Cookiecutter is a computational tool for filtering reads by their matches to specified k-mers. Originally it was created to filter reads with primers and adapters from large sets of sequencing data.
The preprint of the paper describing Cookiecutter is available on bioRxiv.
To compile and use Cookiecutter, the following tools must be installed.
- make;
- gcc 4.7 or higher;
- python 2.7.
Cookiecutter is designed for use on Linux/UNIX and OS X systems.
You may install Cookiecutter from a binary package or compile it from source codes.
Unpack the downloaded archive using tar:
tar -xvzf cookiecutter_osx.tar.gz
Executable files are located in the bin subdirectory. You may
either launch Cookiecutter from the directory to which the archive was
unpacked or copy executable files to any directory specified in the
PATH variable of your environment.
The package should be compiled from its source code using the provided Makefile in the following way.
git clone http://github.com/ad3002/Cookiecutter.git
cd Cookiecutter/src
make
sudo make install
If you do not have root access, you can use Cookiecutter from the src
directory or specify another installation directory using PREFIX:
PREFIX=/my/dir make install
To uninstall Cookiecutter, use make uninstall. If an installation
directory was specified using PREFIX, then it should also be
specified
for uninstalling: PREFIX=/my/dir make uninstall.
Cookiecutter contains a number of subroutines for various tasks:
-
remove searches given k-mers in reads and outputs the reads without any matches to the k-mers;
-
rm_reads is an extension of remove that additionally provides options to filter reads by the presence of (C)n/(G)n tracks or unknown nucleotides, read length or low sequence complexity and outputs both filtered and unfiltered reads;
-
extract searches given k-mers in reads and outputs the reads that matched the k-mers;
-
separate searches given k-mers in reads and outputs both matched and unmatched reads to two separate files.
These subroutines may be launched directly or using the wrapper script cookiecutter. We recommend to use the wrapper because it allows to process multiple input files in parallel mode and provides a convenient command-line interface to the subroutines. Also one may create k-mer libraries from FASTA files using the cookiecutter make_library tool.
Below we give examples of Cookiecutter usage. To get more information
about the program options, use the -h argument: cookiecutter -h.
It can also be applied to a specific subroutine, for example
cookiecutter rm_reads -h.
A library of k-mers is necessary for all Cookiecutter subroutines. It
can be created from a FASTA file using cookiecutter make_library.
For example, the command
cookiecutter make_library -i adapters.fa -o adapters.txt -l 5
will create the file adapters.txt of k-mers of length 5 bp from the FASTA file adapters.fa.
If you are going to create a library from a large dataset or you have limited memory on your machine you can use Jellyfish 2 for the fast k-mer computation with following command:
jellyfish count -m 23 -s 2G -t 4 --text -o kmer_library.dat yourdata.fastq
Let us have a library of k-mers adapters.txt created as described above and a FASTQ file of single-end reads raw_data.fastq, and we would like to remove all reads containing any k-mers from the library. It can be done using remove in the following way.
cookiecutter remove -i raw_data.fastq -f adapters.txt -o filtered
The output FASTQ file raw_data.ok.fastq will be created in the
directory specified by the -o argument. It will contain the reads
that do not include any of the specified matches.
Let us have the same data set as in the subsection above, but now we are to extract the reads matching any of the specified k-mers. For that, one should use extract in the same way as remove:
cookiecutter remove -i raw_data.fastq -f adapters.txt -o filtered
Let us have two FASTQ files of paired-end reads raw_data_1.fastq and raw_data_2.fastq. In addition to the k-mer presence filter, we would also like to filter them by the following criteria: read length, presence of (G)n or (C)n tracks, sequence complexity (DUST) and unknown nucleotides within a read. The rm_reads tool was designed for such filtration.
cookiecutter rm_reads -1 raw_data_1.fastq -2 raw_data_2.fastq
-f adapters.txt -o output_dir --polygc 13 --length 50
--dust --filterN
Since we specified a pair of FASTQ files, the output files will also be paired. Read pairs are maintained if both paired-end read parts passed the filtration. If one part of a read passed the filtration but another failed it, then the passed part will be output to the file which name ends with .se.fastq.
Let us have the same paired-end FASTQ files raw_data_1.fastq and raw_data_2.fastq as in the subsection above. We would like to separate reads matching the k-mer library from reads that do not match it. We will use the separate tool.
cookiecutter separate -1 raw_data_1.fastq -2 raw_data_2.fastq
-f adapters.txt -o output_dir
Cookiecutter supports processing multiple input files (or pairs
of input FASTQ files for paired-end reads) in parallel mode. For
that, one should specify multiple input files in the arguments -1,
-2 or -i (see examples below).
cookiecutter remove -1 reads_a_1.fastq reads_b_1.fastq
-2 reads_a_2.fastq reads_b_2.fastq -f adapters.txt
-o output_dir
cookiecutter extract -i reads_a.fastq reads_b.fastq
-f adapters.txt -o output_dir
Also one may specify multiple input FASTA files for the k-mer library making tool.
cookiecutter make_library -i input_1.fa input_2.fa -o library.txt -l 5
The latest version of Cookiecutter is publicly available at its GitHub repository. If you find any bugs or have any suggestions how to improve the tool, please find free to post issues at the repository. The earliest version of Cookiecutter can also be found at GitHub.