search

sourmash-bio / sourmash_plugin_randsampling

kmers extraction based on weighted-random-sampling
BSD 3-Clause "New" or "Revised" License
1 stars 1 forks source link

readme

sourmash_plugin_RandSampling

The RandSampling plugin performs weighted-random-sampling to extract a max number of k-mers from a signature using their abundance distribution.

The main aim is to extract k-mers from large signatures with different coverages (abundances) to create a smaller signature with a more uniform coverage distribution.

alt text The figure illustrates the relationship between abundance (x-axis) and frequency (y-axis) in the context of k-mer extraction. In the plot, the median abundance frequency is indicated by a horizontal line, and the selected abundances are represented by individual dots. Specifically, this visualization reflects the process of extracting 100,000 k-mers from a pool of approximately 500,000 k-mers, highlighting the weighted-random-sampling technique used to create a more uniform distribution.

Installation

Make sure you have an updated pip version:

# update pip
pip install --upgrade pip

# or 
conda update pip

Then install the plugin:

pip install git+https://github.com/sourmash-bio/sourmash_plugin_randsampling.git

Usage

sourmash scripts randsampling --help
usage:  randsampling [-h] [-q] [-d] --sig SIG -m MAX_KMERS -k K -o OUT [--plot] [--force]

options:
  -h, --help            show this help message and exit
  -q, --quiet           suppress non-error output
  -d, --debug           provide debugging output
  --sig SIG             path to signature file
  -m MAX_KMERS, --max-kmers MAX_KMERS
                        maximum number of kmers to extract
  -k K                  kmer size
  -o OUT, --out OUT     path to output signature file
  --plot                plot abundance distribution
  --force               force write new signature if empty or <max_kmers

The --force option allows you to force the extraction of a signature in case it's empty or has fewer than max_kmers. This is useful in workflows (e.g. snakemake).

The --plot option allows you to plot the abundance distribution of the input and output signatures.

Example

sourmash scripts randsampling  -k 51 --sig large.sig --max-kmers 1000 --plot --force -o subsampled.sig

Motivation

In sourmash, the concurrent use of the scale and num parameters is not supported, and utilizing the num parameter is generally discouraged. This restriction has posed challenges in effectively obtaining a well-distributed, user-defined number of k-mers that would represent the original signature to a satisfactory degree.