Branch: master
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
data
scripts
README.md
Snakefile

README.md

Replicating the experiments

This directory contains the instructions for reproducing the experiments described in our study: "HapCHAT: Adaptive haplotype assembly for efficiently leveraging high coverage in long reads", for which a preprint is available at https://doi.org/10.1101/170225.

The workflow is described as a "Snakefile", which needs to be run with snakemake.

Overview

The workflow is designed to be run from the directory in which this README file is located. In principle, you only need to run snakemake here and then the rest is done automatically, but all dependencies need to be installed first. Further things to know:

  • Approximately 150 GB of hard disk space is needed (better 200 GB to be on the safe side).

  • No data is included in this repository since all required files will be generated (which take up said 150 GB).

  • 8 cores should work, but more is better. Presumably 8 GB RAM works, but more is better. We tested on a 32-core machine with 256 GB RAM.

Obtaining/generaing the input data

More precisely, this workflow replicates what is described in the subsection "Expermental setup" of the above "HapCHAT" study. The data, described in the subsection "Data description", which is used by the former, must be generated first.

In order to do this, go to the child directory /data of this current directory and follow instructions there for generating the input VCF and BAM files required byr this workflow.

Installing dependencies

Install the necessary system packages that the different phasing softwares of this workflow depend on. Some of these dependencies have already been installed as a prerequisite to runng the workflow for obtaining/generating the input data -- so we assume that this workflow has finished successfully.

We also assume you use Debian or Ubuntu. If you do not have root access on your system, you can try to ignore this step. These packages are typically already installed.

sudo apt-get install default-jdk
conda install -y virtualenv networkx

Python 2, and the weave module (supported by python<=2.6) are needed to run ProbHap.

sudo apt-get install python python-pip
sudo pip install weave

Run the workflow

Start the workflow with

nice snakemake -p -j 16

Adjust the 16, which is the number of cores you want to use.

Results

The resulting files for each tool/dataset will be reported in the output directory (grouped by tool), as

  • a .phased.vcf file: the phasing produced by the tool

  • a whatshap diff file (.diff) on this phased VCF file containing the accuracy measures

  • a .time file for the output of /usr/bin/time for time and peak memory used to produce this phased VCF