        # call the SetOfNodes hash for each structure
+        $this->{"CurrentSetOfNodes"} = $distance_matrix->{$structure};
+
+        ###### Reference run: start
+        $this->MainOPTICS( $structure , $distance_matrix, $mutations ); # perform OPTICS for the first time
+        $this->RunSuperClustersID(); # perform Clustering for the reference run
+
+        print "Reference run: Done.\nStart probability calculation for $structure\n";
+
+        $this->getClusterProbabilities( $structure , $distance_matrix, $mutations ); # perform cluster-membership probability calculation
+
+        print "\nProbability Calculation is Done for $structure.\n\n";
+    }
+}
 
 sub MainOPTICS {
     my $this = shift;