LAAVA: Long-read AAV Analysis

A software package and executable bioinformatics workflow for the analysis of recombinant adeno-associated virus (rAAV) products by PacBio long-read sequencing.

• For a full explanation of the methods and example results on public PacBio datasets, see the preprint paper on bioRxiv: Standardized Nomenclature and Reporting for PacBio HiFi Sequencing and Analysis of rAAV Gene Therapy Vectors
• For a summary of technical methods, AAV type/subtype definitions, and interpretation, see: Design and definitions
• For answers to frequently asked questions, see the FAQ

Installation & Usage

LAAVA can be used as an end-to-end Nextflow workflow, an interactive Docker container, or individual scripts in this codebase.

Standard Nextflow (recommended)

This code can be run as a standard Nextflow workflow. When run this way, the workflow will automatically pull in the analysis scripts and their dependencies as a Docker image.

To get started, create a JSON file with your parameter values, similar to params-local-small.json in this repo, and run it with:

nextflow run -profile local -params-file <your-params-file.json> main.nf

Interactive Docker (laava container image)

For exploratory analysis or troubleshooting, you can also run the laava docker image directly on the command line as an interactive container.

Assuming you have Docker installed, fetch the container image:

docker pull ghcr.io/formbio/laava:latest

Then run it interactively in your current directory:

docker run -v $(pwd):$(pwd) -w $(pwd) -it ghcr.io/formbio/laava_dev:latest bash

For developers

You can directly download or clone the repo to use the scripts directly.

$ git clone https://github.com/formbio/laava.git

There are several ways to satisfy the script dependencies locally.

Option 1: Development docker image (laava_dev container image)

The laava_dev.dockerfile in this repo installs the scripts' dependencies, but not the scripts themselves, into a Docker container image that you can then use to run the local copies of the scripts.

To build the container image with the name laava_dev (you can use another name if you prefer):

docker build -t laava_dev:latest -f laava_dev.dockerfile .

To run the container in the current working directory:

docker run -v $(pwd):$(pwd) -w $(pwd) -it laava_dev:latest bash

This opens a Bash shell with the scripts in the PATH, and the original working directory mounted in place.

Option 2: Conda

The conda (or mamba) channels and dependencies are in the configuration files laava.conda_env.yml and laava_dev.conda_env.yml. These two environments are similar, and both will work for running LAAVA itself, but _dev includes some additional developer tools.

First, install conda via Miniconda or Anaconda. Then, for example, suppose you have anaconda installed and the binary is in $HOME/anaCogentPy37/bin. To make the installed scripts available in your environment, you would add the binary to $PATH if it isn't there already:

$ export PATH=$HOME/anaCogentPy37/bin:$PATH

Next, use the YAML configuration file to create a new conda environment and install its dependencies:

$ conda env create -f laava.conda_env.yml

Finally, once installation completes, activate the new environment:

$ source activate laava

At this point the prompt should change to (laava) $ and the executable scripts should be available in your PATH.

Option 3: Manual installation from scratch

Bypassing all the above, you can use other package managers to install the dependencies individually.

The prerequisites to run these scripts include:

• Python 3.7 or later
• R 3.6 or later

Python packages:

• biopython
• pysam
• parasail-python

R packages:

• tidyverse
• flextable
• Rmarkdown

Testing

The test/ subdirectory in this repo contains small example input files and a Makefile to run the scripts to reanalyze them and produce example HTML and PDF reports.

Once you've completed installation (above), activate your conda environment or Docker container and change to the test directory:

cd test

To generate the HTML and PDF reports from the test dataset included in the repo (this takes about 1-2 minutes):

make