/* * BioJava development code * * This code may be freely distributed and modified under the * terms of the GNU Lesser General Public Licence. This should * be distributed with the code. If you do not have a copy, * see: * * http://www.gnu.org/copyleft/lesser.html * * Copyright for this code is held jointly by the individual * authors. These should be listed in @author doc comments. * * For more information on the BioJava project and its aims, * or to join the biojava-l mailing list, visit the home page * at: * * http://www.biojava.org/ * */ package org.biojava.bio.seq.io; import java.io.*; import java.util.*; import org.biojava.bio.*; import org.biojava.bio.seq.*; import org.biojava.bio.seq.impl.*; import org.biojava.bio.seq.io.*; import org.biojava.bio.seq.genomic.*; import org.biojava.bio.symbol.*; import org.biojava.utils.*; /** * Translate Sequence object into sgv file * @author Hanning Ni Doubletwist Inc, 2001-7-22 */ public class SVGGenerator { /** source code version */ public static final String VER_ID = "$Header: $"; // this part should not be hard coded. will change later ... public static final String CONTAMINATION = "contamination" ; public static final String REPEAT= "repeat" ; public static final int LABEL_LENGTH = 50 ; public static final int WINDOW_WIDTH = 800 + LABEL_LENGTH; public static final int WINDOW_HEIGHT = 600 ; /** The XML writer */ protected PrintWriter mOut; //define some global variable, some of them will be removed in the future /** the height of each row */ protected int row_height = 0 ; /** ratio of sequence length to WINDOW_WIDTH */ protected float ratio = 0.0f ; /** contains label of sequence_map --> the maximum rows this label will occupy */ Hashtable table_forward ; Hashtable table_complement ; /** contain label of sequence_map --> the i-th of row in view */ Hashtable forward_row_loc ; Hashtable complement_row_loc ; /** contains all the components which will be draw in the final view */ Vector comp2svg ; private int mId ; private int total_length ; private int total_rows ; /** * Constructor. * @param out Writer for the XML output. */ public SVGGenerator(Writer out) { mOut = new PrintWriter(out); mId = 0 ; total_length = 0 ; total_rows = 0 ; table_forward = new Hashtable() ; table_complement = new Hashtable(); forward_row_loc = new Hashtable() ; complement_row_loc = new Hashtable(); comp2svg = new Vector() ; } public void close() { mOut.close() ; } /** * Write XML version information and DOCTYPE tag for svg */ private void writeDocumentType() { mOut.println(""); mOut.println(""); } /** * Write object. * */ public void write(Sequence seq) throws IOException { try { init( seq ) ; writeDocumentType() ; writeHeader( ); writeBody( ); writeFooter(); } catch(Exception e) { throw new IOException(e.getMessage() ); }finally{ clear() ; } } private void init(Sequence sequence) throws Exception { //get the total dna length total_length = getLength( sequence ) ; //get the total rows which will show in the view total_rows = getRows( sequence ) ; row_height = WINDOW_HEIGHT / total_rows ; ratio = WINDOW_WIDTH / (float) total_length ; } private void clear() { mId = 0 ; total_length = 0 ; total_rows = 0 ; row_height = 0 ; ratio = 0.0f ; table_forward.clear(); table_complement.clear(); forward_row_loc.clear(); complement_row_loc.clear(); comp2svg.clear() ; } /** dump the svg header which defines how to draw each type of gene component */ private void writeHeader( ) throws Exception { //currently we use one basepair <-> one pixel // LABEL_LENGTH = (total_length / 5 > 50 ? total_length /5 : 50 ) ; int x = WINDOW_WIDTH ; // total_length + LABEL_LENGTH ; //total_row does not include dna row yet int y = WINDOW_HEIGHT ; // ( total_rows + 1 ) * ROW_HEIGHT ; String head ="" ; mOut.println( head ); //javascript here mOut.println(" " ) ; mOut.println("") ; mOut.println(" ") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println(" ") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println(" ") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; mOut.println("") ; } /** * first write all the labels of sequence_map * then write all the gene component * pre-condition : after call write_header, which all ready store all the infomation into Vector */ private void writeBody() { //write all the labels write_label() ; //write all the compoments for(int i = 0 ; i < comp2svg.size() ; i ++ ) { //can be dna fragment, gene , exon and intron Component comp = (Component) comp2svg.elementAt(i) ; //add the length of label comp.start_x = Math.round( comp.start_x * ratio ) + LABEL_LENGTH ; //calculate the real start_y value , which need the information about algorihtm type and row if( comp.isForward && forward_row_loc.get( comp.row_type ) != null ) comp.start_y = ( (Integer) forward_row_loc.get( comp.row_type )).intValue() * row_height ; else if(( ! comp.isForward ) && complement_row_loc.get( comp.row_type ) != null ) comp.start_y = ( (Integer) complement_row_loc.get( comp.row_type )).intValue() * row_height ; //now , dumping // System.out.println("writing : " + comp.start_y + " : " + comp.row_type ); mOut.println( comp.toString() ) ; } } /** * first calculate all the location of sequence_map in the final view * then dumping */ private void write_label() { // calculate all the location of sequence_map cal_row_loc() ; //dumping the labels in the forward strand, include DNA Enumeration nums = forward_row_loc.keys() ; while( nums.hasMoreElements() ) { String this_label = (String) nums.nextElement() ; int row = ( (Integer) forward_row_loc.get( this_label )).intValue() ; print_text( this_label, 0 , row * row_height ) ; } //dumping the labels in the complement strand nums = complement_row_loc.keys() ; while( nums.hasMoreElements() ) { String this_label = (String) nums.nextElement() ; int row = ( (Integer) complement_row_loc.get( this_label )).intValue() ; print_text( this_label, 0 , row * row_height ) ; } } /**** * dumping the svg text **/ private void print_text(String label, int x, int y) { int id = mId ++ ; mOut.println( " " + "" + label + "" + "" ); } /** * calculate the real location of each sequence_map * must put nr pro into the out most level * and it is nice to put comtamination /repeat /genbank near dna sequence */ private void cal_row_loc() { //write label first (dont care ) int row = 0 ; Object o ; String contamination =""; String repeat = "" ; Enumeration nums = table_forward.keys() ; while( nums.hasMoreElements() ) { String this_label = (String) nums.nextElement() ; if( this_label.indexOf( CONTAMINATION ) != -1 ) { contamination = this_label ; continue ; } if( this_label.indexOf( REPEAT ) != -1 ) { repeat = this_label ; continue ; } o = table_forward.get( this_label ) ; forward_row_loc.put( this_label , new Integer( row ) ) ; //System.out.println(this_label + " : " + row ) ; row += ((Integer)o ).intValue() ; } //put contamination , repeat near dna o = table_forward.get( contamination ) ; if( o != null ) { forward_row_loc.put( contamination , new Integer( row ) ) ; row += ((Integer)o).intValue() ; } o = table_forward.get( repeat ) ; if( o != null ) { forward_row_loc.put( repeat , new Integer( row ) ) ; //System.out.println(REPEAT + " : " + row ) ; row += ((Integer)o).intValue() ; } //put dna into forward_row_loc for easy use. forward_row_loc.put( "DNA", new Integer( row ) ) ; //System.out.println("DNA" + " : " + row ) ; row ++ ; //put complement label in the reverse order o = table_complement.get( repeat ) ; if( o != null ) { complement_row_loc.put( repeat , new Integer( row ) ) ; //System.out.println(REPEAT + " -: " + row ) ; row += ((Integer)o).intValue() ; } o = table_complement.get( contamination ) ; if( o != null ) { complement_row_loc.put( contamination , new Integer( row ) ) ; row += ((Integer)o).intValue() ; } nums = table_complement.keys() ; while( nums.hasMoreElements() ) { String this_label = (String) nums.nextElement() ; if( this_label.indexOf( CONTAMINATION ) != -1 || this_label.indexOf( REPEAT ) != -1 ) { continue ; } o = table_complement.get( this_label ) ; complement_row_loc.put( this_label , new Integer( row ) ) ; //System.out.println(this_label + " -: " + row ) ; row += ((Integer)o ).intValue() ; } } /** * Write </sciobj> */ private void writeFooter() { mOut.println(""); } /** * get the total length of dna sequence * */ private int getLength(Sequence seq) throws Exception { return seq.length() ; } /** * actaully two function together, it is not good for design, just for performance * . (1) get the maximun rows each feature * (2) put every gene component into list, for easy dumping */ private int getRows(Sequence seq) throws Exception { int rows = 0 ; int length = 0 ; /** store seq_map --> max rows **/ if( seq instanceof SimpleAssembly) cal_sequence_row((SimpleAssembly) seq, length) ; else cal_sequence_row( seq , length) ; //sum up the total rows Enumeration nums = table_forward.keys() ; while( nums.hasMoreElements() ) { rows += ((Integer)table_forward.get( nums.nextElement() ) ).intValue() ; } nums = table_complement.keys() ; while( nums.hasMoreElements() ) { rows += ((Integer)table_complement.get( nums.nextElement() ) ).intValue() ; } return rows ; } private void cal_sequence_row(Sequence seq, int length) { String prop = annot2string( seq.getAnnotation() ) ; int x = 0 ; int y = -1 ; int len = seq.length() ; String row_type = "DNA" ; comp2svg.addElement( new Component(x, y, len, Component.FORWARD, prop, row_type, true, mId++) ); cal_feature_row( seq, length) ; } private void cal_sequence_row(SimpleAssembly seq , int length) { for(Iterator i = seq.features(); i.hasNext(); ) { ComponentFeature child = (ComponentFeature) i.next(); int type = (( child.getStrand() == StrandedFeature.NEGATIVE ) ? Component.REVERSE : Component.FORWARD ); String prop = annot2string( child.getAnnotation() ) ; int x = child.getLocation().getMin() ; int y = -1 ; int len = seq.length() ; String row_type = "DNA" ; comp2svg.addElement( new Component(x, y, len, type, prop, row_type, true, mId++) ); cal_feature_row( child.getComponentSequence(), x ) ; } } /** * return : feature source -> list of features */ private HashMap clusterSubFeatures(FeatureHolder feature) { HashMap maps = new HashMap(); for(Iterator i = feature.features(); i.hasNext(); ) { Feature child = (Feature) i.next(); String source = child.getType() ; if( source == null ) source = child.getSource() ; if( source == null ) source = "UNKOWN" ; // System.out.println( source ) ; Object o = maps.get( source ) ; if( o == null ) { List list = new ArrayList() ; list.add (child ) ; maps.put( source, list ) ; }else{ ((List)o).add( child ) ; } } // System.out.println( maps.size()); return maps; } /** * (1) calculate the rows for contig * (2) put info of each genecomponent inot Component */ private void cal_feature_row(Sequence seq, int length) { //decide strandness HashMap maps = clusterSubFeatures((FeatureHolder)seq ) ; for(Iterator it = maps.keySet().iterator(); it.hasNext();) { String label = (String) it.next() ; cal_map_row( (List) maps.get( label ), label , length) ; } } /** * (1) calculate the max row in this fragment * (2) store all the gene /exon of this map type in this fragment */ private void cal_map_row(List list, String label , int length) { //first store sequence_map label at this level Vector regions_for = new Vector() ; Vector regions_com = new Vector() ; Iterator fs = list.iterator() ; while(fs.hasNext()) { Feature f = (Feature) fs.next(); cal_map_row(f, regions_for, regions_com, label, length) ; } // get the maximun rows in this fragment int rows = cal_row_each_map_in_fragment( regions_for ) ; Object o = table_forward.get( label ) ; //store the max value for all BAC if( rows >=1 ) { if( o != null ) { if( ((Integer)o).intValue() < rows ) table_forward.put( label, new Integer( rows )) ; } else table_forward.put( label, new Integer( rows ) ); } // get the maximun rows in this fragment rows = cal_row_each_map_in_fragment( regions_com ) ; o = table_complement.get( label ) ; //store the max value for all BAC if( rows >=1 ) { if( o != null ) { if( ((Integer)o).intValue() < rows ) table_complement.put( label, new Integer( rows )) ; } else table_complement.put( label, new Integer( rows ) ); } } private void cal_map_row(Feature f, Vector region_for, Vector region_com, String label, int length) { Location loc = f.getLocation() ; boolean forward = true ; if( f instanceof StrandedFeature ) { forward = (((StrandedFeature)f).getStrand() != StrandedFeature.NEGATIVE ) ; } if( forward ) region_for.addElement( loc ); else region_com.addElement( loc ); dump_result( f, length, label , forward) ; for(Iterator it = f.features(); it.hasNext() ;) { cal_map_row( (Feature) it.next() , region_for, region_com, label, length) ; } } /** * store the info of comp_result into Component * currently I only choose some info. for final show */ private void dump_result(Feature f, int length, String row_type, boolean forward) { String prop = annot2string( f.getAnnotation() ) ; int start = f.getLocation().getMin() ; int end = f.getLocation().getMax() ; String type = f.getType() ; int x = length + start; int y = -1 ; int len = end - start + 1 ; int comtype = Component.EXON ; //default value if( type.indexOf("gene") != -1 || f instanceof Gene ) comtype = Component.GROUP ; else if( type.indexOf("exon") != -1 || f instanceof Exon) comtype = Component.EXON ; else if( type.indexOf("intron") != -1 ) comtype = Component.INTRON ; comp2svg.addElement( new Component(x, y, len, comtype, prop, row_type, forward, mId++) ); } /** cal the #rows this set of regions will occupy */ private int cal_row_each_map_in_fragment( Vector regions ) { if( regions.size() <2 ) return regions.size() ; return cal_bins_for_one_vector( regions ).size() ; } /** regions must have at least two elements * algorithm : *

init : create one bin, and put first region into it. *

for each region aa in regions-A *

for each bin bb in bins-B *

check if aa can put into bb withou overlap any element *

if can, put aa into bb and break ; *

if can not, create new bin bb-new ,and put bb-new into bins-B *

put aa into bb-new *

finally, the number of bins in bins-B is the number of rows *

need. *

**/ private Vector cal_bins_for_one_vector(Vector regions ) {//System.out.println("entering bin : " + regions.size()); //init Vector bins = new Vector() ; Vector one_bin = new Vector() ; one_bin.addElement( regions.elementAt(0) ) ; bins.addElement( one_bin ) ; //loop... try{ for(int i = 1 ; i < regions.size() ; i++) { Location region = (Location) regions.elementAt(i) ; int start = region.getMin() ; int end = region.getMax() ; boolean found_bin = false ; //check the already created bin for( int j = 0 ; j < bins.size() ; j++) { Vector this_bin = (Vector) bins.elementAt( j ) ; //check if region can be put into this bin boolean overlap = false ; for(int k = 0 ; k < this_bin.size() ; k++) { Location this_region = (Location) this_bin.elementAt( k ) ; int ov = Math.min(end, this_region.getMax() ) - Math.max(start, this_region.getMin() ); if( ov > 0 ) //can not fit into this bin { overlap = true ; break ; } } if( ! overlap ) //can fit into this bin { this_bin.addElement( region ) ; found_bin = true ; break ; } } if( ! found_bin ) //can not found a suitable bin, so create a new one.. { Vector new_bin = new Vector() ; new_bin.addElement( region ) ; bins.addElement( one_bin ) ; } } }catch(Exception e ){} // System.out.println("leaving bin : " + bins.size()); return bins ; } private String annot2string(Annotation seqAn) { StringBuffer sb = new StringBuffer() ; for (Iterator i = seqAn.keys().iterator(); i.hasNext(); ) { Object key = i.next(); Object value = seqAn.getProperty(key); sb.append( key + " : " + filter( value.toString() ) + "\n") ; } return sb.toString(); } private String filter( String target) { StringBuffer sb = new StringBuffer( ) ; for(int i = 0 ; i < target.length() ; i++) { char chr = target.charAt(i) ; if( chr == '>' ) sb.append(">") ; else if ( chr == '<' ) sb.append("<") ; else if ( chr == '&' ) sb.append("&") ; else sb.append(chr); } return sb.toString() ; } /** * private class used for store location info for each gene components * for performance issue, I use all the public variables */ class Component { //start point in view public int start_x ; //start point in view public int start_y ; //should be label of sequence_map public String row_type ; //should be type of genecomponent public int type ; public int id; public static final int GROUP = 0 ; public static final int EXON = 1 ; public static final int INTRON = 2 ; public static final int FORWARD = 3; public static final int REVERSE = 4 ; public static final int UNORDER = 5 ; //should be the length of this component public int length ; //currently used for view to show the info public String property ; //which side public boolean isForward ; //construct.. public Component(int x, int y, int len, int type, String prop , String row, boolean forward, int id){ start_x = x; start_y = y ; length = len ; this.type = type ; property = prop ; row_type = row ; isForward = forward; this.id = id ; } //help method to get the type of svg element private String getSVGType(int type) { String s = " "; switch(type) { case GROUP : s = "genebox"; break; case EXON : s = "exonbox"; break; case INTRON : s = "intronbox"; break; case FORWARD : s = "forwardfragment"; break; case REVERSE : s = "reversefragment"; break; case UNORDER : s = "unorderedfragment"; break; } return s; } public String toString() { String svgtype = getSVGType( type ) ; int len = ( length / 10 > 1 ? length/10 : 1 ) ; StringBuffer prop = new StringBuffer("") ; int index1 , index2; String breakline = System.getProperty( "line.separator" ) ; index1 = index2 = 0 ; if( property != null ) if( property.indexOf("\n") == -1 ) prop.append( property ) ; else while( (index1 = property.indexOf("\n", index2) ) != -1 ) { prop.append( "" + property.substring(index2, index1) + "\n" ) ; index2 = index1 + 1 ; } return " " + "\"" + "" + "" + "" + prop.toString() + "" + "" ; } } }