No commit message

AUTHORS

@@ -0,0 +1 @@
+Sébastien "sebhtml" Boisvert  http://genome.ulaval.ca/users/boisvert/

ChangeLog

@@ -0,0 +1 @@
+http://denovoassembler.svn.sourceforge.net/viewvc/denovoassembler/

INSTALL

@@ -0,0 +1,6 @@
+You need a 64-bit machine.
+you need also a c++ compiler (standard: c++98) 
+and a UNIX box (mkdir, etc.).
+
+If you use GNU/Linux, simply type 'bash scripts/build.sh'. The executables
+will be in build/bin.

LICENSE

@@ -0,0 +1 @@
+COPYING

Makefile.am

@@ -0,0 +1 @@
+SUBDIRS = src

NEWS

@@ -0,0 +1,2 @@
+2008-12-14
+	NEWS file created.

README

@@ -0,0 +1,13 @@
+http://DeNovoAssembler.sf.net
+
+DNA assembly is now a challenge because of the
+ overwhelming amount of data produced by sequencing
+ facilities. De Novo Assembler is an assembler
+ to perform De Novo assembly with new sequencing 
+technologies (i.g. Titanium).
+
+
+Also to ensure that nothing goes wrong, you should limit the memory usage.
+
+
+ulimit -v  230000000 # in kB

configure.ac

@@ -0,0 +1,6 @@
+AC_INIT(dna,1.0,sebastien.boisvert.3@ulaval.ca)
+AC_CONFIG_SRCDIR([src/Read.cpp])
+AM_INIT_AUTOMAKE
+AC_PROG_CXX
+AC_OUTPUT(Makefile)
+AC_OUTPUT(src/Makefile)

scripts/LinuxDropCaches.sh

@@ -0,0 +1,3 @@
+sync
+echo 3 > /proc/sys/vm/drop_caches
+echo 0 > /proc/sys/vm/drop_caches

scripts/build.sh

@@ -0,0 +1,12 @@
+aclocal
+autoconf
+automake --add-missing
+autoreconf
+# -pedantic 
+# 
+
+export CXXFLAGS="-O6 -Wall -std=c++98 -fomit-frame-pointer -static  "
+export LDFLAGS=$CXXFLAGS
+mkdir -p build
+./configure CXXFLAGS="$CXXFLAGS" LDFLAGS="$LDFLAGS" --prefix=$(pwd)/build
+make install

scripts/doHuman.sh

@@ -0,0 +1,7 @@
+# 230000000
+# 230000000/1024/1024 ->  219 GB
+# 606 fastq files
+
+ulimit -v  230000000
+
+nohup dna -buckets 1000000000 -assemblyDirectory HumanGenome $(ls ~/Datasets/SRA000271/*fastq|head -n100) > /dev/null &

scripts/generateRandom.rb

@@ -0,0 +1,15 @@
+#!/usr/bin/ruby
+
+puts ">"
+10000.times do 
+	i=rand(4)
+	if i==0
+		print 'A'
+	elsif i==1
+		print 'T'
+	elsif i==2
+		print 'C'
+	elsif i==3
+		print 'G'
+	end
+end

scripts/keepLargeContigs.rb

@@ -0,0 +1,42 @@
+#!/usr/bin/ruby
+
+if ARGV.size!=3
+	puts "usage"
+	puts "keepLargeContigs.rb <contigsFile> <minimumContigSize> <largeContigsFile>"
+	exit
+end
+
+seq=""
+
+contigs=[]
+f=File.open ARGV[0]
+while l=f.gets
+    if l[0..0]=='>'
+        contigs<< seq
+        seq=""
+    else
+        seq<< l.strip
+    end
+end
+
+contigs<< seq
+f.close
+
+threshold=ARGV[1].to_i
+k=1
+out=File.open ARGV[2],"w+"
+contigs.each do |i|
+    if i.length<threshold
+        next
+    end
+    j=0
+    out.puts ">#{k} #{i.length}"
+    columns=60
+    while j<i.length
+        out.puts i[j..(j+columns-1)]
+        j+=columns
+    end
+    k+=1
+end
+
+out.close

scripts/pairedInfo.txt

@@ -0,0 +1,2 @@
+1 
+200x36x36-071113_EAS56_0053-s_1_1.fastq 200x36x36-071113_EAS56_0053-s_1_2.fastq 300

scripts/runTests.sh

@@ -0,0 +1,10 @@
+# 454 S. pneumoniae
+nohup dna -assemblyDirectory 1020 ~/Datasets/SRA001020/sff/ETJITFZ02.sff > /dev/null &
+# 454 S. cerevisiae
+nohup dna -assemblyDirectory 257 ~/Datasets/SRA000257/sff/*.sff > /dev/null &
+# Solexa S. cerevisiae
+nohup dna -assemblyDirectory 1177 ~/Datasets/SRA001177/SRR003681.*fastq* > /dev/null &
+# 454 L. tarentolae
+nohup dna -assemblyDirectory tar ~/Datasets/tar-454/sff/*.sff > /dev/null &
+# 454 r6 S pneumoniae
+nohup dna -assemblyDirectory r6  ~/Datasets/Marc/r6/sff/E*.sff > /dev/null &

scripts/simulate-454.rb

@@ -0,0 +1,97 @@
+#!/usr/bin/ruby
+
+if ARGV.size==0
+	puts "You must provide a file"
+	exit
+end
+
+def revComp a
+	b=""
+	i=a.length-1
+	while i>=0
+		s=a[i..i]
+		if s=='A'
+			b<< 'T'
+		elsif s=='T'
+			b<< 'A'
+		elsif s=='C'
+			b<< 'G'
+		elsif s=='G'
+			b<< 'C'
+		end
+		i-=1
+	end
+	b
+end
+
+chromosomes=[]
+f=File.open  ARGV[0]
+seq=""
+while l=f.gets
+	l=l.upcase
+	if l[0..0]=='>'
+		if seq!=""
+			chromosomes<< seq
+		end
+		seq=""
+	else
+		seq<< l.strip
+	end
+end
+
+chromosomes<< seq
+f.close
+
+coverage=25
+readLength=250
+errors=4
+readID=1
+chromosomes.each do |genome|
+	gSize=genome.length
+	position=0
+	while position<gSize
+		coverage.times do |t|
+			read_length=readLength+rand(100)-50
+			start=position+rand(read_length)-read_length/2
+			if start<0
+				start=0
+			end
+			sequence=genome[start..(start+read_length)]
+			if sequence.nil?
+				next
+			end
+			errorsInRead=errors+rand(4)-2
+			errorsInRead.times do
+				break
+				n=rand(4)
+				p=rand(sequence.length)
+				if n==0
+					sequence[p..p]='A'
+				elsif n==1
+					sequence[p..p]='T'
+				elsif n==2
+			sequence[p..p]='C'
+		elsif n==3
+			sequence[p..p]='G'
+				end
+			end
+			if rand(2)==0
+		puts "@#{readID}_#{start}_#{read_length}_F_#{errorsInRead}"
+		puts sequence
+		puts "+#{readID}_#{start}_#{read_length}_F_#{errorsInRead}"
+			else
+		puts "@#{readID}_#{start}_#{read_length}_R_#{errorsInRead}"
+		puts revComp(sequence)
+		puts "+#{readID}_#{start}_#{read_length}_R_#{errorsInRead}"
+			end
+			readID+=1
+	j=0
+			while j<sequence.length
+		print 'F'
+		j+=1
+			end
+	puts ""
+		end
+		position+=readLength
+	end
+end

scripts/simulate-perfect.rb

@@ -0,0 +1,89 @@
+#!/usr/bin/ruby
+
+if ARGV.size==0
+	puts "You must provide a file"
+	exit
+end
+
+def revComp a
+	b=""
+	i=a.length-1
+	while i>=0
+		s=a[i..i]
+		if s=='A'
+			b<< 'T'
+		elsif s=='T'
+			b<< 'A'
+		elsif s=='C'
+			b<< 'G'
+		elsif s=='G'
+			b<< 'C'
+		end
+		i-=1
+	end
+	b
+end
+
+chromosomes=[]
+f=File.open  ARGV[0]
+seq=""
+while l=f.gets
+	l=l.upcase
+	if l[0..0]=='>'
+		if seq!=""
+			chromosomes<< seq
+		end
+		seq=""
+	else
+		seq<< l.strip
+	end
+end
+
+chromosomes<< seq
+f.close
+
+coverage=10
+readLength=250
+readID=1
+errors=0
+chromosomes.each do |genome|
+	gSize=genome.length
+	position=0
+	while position<gSize
+		coverage.times do |t|
+			read_length=readLength
+			start=position
+			if start<0
+				start=0
+			end
+			sequence=genome[start..(start+read_length)]
+			if sequence.nil?
+				next
+			end
+			errorsInRead=errors+rand(4)-2
+			errorsInRead.times do
+				break
+				n=rand(4)
+				p=rand(sequence.length)
+				if n==0
+					sequence[p..p]='A'
+				elsif n==1
+					sequence[p..p]='T'
+				elsif n==2
+			sequence[p..p]='C'
+		elsif n==3
+			sequence[p..p]='G'
+				end
+			end
+			if rand(2)==0
+		puts ">#{readID}_#{start}_#{read_length}_F_#{errorsInRead}"
+		puts sequence
+			else
+		puts ">#{readID}_#{start}_#{read_length}_R_#{errorsInRead}"
+		puts revComp(sequence)
+			end
+			readID+=1
+		end
+		position+=readLength
+	end
+end