in progress

[jalview.git] / forester / java / src / org / forester / application / surfacing.java

diff --git a/forester/java/src/org/forester/application/surfacing.java b/forester/java/src/org/forester/application/surfacing.java
index fee80e0..4218c4b 100644 (file)
--- a/forester/java/src/org/forester/application/surfacing.java
+++ b/forester/java/src/org/forester/application/surfacing.java
@@ -33,6 +33,7 @@ import java.io.IOException;
 import java.io.Writer;
 import java.util.ArrayList;
 import java.util.Date;
+import java.util.HashMap;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
@@ -98,6 +99,7 @@ import org.forester.util.ForesterUtil;
 
 public class surfacing {
 
+    private static final int                                  MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING                           = 1000;
     public final static String                                DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS                    = "graph_analysis_out";
     public final static String                                DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS                = "_dc.dot";
     public final static String                                PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_fitch_present_dc.dot";
@@ -233,7 +235,7 @@ public class surfacing {
     final static private String                               SEQ_EXTRACT_OPTION                                                     = "prot_extract";
     final static private char                                 SEPARATOR_FOR_INPUT_VALUES                                             = '#';
     final static private String                               PRG_VERSION                                                            = "2.210";
-    final static private String                               PRG_DATE                                                               = "2011.11.25";
+    final static private String                               PRG_DATE                                                               = "2011.12.08";
     final static private String                               E_MAIL                                                                 = "czmasek@burnham.org";
     final static private String                               WWW                                                                    = "www.phylosoft.org/forester/applications/surfacing";
     final static private boolean                              IGNORE_DUFS_DEFAULT                                                    = true;
@@ -2073,11 +2075,19 @@ public class surfacing {
         DescriptiveStatistics pw_stats = null;
         try {
             String my_outfile = output_file.toString();
-            if ( !my_outfile.endsWith( ".html" ) ) {
+            Map<Character, Writer> split_writers = null;
+            Writer writer = null;
+            if ( similarities.size() > MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING ) {
+                if ( my_outfile.endsWith( ".html" ) ) {
+                    my_outfile = my_outfile.substring( 0, my_outfile.length() - 5 );
+                }
+                split_writers = new HashMap<Character, Writer>();
+                createSplitWriters( out_dir, my_outfile, split_writers );
+            }
+            else if ( !my_outfile.endsWith( ".html" ) ) {
                 my_outfile += ".html";
+                writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
             }
-            final Writer writer = new BufferedWriter( new FileWriter( out_dir == null ? my_outfile : out_dir
-                    + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
             List<Species> species_order = null;
             if ( species_matrix ) {
                 species_order = new ArrayList<Species>();
@@ -2097,6 +2107,7 @@ public class surfacing {
                     .writeDomainSimilaritiesToFile( html_desc,
                                                     new StringBuilder( number_of_genomes + " genomes" ),
                                                     writer,
+                                                    split_writers,
                                                     similarities,
                                                     number_of_genomes == 2,
                                                     species_order,
@@ -2363,6 +2374,65 @@ public class surfacing {
         System.out.println();
     }
 
+    private static void createSplitWriters( final File out_dir,
+                                            final String my_outfile,
+                                            final Map<Character, Writer> split_writers ) throws IOException {
+        split_writers.put( 'a', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_A.html" ) ) );
+        split_writers.put( 'b', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_B.html" ) ) );
+        split_writers.put( 'c', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_C.html" ) ) );
+        split_writers.put( 'd', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_D.html" ) ) );
+        split_writers.put( 'e', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_E.html" ) ) );
+        split_writers.put( 'f', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_F.html" ) ) );
+        split_writers.put( 'g', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_G.html" ) ) );
+        split_writers.put( 'h', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_H.html" ) ) );
+        split_writers.put( 'i', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_I.html" ) ) );
+        split_writers.put( 'j', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_J.html" ) ) );
+        split_writers.put( 'k', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_K.html" ) ) );
+        split_writers.put( 'l', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_L.html" ) ) );
+        split_writers.put( 'm', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_M.html" ) ) );
+        split_writers.put( 'n', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_N.html" ) ) );
+        split_writers.put( 'o', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_O.html" ) ) );
+        split_writers.put( 'p', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_P.html" ) ) );
+        split_writers.put( 'q', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_Q.html" ) ) );
+        split_writers.put( 'r', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_R.html" ) ) );
+        split_writers.put( 's', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_S.html" ) ) );
+        split_writers.put( 't', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_T.html" ) ) );
+        split_writers.put( 'u', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_U.html" ) ) );
+        split_writers.put( 'v', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_V.html" ) ) );
+        split_writers.put( 'w', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_W.html" ) ) );
+        split_writers.put( 'x', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_X.html" ) ) );
+        split_writers.put( 'y', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_Y.html" ) ) );
+        split_writers.put( 'z', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_Z.html" ) ) );
+        split_writers.put( '0', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+                + "_domains_0.html" ) ) );
+    }
+
     private static void printOutPercentageOfMultidomainProteins( final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
                                                                  final Writer log_writer ) {
         int sum = 0;