snafees / spid.jl

SPID: SNP Pipeline for Infectious Disease

This package implements a simple alignment and SNP-calling pipeline for detecting viral and bacterial infectious disease outbreak clusters.

Installation

The easiest way to use spid is through the docker container czbiohub/spid.

Manual installation

If you prefer not to use the docker container, you can manually install as follows:

1. Install dependencies:
   • minimap2 >= 2.15
   • samtools >= 1.9
   • julia >= 1.0
2. Add the top-level folder as a local julia package
   • You can do this by typing ] and then add . in julia
3. (Optional) Add the executable bin/spid.jl to your $PATH.

Usage

Spid can be run from julia by importing it with using Spid, or from the command-line as spid.jl [command].

There are 3 main commands:

• align_short_reads: Align short reads against a reference.
• align_assembly: Align already assembled fasta against a reference.
• merge_alignments: Combine samples generated from align_short_reads and align_assembly into a single dataset, and compute pairwise distances between samples.

To see the help for these functions, run spid.jl [command] --help from the command line, or ?{command} within julia.

Running through docker

First pull the image from dockerhub:

docker pull czbiohub/spid:<version>

To run the command-line utility, with the current directory mounted at /data inside the docker image, do:

docker run -v $PWD:/data czbiohub/spid spid.jl [...]

To run julia through the docker container, with the current directory mounted at /data, do:

docker run -it -v $PWD:/data czbiohub/spid julia

Implementation Details

The pipeline aligns samples against a reference genome using minimap2, then uses samtools to perform an mpileup, and then runs julia code to call a consensus allele at each position according to user-specified filters. It also runs julia code to compute the SNP distances between every pair of samples.

The pipeline is meant to work well in an online setting where samples are added incrementally, without affecting previous results. Thus, alleles are called individually rather than jointly across samples. It also reports a whole-genome SNP distance between every pair, which does not change as new samples are added, in addition to the more traditional core-genome SNP distance which does change as samples are added (since the core genome size shrinks with sample size).

This pipeline is only meant for viral and bacterial haploid sequences. It will not work for diploid or polyploid eukaryotic sequences.