Skip to content

sfu-compbio/deez

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DeeZ: Reference Based Compression by Local Assembly

Warning: Current version is 1.9 beta 2. It will eventually become 2.0. It is huge improvement over 1.x series, but it is still in developement, and it might contain some rough edges. If you need absolute stability, please use 1.1 until 2.0 is released.

So what is DeeZ?

DeeZ is a tool for compressing SAM/BAM files.

How do I get DeeZ?

Just clone our repository and issue make command:

git clone https://github.com/sfu-compbio/deez.git
cd deez && make -j 

Note: You will need at least g++ 4.8 with C++11 support to compile the sources, as well as BZ2, CURL, and OpenSSL libraries.

How do I run DeeZ?

DeeZ is invoked as following:

  • Compression

    deez -r [reference] [input.sam] -o [output]

    This will compress input.sam to input.sam.dz.

  • Decompression

    deez -r [reference] [input.dz] -o [output] ([region])

    This will decompress input.dz to input.dz.sam. [region] is optional.

  • Random Access

    You can also specify the region of interest while decompressing (i.e. randomly access the region). For example, to extract some reads from chr16 to standard output, you should run: deez -r [reference] input.dz -c chr16:15000000-16000000

DeeZ supports URLs as input files. Supported protocols are file://, ftp://, http(s):// and s3://. For Amazon S3 access, DeeZ accepts AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables if you need to access private repositories.


Parameter explanation

  • --threads, -t [number]

    Set up the number of threads DeeZ may use for compression and decompression.

    Default value: 4

  • --header, -h

    Outputs the SAM header.

  • --reference, -r [file|directory]

    Specify the FASTA reference file. If no reference file is provided, DeeZ will either try to download reference specified in SAM @SQ:UR field. If such field is not available or if reference specified therein is not accessible, DeeZ will use no reference.

    If no reference is being used, compression rate will not be optimal.

    Note: Chromosome names in the SAM and FASTA files must match. Also, instead of one big FASTA file, DeeZ supports reference lookup in the given directory for chr*.fa files, where chr* is the chromosome ID from the SAM file.

  • --force, -!

    Force overwrite of exiting files.

  • --verbosity, -v [level]

    Set the logging level. Level 0 shows only errors and warnings. Level 1 shows many other useful information (e.g. download progress, compression progress, per-field data usage etc.) Level 2 shows debug information, including detailed CURL output.

    Default valie: 0

  • --stdout, -c

    Compress/decompress to the stdout.

  • --output, -o [file]

    Compress/decompress to the file.

  • --lossy, -l

    Set lossy parameter for quality lossy encoding (for more information, please check SCALCE).

  • --quality, -q [mode]

    If mode is 1 or samcomp, DeeZ will use sam_comp quality model to encode the qualities. Quality random access is not supported on those files. If mode is 2, DeeZ will use rANS order-2 model. This will result in slightly better compression, but decompression will be slower.

  • --withflag, -f [flag]

    Decompress only mappings which have flag bits set.

  • --withoutflag, -F [flag]

    Decompress only mappings which do not have flag bits set.

  • --stats, -S

    Display mapping statistics (needs DeeZ file as input).

  • --sort, -s

    Sort the input SAM/BAM file by mapping location.

  • --sortmem, -M [size]

    Maximum memory used for sorting.

    Default value: 1G

  • --noqual, -Q

    Disable quality score output during decompression.


Contact & Support

Feel free to drop any inquiry to inumanag at sfu dot ca.

Publication

DeeZ was publised in Nature Methods in November 2014.

Authors

  • Dr. Faraz Hach
  • Ibrahim Numanagić
  • Dr. S. Cenk Şahinalp

Many thanks to...