Strelka Workflow -- 1.0.11
---------------

Christopher Saunders (csaunders@illumina.com)


Strelka is an analysis package designed to detect somatic SNVs and
small indels from the aligned sequencing reads of matched tumor-normal
samples.

For additional documentation and mailing list info, please see:

https://sites.google.com/site/strelkasomaticvariantcaller


WORKFLOW SUMMARY:

The inputs of the workflow are BAM files for each sample containing
aligned sequencing reads. These must be sorted and indexed. PCR
duplicate removal is recommended. Strelka does its own read
realignment around indels so it is not necessary to pre-process the
data with any such step.

The output of the workflow are a pair of VCF files describing the
somatic SNVs and indels. The workflow takes advantage of the VCF file
format to report both filtered and passed variants, but it is strongly
recommended that only the passed variants (or a subset thereof) be
used.



INSTALLATION:

The strelka workflow is designed to run on *nux-like platforms. It has
been specifically tested on the following distributions:

* Amazon Linux AMI 2011.09
* Centos 5.3
* Centos 6.2
* Mac OS X 10.5
* Ubuntu 11.10
* Ubuntu 12.04 LTS


For Centos/Amazon Linux and other rpm-based distributions, the following
packages (and their dependencies) are known requirements on top of
the base Amazon Linux AMI 2011.09:

* make
* gcc
* gcc-c++
* zlib
* zlib-devel


For Ubuntu and other Debian package distributions, the following
packages (and their dependencies) are known requirements on top of
the base Ubuntu 11.10 distribution:

* g++
* zlib1g-dev


Other compilation and runtime prerequisites are included in the distribution
tarball. These included prerequisites are:

* boost 1.44 (modified to include only the subset of the library used by strelka)
* samtools 0.1.18 (modified to remove the 'curses' library requirement)
* vcftools r837 
* codemin 1.0.2
* tabix 0.2.5 (makefile modified to reorder libz library to last place, as required for Ubuntu 12 build)


The build process for strelka workflow will compile the included prerequisite packages,
ancillary workflow programs and the strelka variant caller itself. To build the workflow,
go to the root of the strelka_workflow directory tree and use configure and make as
in the example below: 

Example build:
"""
tar -xzf strelka_workflow-1.2.3.tar.gz
cd strelka_workflow-1.2.3
./configure --prefix=/path/to/install/to
make
"""

If the installation succeeds, all files for the above example will be
installed to "/path/to/install/to".

The final installation step is to run the workflow on the
demonstration data included in the distribution. This verifies that the
workflow completes without error and produces the expected answer.
To do this, first complete the installation steps above. 

Next, given a strelka installation path of $STRELKA_INSTALL_DIR, execute
the following script:

$STRELKA_INSTALL_DIR/bin/demo/run_demo.bash

The demo will create a 'strelkaDemoAnalysis' directory under the current
working directory, then it will run an analysis for a very small chromosome
segment.

The demo should complete in less than a minute.  If the demo script succeeds,
you should see the following lines as the final output:

"""
**** No differences between expected and computed results.


**** Demo/verification successfully completed
"""



CONFIGURATION & ANALYSIS:

A strelka run is accomplished in two steps: (1) configuration and (2)
analysis.

The configuration is a fast step designed to setup the run and perform
some preliminary validation (eg. check that the chromosome names match
in the BAM header and reference genome). The configuration will create
a makefile to control the analysis and a directory structure to hold
all intermediate and final results. All configuration output is
written to the analysis directory. This step requires a configuration
initialization file containing strelka's default parameters. Template
configuration files can be found in the etc/ subdirectory under
the strelka_workflow installation root (see example run setup below).

Following configuration, the makefile can be used to run the analysis
itself using make, qmake, or a similar tool. For whole genome sequence
data from human samples, strelka requires approximately 1 core-hour
per 2x of combined sample reference coverage e.g. analyzing a human
tumor-normal sample pair sequenced to 60x and 40x respectively should
take approximately 50 core-hours.

Example run:
"""
# example location where strelka was installed:
#
STRELKA_INSTALL_DIR=/opt/strelka_workflow

# example location where analysis will be run:
#
WORK_DIR=/data/myWork

# Step 1. Move to working directory:
#
cd $WORK_DIR

# Step 2. Copy configuration ini file from default template set to a local copy,
#         possibly edit settings in local copy of file:
#
cp $STRELKA_INSTALL_DIR/etc/strelka_config_eland_default.ini config.ini

# Step 3. Configure:
#
$STRELKA_INSTALL_DIR/bin/configureStrelkaWorkflow.pl --normal=/data/normal.bam --tumor=/data/tumor.bam --ref=/data/reference/hg19.fa --config=config.ini --output-dir=./myAnalysis

# Step 4. Run Analysis
#         This example is run using 8 cores on the local host:
#
cd ./myAnalysis
make -j 8