Introduction

This repository aims to provide the tools to bioinformatics results reported in the manuscript Conbase: a software for unsupervised discovery of clonal somatic mutations in single cells through read phasing by Hård et al. (Genome Biology). These tools is set to work in a Unix-like operative system (Linux, Unix, MacOSX, cygwin, etc.) from a Unix-terminal.

The main aim is documentation of the data analysis pipelines used. The data used in the manuscript is available on https://ega-archive.org/studies/EGAS00001003108.

Results

Snakemake procedures for data processing and simulations

The conda environments

We have used conda environments to set up the software framework needed to perform the analyses (https://docs.conda.io). The required conda-packages are listed in pipeline-spcific (see below) requirements.yaml/txt files

NB! GATK cannot be installed completely from bioconda since it requires a license. Therefore you have to download GATK (v3.6 or which ever is specified in requirements.txt) from the website. After activating the conda environment you need to run gatk-register and refer to the downloaded GATK.

The snakemake files

There are three separate pipelines:

The main variant-calling pipeline

This is defined in Snakefile in the main/base directory perform read-mapping, QC and initial variant-calling of single-cell data. Conda requirements can be found in requirements.yaml.

This pipeline was used for the analyses of both experimental single-cell datasets (human T-cell and fibroblasts, respectively) in the manuscript. However, different configuration files were used for the analyses: config_Tcell.yaml, config_Fibs.yaml, respectively.

The LiRA pipeline

This is defined in pipeline_lira/Snakefile and describes how analyses using the program LiRA were run. The folder pipeline_lira/Snakefile also functions as the relevant documentation. See further the README.md file in the folder pipeline_lira.

The simulation pipeline

This is defined in pipeline_simulation/Snakefile and performs the simulation study from the manuscript, that is, generates synthetic single-cell data based on observed bulk data, runs Conbase, Monovar, LiRA, and SCcaller variant calling on this data and summarizes the results. See further the README.md file in the folder pipeline_simulation.

The simulation model description and the scripts used to generate synthetic data can be found in the folder simulation