ray-assembly.txt

Metagenomics Practical
======================

Metagenomics assemblies with Ray
--------------------------------

Ray is a particularly interesting genome assembler due to several
unusual features:

- It can scale to arbitrary numbers of processors and machines by distributing its assembly graph
- It has several functions specific to metagenome assembly 'Ray Meta'
- Ray's author, ``@sebhtml`` is incredibly responsive on Twitter :)
- Ray will happily mix input from several different sequencing techniques, e.g. Illumina and 454
- If run with the ``write-kmers`` option enabled, the resulting assembly graph may be viewed using the separate Ray Cloud Browser software

Installing Ray
--------------

Dependencies
~~~~~~~~~~~~

::

    sudo apt-get install build-essential
    sudo apt-get install git
    sudo apt-get install openmpi1.6-bin openmpi1.6-common libopenmpi1.6-dev

Installing Ray from source code
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

::

    git clone https://github.com/sebhtml/ray
    git clone https://github.com/sebhtml/RayPlatform

    cd ray
    HAVE_LIBZ=y MAXKMERLENGTH=64 make

You can add this to your PATH:

::

    export PATH=$PATH:`pwd`

A simple command-line for multi-processor execution:

For paired-end reads:

::

    mpirun -np 8 Ray -k 31 -p pair1.fastq.gz pair2.fastq.gz -o output_directory

For interleaved paired-end reads:

::

    mpirun -np 8 Ray -k 31 -i pairs.fastq.gz -o output_directory

For single-end reads:

::

    mpirun -np 8 Ray -k 31 -s reads.fastq.gz -o output_directory

If you want to run Ray Cloud Browser, you will want the ``-write-kmers``
option:

::

    mpirun -np 8 Ray -write-kmers -k 31 -p pair1.fastq.gz pair2.fastq.gz -o output directory

If you run via a cluster, i.e. StarCluster, mpirun can be set to execute
on multiple machines, e.g.:

::

    mpirun -np 8 -H host1,host2,host3,host4 -k 31 -p pair1.fastq.gz pair2.fastq.gz -o x

For more command-line options, see:

https://github.com/sebhtml/ray/blob/master/MANUAL\_PAGE.txt

Ray Cloud Browser
~~~~~~~~~~~~~~~~~

Here is a useful script to set up Ray Cloud Browser from a kmers.txt and Contigs.fasta file:

::

    #!/bin/bash
    tag=$1
    kmerfile=$2
    contigfile=$3
    mapid=$4
    sectionid=$5

    RayCloudBrowser-client create-map $kmerfile $tag.dat
    RayCloudBrowser-client add-map config.json "$tag" $tag.dat
    RayCloudBrowser-client create-section $contigfile $tag-contigs.dat
    RayCloudBrowser-client create-map-annotations-with-section $tag.dat $tag-contigs.dat $sectionid
    RayCloudBrowser-client add-section config.json $mapid "$tag Contigs" $tag-contigs.dat