DESCHRAMBLER

DESCHRAMBLER.pl

@@ -0,0 +1,81 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+use FindBin qw($Bin);
+use Cwd;
+use Cwd 'abs_path';
+
+# check the number of argument
+if ($#ARGV+1 != 1) {
+	print STDERR "Usage: ./DESCHRAMBLER.pl <parameter file>\n";
+	exit(1);
+}
+
+my $params_f = $ARGV[0];
+
+# parse parameter file
+my %params = ();
+open(F,"$params_f");
+while(<F>) {
+	chomp;
+	my $line = trim($_);
+	if ($line =~ /^#/ || $line eq "") { next; }
+	my ($name, $value) = split(/=/);
+	$name = trim($name);
+	$value = trim($value);
+	if (-f $value || -d $value) {
+		$params{$name} = abs_path($value);
+	} else {
+		$params{$name} = $value;
+	}
+}
+close(F);
+
+check_parameters(\%params);
+
+my $sf_dir = $params{"OUTPUTDIR"}."/SFs";
+`mkdir -p $params{"OUTPUTDIR"}`;
+
+# make blocks
+print STDERR "\n## Constructing syntenic fragments ##\n"; 
+my $cwd = getcwd();
+`mkdir -p $sf_dir`;
+`sed -e 's:<resolutionwillbechanged>:$params{"RESOLUTION"}:' $params{"CONFIGSFSFILE"} > $sf_dir/config.file`;
+`sed -e 's:<willbechanged>:$Bin/code/makeBlocks:;s:<treewillbechanged>:$params{"TREEFILE"}:' $params{"MAKESFSFILE"} > $sf_dir/Makefile`;
+#}
+chdir($sf_dir);
+`make all`;
+
+chdir($cwd);
+`$Bin/script/create_blocklist.pl $params{"REFSPC"} $sf_dir`; 
+
+# reconstruct APCFs
+`$Bin/script/wrap_recon_apcf.pl $params{"TREEFILE"} $params{"RESOLUTION"} $params{"REFSPC"} $params{"MINADJSCR"} $sf_dir $params{"OUTPUTDIR"}`; 
+
+###############################################################
+sub check_parameters {
+	my $rparams = shift;
+	my $flag = 0;
+	my $out = "";
+	my @parnames = ("REFSPC","OUTPUTDIR","RESOLUTION","TREEFILE","CONFIGSFSFILE","MAKESFSFILE","MINADJSCR"); 
+
+	foreach my $pname (@parnames) {
+		if (!defined($$rparams{$pname})) {
+			$out .= "$pname "; 
+			$flag = 1;
+		}
+	}
+
+	if ($flag == 1) {
+		print STDERR "missing parameters: $out\n";
+		exit(1);
+	}	
+}
+
+sub trim {
+	my $str = shift;
+	$str =~ s/^\s+//;
+	$str =~ s/\s+$//;
+	return $str;
+}

Makefile

@@ -0,0 +1,13 @@
+BIN ?= $(shell pwd)
+
+all:
+	sed -e 's:<pathtochainNet>:$(BIN)/examples/chainNet:g' examples/config.SFs.tmp > examples/config.SFs
+	cd lib/kent/src/lib && ${MAKE}
+	cd code/makeBlocks && ${MAKE}
+	cd code && ${MAKE}
+
+clean:
+	cd lib/kent/src/lib && ${MAKE} clean
+	cd code/makeBlocks && ${MAKE} clean
+	cd code && ${MAKE} clean
+	cd examples && rm -rf APCFs.300K config.SFs

README

@@ -0,0 +1,185 @@
+
+                       DESCHRAMBLER 
+          ======================================
+
+1. How to compile?
+------------------
+
+    1.1. Type make
+
+        Simple type make to compile the DESCHRAMBLER package.
+
+
+2. How to run?
+--------------
+
+    2.1. Mandatory input files
+
+        DESCHRAMBLER requires the following four files as input. Example files are in 
+        the "examples" directory.
+
+        - config.SFs: a configuration file for chain/net directories and species
+        - Makefile.SFs: a makefile for generating syntenic fragments
+        - tree.txt: a newick tree for used species and a target ancestor
+        - params.txt: a parameter file 
+
+        The above four files need to be placed in the same directory.
+
+    2.1.1. config.SFs
+
+        In the file, the following three values need to be specified. Please refer to the example
+        file in the "examples" directory.
+
+        - >netdir: a path to the directory that contains net files
+                   This directory needs to have a sub-directory structure as below.
+
+                   reference/descendant1/chain/separate chain files for each chromosome (or scaffold)
+                   reference/descendant1/net/separate net files for each chromosome (or scaffold)
+                   reference/descendant2/chain/separate chain files for each chromosome (or scaffold)
+                   reference/descendant2/net/separate net files for each chromosome (or scaffold)
+                   reference/descendant3/chain/separate chain files for each chromosome (or scaffold)
+                   reference/descendant3/net/separate net files for each chromosome (or scaffold)
+
+                   An example directory, called "chainNet", is in the "examples" directory.
+
+        - >chaindir: a path to the directory that contains chain files
+					 Usually, this directory is the same as the one used in >netdir above.
+
+       The chain/net files can be downloaded from the UCSC genome browser (http://genome.ucsc.edu/), 
+       or generated by using tools provided by the UCSC genome browser. Please refer to 
+       http://genomewiki.ucsc.edu/index.php/Whole_genome_alignment_howto.
+
+        - >species: the list of reference, ingroup and outgroup species 
+                    The species information is listed under ">species". Each line has the following 
+                    three values delimited by a space.   
+
+                    - Species name: this name must match with the directory name in the chain/net
+                      directory and the name in the newick tree in the "tree.txt" file.
+                    - Flag indicating whether the species is a reference (0), descendant (1), 
+                      or outgroup (2) species
+                    - Flag indicating whether the assembly of the species consists of chromosomes (1) 
+                      or not (0) 
+
+        Please do not change the ">resolution" section.
+
+    2.1.2. Makefile.SFs
+
+        You don't need to edit this file. Just copy this file to your working directory. Please use
+        the example file in the "examples" directory.
+
+    2.1.3. tree.txt
+
+        This file contains the newick tree for the species listed in the config.SFs file. 
+        The target ancestor must be specified by the "@" symbol. Please refer to the example tree
+        file in the "examples" directory.       
+
+    2.1.4. params.txt
+
+        The following parameters need to be specified.
+
+        - REFSPC: the name of a reference species
+                  This name must match with the name used in the chain/net directory, species list in 
+                  the config.SFs file, and the tree.txt file.
+        - OUTPUTDIR: the output directory 
+        - RESOLUTION: block resolution
+        - TREEFILE: a path to the newick tree file
+        - MINADJSCR: the minimum scores of adjacent syntenic fragments used in reconstruction
+        - CONFIGSFSFILE: a path to the config.SFs file          
+        - MAKESFSFFILE: a path to the Makefile.SFs file          
+
+
+    2.2. Run DESCHRAMBLER 
+
+        There is a wrapper Perl script, 'DESCHRAMBLER.pl'. To run DESCHRAMBLER, type as:
+
+            <path to DESCHRAMBLER>/DESCHRAMBLER.pl <path to the parameter file: params.txt> 
+
+
+3. What are produced?
+---------------------
+
+    Many files are generated in the output directory that is specified in the parameter file.
+    Among them the following files are the most important and useful.
+
+    3.1. block_list.txt (in the "SFs" subdirectory)
+
+        This file contains the coordinates and identifiers of syntenic fragments (SFs) used in 
+        reconstruction.
+
+            Column 1: chromosome or scaffold
+            Column 2: start position (0-based); add 1 to obtain the actual start position
+            Column 3: end position (1-based) 
+            Column 4: orientation 
+            Column 5: identifier 
+
+    3.2. Conserved.Segments (in the "SFs" directory)
+
+        This file contains the coordinates of genomic regions of each species belonging to each SF.
+
+            Lines starting with ">": ><identifier of an SF>
+            Other lines: the coordinates of genomic regions of each species belonging to the SF. The
+                format is as follows.
+                    <species>.<chromosome or scaffold>:<start position>-<end position> <orientation>
+
+                Here, the start position is 0-based, and the end position is 1-based. You need to
+                add 1 to the start position to obtain the actual start position.
+
+    3.3. Ancestor.APCF
+
+        This file contains the list of ancestral predicted chromosome fragments (APCFs), and the order 
+        and orientation of SFs in each APCF. 
+
+            Line 1: >ANCESTOR <total number of SFs>
+            Lines starting with "#": # APCF <identifier of APCF>
+            Other lines: the order and orientation of SFs belonging to the APCF shown immediately above line.
+                SFs are shown by using their identifiers (refer to the block_list.txt file).  
+
+    3.4. Ancestor.ADJS
+
+        This file contains the predicted adjacencies of pairs of SFs and their scores used in the 
+        Ancestor.APCF file. SFs are shown by using their identifiers (refer to the block_list.txt file).
+
+            Column 1: the first SF identifier
+            Column 2: the second SF identifier
+            Column 3: adjacency score of the first and the second identifiers
+
+        In the first and second columns, 0 means the end of APCFs. For example, "0 64" means the SF 64 
+        is placed at the end of a current APCF.      
+
+    3.5. APCF_size.txt
+
+        This file contains the total length of APCFs.
+
+            Column 1: APCF identifier
+            Column 2: total length
+
+        The last line shows the sum of total APCF lengths.
+
+    3.6. APCFs
+
+        This file contains the list of coordinates of genomic regions of each species matching to each APCF.
+
+            Lines starting with "#": #<identifier of APCF>
+            The coordinates of genomic regions of each species (grouped by species) follow this line. 
+
+    3.7. APCF_<SPC>.merged.map
+
+        This file contains the mapping genomic regions of the SPC species for APCFs. Each mapping is shown
+        by using the following three lines.
+
+            ><identifier of mapping>
+            APCF.<identifier of APCF>:<start position in APCF>-<end position in APCF> <orientation>
+            <SPC>.<chromosome or scaffold>:<start position in SPC>-<end position in SPC> <orientation>
+
+        The start position is 0-based, and the end position is 1-based. You need to add 1 to the start position to 
+        obtain the actual start position. 
+
+
+4. Supplementary data
+---------------------
+
+    The chain/net files and reconstruction results are available at our supplementary website
+    (http://bioinfo.konkuk.ac.kr/DESCHRAMBLER). 
+
+
+

code/Makefile

@@ -0,0 +1,30 @@
+CC = gcc
+GCC = g++
+CDEBUG = -g#-ggdb -g -pg
+OPTM =-O3
+WARN = -W
+KLIB = ../lib/kent/src/lib/
+KINC = ../lib/kent/src/inc
+CFLAGS = $(WARN) $(OPTM) -I. -I$(KINC)
+CLIB = $(KLIB)/jkweb.a -lm
+
+RM = rm -rf
+
+ALLSRC = inferAdjProb deschrambler 
+
+all: $(ALLSRC)
+
+%: %.c
+	$(CC) $(CDEBUG) $(CFLAGS) $+ $(CLIB) -o $@
+
+%: %.cpp 
+	$(GCC) $+ -o $@
+
+.PHONY: tags
+tags:
+	ctags *.[hc] lib/*.[hc]	inc/*.h
+
+.PHONY: clean
+clean:
+	$(RM) $(ALLSRC) *.o *.dSYM
+