Saturn is a tool for assessing the library saturation without alignment. It allows a sequencing platform to assess the saturation, to predict the yield of distinct reads from deeper sequencing and to adjust the sequencing effort even without any reference genome. Saturn uses an estimation of reads duplication and an equation model estimating potential errors along with standard saturation. As sequencing errors introduce new distinct reads and leads to overestimate the complexity of a readset, we modelize them to predict the correct amount of new sequences obtained by additionnal sequencing from a library.
Saturn is distributed open-source under CeCILL FREE SOFTWARE LICENSE. Check out http://www.cecill.info/ for more information about the contents of this license.
Saturn home on the web is http://www.genoscope.cns.fr/externe/saturn/
Usage : saturn --fq1 <fastq file 1> --fq2 <fastq file 2> --o <output directory> {Options}
--fq1 : illumina reads (R1) in fastQ format
--fq2 : illumina reads (R2) in fastQ format (Optional)
--nparts : number of parts to split file
--o : Output directory, default is Saturn_date_pid
--nb_proc : Number of parallel task, default is 16
--h : help message
Several directory are created in the output directory :
- files : contain sample files of increasing size generated from the input fastq files
- data : contain the output files of fastx_estimate_duplicatedReads on each sample and a file 'duplicates_data.csv' which sumarize level of duplication. This last file is used in input for neoreg which create a estimation of the saturation
- output : contain results images and stats of saturn (via neoreg).
File 'summary.txt' in output directory contain a description of the regression model, an estimation of the number of fragments in the library and a table depicting the complexity of sequenced fragments (X<1) and benefit of additionnal sequencing (X>1) until having 95% of the unique fragments of the library :
- X : factor of the number of reads in the input fastq file. For example the line X=2 give the estimation of the complexity if the library is sequenced one more time.
- nbReads : number of reads or pair reads sequenced
- nUniq : estimation of the number of unique read or pair reads if nbReads were sequenced
- nUniqError : estimation of the number of unique fragment if nbReads were sequenced. This correspond to nUniq minus the estimated number of read or pair reads with errors.
- %dupl : duplication rate if nbReads were sequenced
- pbUniq : probability of sequencing a new fragment after X sequencing
Two jpg files depicting the curves nUniqError (dotted curve) and nUniq (plain curve) according to the sequenced number of reads (nbReads) and the estimated number of fragment in the library (dotted line). File saturation.jpg until 3 times the number of reads in the input file, file saturation_80.jpg until nUniqError reach 80% of the estimated number of fragment in the library.
- A Linux based operating system.
- Binaries are provided for the following platform : Linux x86_64
- Perl 5.8.0 or higher installed.
- Perl Compress' Zlib module (http://search.cpan.org/~pmqs/IO-Compress-2.068/lib/Compress/Zlib.pm)
- Perl GetOpt module (http://search.cpan.org/dist/Getopt-Long/)
- R version 2.4.1 or higher installed.
- g++ with gcc 4.1.2 or higher
- parallel installed (working with ver. 20130922-1)
- Download the current tarball archive from http://www.genoscope.cns.fr/externe/saturn/
wget http://www.genoscope.cns.fr/externe/saturn/saturn_latest.tar.gz - Untar/unzip the archive
tar -zxvf saturn_latest.tar.gz - compile sources
cd saturn_latest; make; make install - Modify if needed the Perl , R and sh interpreters that have been set to /usr/bin/perl , /usr/bin/env Rscript, and /bin/bash
If you have questions about Saturn, you may ask them to sengelen [at] genoscope [.] cns [.] fr and jmaury [at] genoscope [.] cns [.] fr . You may also create an issue to ask questions on github website: https://github.com/institut-de-genomique/saturn/issues.
Stefan Engelen, Cyril Firmo, Amin Madoui and Jean-Marc Aury - Saturn's authors
This work was financially supported by the Genoscope, Institut de Genomique, CEA and Agence Nationale de la Recherche (ANR), and France Génomique (ANR-10-INBS-09-08).