FunPipe: a python library for building best practice fungal genomic analysis pipeline

FunPipe is a python library designed for efficient implementation of bioinformatic tools and pipelines for fungal genomic analysis. It contains wrapper functions to popular tools, customized functions for specific analyses tasks, and command line tools developed using those functions. This package is developing to facilitate fungal genomics, but many of the functions are generally applicable to other genomic analysis as well.

Synposis

• funpipe: a directory that contains python library
• scripts: tools and established pipelines, doc here
• tests: unit tests
• docs: API documentation
• README.md: this file
• setup.py: pip setup script
• conda_env.yml: spec file for setting up conda environment
• Dockerfile: docker images
• requirements.txt: sphinx requirement file (not requirement for this package)
• LICENSE: MIT license

Installation

Install with Conda

It is recommended to install funpipe via conda, as it automatically setup all required bioinformatic tools. This is very useful on servers or clusters without root privilage. Make sure conda is available in your environment via which conda. If conda is not available in your system, install Python3.7 version of it here.

HTTP errors sometimes occur when creating the conda environment, simply rerun the conda env create -f conda_env.yml to continue creating the environment.

# clone this repo
git clone git@github.com:broadinstitute/funpipe.git

# setup conda environment
cd funpipe

conda env create -f conda_env.yml # this will take about 10 min
conda list  # verify new environment was installed correctly

# activate funpipe environment
conda activate funpipe

# the latest stable version of funpipe is available in this environment
# to use the latest funpipe version, do
pip install .

# deactivate the environment when done
conda deactivate


# to complete remove the environment
conda remove -n funpipe --all

Note:

• diamond=0.9.22 uses boost library, which depends on python 2.7. This conflicts with funpipe's python version. To use diamond, use it via docker.

Install via Docker

There's a bit more overhead using Docker, but it came along with the benefits of consistent environment (i.e.: including the operation systems). It's very useful when using funpipe on the cloud.

To use docker:

# Download docker
docker pull broadinstitute/funpipe:latest

# Run analysis interactively
docker run --rm -v $path_to_data/data -t broadinstitute/funpipe \
    /bin/bash -c "/scripts/vcf_qc_metr.py \
        -p prefix --jar /bin/GenomeAnalysisTK.jar \
        --fa /data/reference.fa
    "

You can use Dockerfile to compile the docker from scratch:

cd funpipe
docker build funpipe .

Install with PIP

This approach is for advanced users who don't like conda and want to integrate funpipe into their current working environment. Before starting pip installation, make sure the following list of bioinformatic tools (or a subset of tools of interest) are properly installed and add to your PATH. Path to Java tools (JARs) need to be specified when evocaking specific functions.

Requirements

• Python >= 3.7
• Bioinformatic tool collections: can be automatically installed via conda here
  • Basic functions:
    • samtools>=1.9
    • bwa>=0.7.8
    • gatk>=3.8
    • picard>=2.18.17
  • Phylogenetics:
    • raxml>=8.2.12
    • readseq>=2.1.30
  • CNV:
    • breakdancer>=1.4.5
    • cnvnator>=0.3
  • Microbiome:
    • pilon>=1.23
    • diamond>=0.9.22

To install with pip:

# install latest stable release
pip install funpipe

# install a specific version
pip install funpipe==0.1.0

To install the latest version: funpipe

git clone git@github.com:broadinstitute/funpipe.git
cd funpipe
pip install .

Documentation

Major analysis pipelines/tools:

• Quality control modules
  • Reference genome quality evaluation with Pilon.
  • FASTQ quality control with fastqc.
  • BAM quality control using Picard.
  • VCF quality control using GATK VariantEval.
• Variant Annotation with snpEff.
• Genomic Variation
  • Coverage analysis
  • Mating type analysis
  • Copy number variation with CNVnator
• Phylogenetic analysis
  • Dating analysis with BEAST.
  • Phylogenetic tree with FastTree, RAxML and IQTREE.
• GWAS analysis with GEMMA.

Here are scripts to run each of the above pipelines, use <toolname> -h to see the manuals.

##### Quality control #####
run_pilon.py          # Evaluate reference genome quality with pilon
fastqc.py             # Fastq quality control
bam_qc_metr.py        # Quality control of BAMs
vcf_qc_metr.py        # Quality control of VCFs

##### Variant Annotation #####
run_snpeff.py         # Annotation genomic variants with snpEff
phylo_analysis.py     # Phylogenetic analysis

##### Genomic Variations #####
coverage_analysis.py  # Hybrid coverage and ploidy analysis

You can also use out APIs to build your customized analysis scripts or pipelines. The docs will be available here: https://funpipe.readthedocs.io