klust

Efficient DNA/RNA sequence clustering using k-mers as an approximation for sequence similarity.

Compiling and running klust

The klust program has been developed and tested on GNU/Linux systems and compiled with the g++ compiler (version 5.1.0), which is part of GCC. The program has also been tested to work with the clang++ compiler (version 3.6.1). The program is expected to compile and run on any POSIX compliant system with a recent g++ or clang++ compiler supporting C++11.

The following commands will download and compile the klust program, assuming git is installed:

git clone https://github.com/Rathcke/klust.git
cd klust/src
make

Simply running klust with no arguments will print instructions for usage:

$ ./klust
Usage: ./klust <FASTA input file>

Options:
-o, --centroids file   Output FASTA file for centroids
-u, --clusters file    Output file for clustering results
-t, --id t             Set similarity threshold/identity to t (in [0,1])
-k k                   Set the k in k-mer, used in similarity metric
-c, --count n          Read and process n sequences
-d, --sort_decr        Sort sequences by decresing length
-i, --sort_incr        Sort sequences by increasing length
-l, --depth d          Set the depth of the tree of divides, i.e.
                       cluster 2^d subsets of sequences and combine
-m, --max_rejects x    Max number of rejects when searching for centroid

--springy file         Generate springy JavaScript code

Example

The following example will cluster the sequences in the FASTA format file ../data/SILVA_10k.fasta which contains 10,000 RNA sequences (the first 10k from the SILVA_119_SSURef_tax_silva.fasta dataset). The program is run with default clustering parameters and the input sequences are sorted by increasing sequence length. The centroids will be written to the file centroids.fasta and the clustering results to the file clusters:

./klust ../data/SILVA_10k.fasta -o centroids.fasta -u clusters --sort_incr