Installation

Install

SweepLink installs with conda, which brings the executable and all four companion scripts in one go:

$ conda install -c CHANNEL sweeplink

Note

The conda package is not published yet. Until it is, use Building from source below — the channel name will be filled in here on release.

That gives you five commands:

Command	What it is
sweeplink	The inference program itself — see sweepLink tasks.
sweeplink-vcf2input	Script that converts VCF → allele counts and meta file.
sweeplink-extract	Script to extract the loci under selection from the results.
sweeplink-plot	Script to plot the results of the inference.
sweeplink-plot-posteriors	Script to plot the posterior of the population size.

The four sweeplink-* scripts are documented under Companion scripts.

Check it works

Run the executable with no arguments to see the version and the available tasks:

$ ./sweeplink

                 sweepLink 1.0.0
                -----------------

            University of Fribourg
  https://bitbucket.org/wegmannlab/sweepLink

Commit 574598274fbe838e3655a47e6d38b417a038eab7

  - Used executable: ./sweeplink
  - Available tasks:
        - calculateLL        Calculate the log-likelihood for a set of parameters.
        - calculateLLForN    Calculate the log-likelihood for each site for a given grid on N.
        - calculateLLForS    Calculate the log-likelihood for each site for a given grid on s.
        - calculateLLForSH   Calculate the log-likelihood for each site for given grids on s and h.
        - help               Print a categorized reference of all command-line options.
        - infer              Inferring selection from time-series data.
        - simulate           Simulating under the SweepLink model.

  - Usage: ./sweeplink taskName [options]
  - sweepLink terminated successfully in 0 seconds!

The version and commit will be whatever you installed. Then simulate a small dataset and infer on it, which exercises the whole path:

$  # Simulate an allele counts file with 100 loci sampled from one population
$ ./sweeplink simulate --numLoci 100 --numPop 1 --binomialN 1000 --mu_a_A 1e-8
$  # Run inference of selection across those loci
$ ./sweeplink infer --meta sweeplink_meta.txt --counts sweeplink_alleleCounts.txt --mu_a_A 1e-8 --burnin 10 --iterations 100

Both should end with sweepLink terminated successfully, and the second writes the output files described in File formats.

Note

--mu_a_A is mandatory for every task — there is no sensible default mutation rate, so a command without it stops immediately.

Building from source

Needed if you want to develop sweepLink, or are on a platform the conda package does not cover.

Requirements, download and compile

SweepLink requires:

• A compiler compatible with C++17 or higher. We recommend gcc version 9 or higher — check yours with gcc --version.

• cmake version 3.14 or higher — check with cmake --version.

Linux

You should be able to build sweepLink without any problems.

MacOS

Use Homebrew to install cmake, autoconf and automake:

$ brew install cmake autoconf automake

Windows

Build sweepLink under Windows Subsystem for Linux.

Cluster

On most computing clusters you can load recent cmake and gcc versions using modules.

Download the repository:

$ git clone --depth 1 https://bitbucket.org/wegmannlab/sweeplink.git

Compile it with cmake:

$ cd sweeplink
$ mkdir -p build
$ cd build
$ cmake .
$ cmake --build .

This leaves a compiled sweeplink program in the repository directory.

The companion scripts are a separate Python package in the tools directory, and are not built by cmake. Install them with pip:

$ pip install ./tools

SweepLink itself only reads allele counts, not VCF files, and only writes raw posterior files, not figures — the scripts fill both gaps and are used throughout the tutorial.

Next

The tutorial walks through a complete analysis on a small bundled dataset.