Strainberry is a method that performs strain separation in low-complexity metagenomes using error-prone long-read technologies. It exploits state-of-the-art tools for variant calling, haplotype phasing, and genome assembly, in order to achieve single-sample assembly of strains with higher quality than other state-of-the-art long-read assemblers.

• System requirements
• Installation
• Usage
• Output files
• Example
• Command-line options
• Reference

Strainberry has been developed and tested under a Linux environment. It requires certain packages/tools in order to be installed/used:

• GNU bash (version 4 or later recommended)
• miniconda3

The simplest (and recommended) way to install Strainberry dependencies is through conda in an isolated environment (e.g., named sberry):

git clone https://github.com/rvicedomini/strainberry.git
cd strainberry
conda env create -n sberry --file environment.yml

The whole installation process should take about 5-10 minutes.

To make the strainberry command available, it is advised to include the absolute path of Strainberry's directory in your PATH environment variable by adding the following line to your ~/.bashrc file:

export PATH=/absolute/path/to/strainberry:${PATH}

Assuming Strainberry conda environment has sberry name, the following commands allow to update Strainberry to the latest (development) version.

cd strainberry
git pull
conda env create -n sberry --file environment.yml --force

Activate Strainberry conda environment:

conda activate sberry

Running Strainberry:

strainberry [options] -r FASTA -b BAM -o OUTPUT_DIR

where FASTA is a strain-oblivious metagenome assembly (e.g., generated with metaFlye) and BAM is a coordinate-sorted long-read alignment in BAM format. Both FASTA and BAM files are expected to be indexed with the samtools faidx and samtools index commands, respectively. Strainberry's output is stored in OUTPUT_DIR.

After Strainberry execution the conda environment can be deactivated with the command:

conda deactivate

The output directory of Strainberry has the following structure:

OUTPUT_DIR/
├── strainberry_n2/
├── strainberry_n3/
├── ...
├── strainberry_nK/
├── assembly.scaffolds.bam
├── assembly.scaffolds.bam.bai
├── assembly.scaffolds.fa
└── assembly.scaffolds.fa.fai

Strainberry output assembly is stored in the assembly.scaffolds.fa file. A minimap2-based alignment of input reads on the output assembly is also available in the assembly.scaffolds.bam file.

All sub-directories named strainberry_n{k} contain intermediate results of Strainberry iterations (e.g. the VCF of variants found by mapping reads to contigs) and log files. After a successful run of Strainberry, they could be deleted.

In order to verify that Strainberry has been correctly installed, it is possible to test it on a small dataset in the example sub-directory.

In order to generate a strain-oblivious assembly and a read alignment, we recommend to use metaFlye, minimap2, and samtools. Assuming these tools are available, it is possible to run the following commands, using 12 threads:

cd example
flye --meta --pacbio-raw reads.fq.gz --out-dir flye_out --genome-size 300k --threads 12
minimap2 -ax map-pb -t 12 ./flye_out/assembly.fasta reads.fq.gz | samtools sort >./flye_out/alignment.sorted.bam
samtools faidx ./flye_out/assembly.fasta
samtools index ./flye_out/alignment.sorted.bam

where --genome-size provides an estimate of the metagenome to metaFlye (not required from version 2.8). The assembly and read alignment are then available in the flye_out directory as assembly.fasta and alignment.sorted.bam respectively.

Given a strain-oblivious assembly (file assembly.fasta) and a long-read mapping (file alignment.sorted.bam), it is possible to run Strainberry using 4 threads as follows:

$ cd example
$ strainberry -r assembly.fasta -b alignment.sorted.bam -o sberry_out -c 4

Strainberry should take around 5 minutes to finish. The file assembly.fasta contains a single sequence which is a consensus of a small region of E. coli strains K12 and W. After a successful run of Strainberry, in the sberry_out directory, the file assembly.scaffolds.fa should contain two scaffolds (one closer to strain K12, the other closer to strain W).

usage: strainberry -r PATH -b PATH -o PATH [--nanopore] [-n int] [-s float]
                   [-c int] [-h] [-V] [-v]

Automated strain separation of low-complexity metagenomes

Required arguments:
  -r PATH, --reference PATH
                        Strain-oblivious assembly in FASTA format
  -b PATH, --bam PATH   Read alignment in BAM format
  -o PATH, --out-dir PATH
                        Output directory of Strainberry assemblies

Optional arguments:
  --nanopore            Input consists of Oxford Nanopore reads
  -n int, --max-strains int
                        Attempt strain-separation at most for the provided
                        strain multiplicity [5]
  -s float, --snv-density float
                        Minimum SNV percentage to consider haplotype blocks
                        [0.1]
  -c int, --cpus int    Maximum number of CPUs to be used [1]

Other arguments:
  -h, --help            Show this help message and exit
  -V, --version         Show version number and exit
  -v, --verbose         Verbose output

If you use Strainberry in your work, please cite:

R. Vicedomini, C. Quince, A. E. Darling, R. Chikhi, Strainberry: automated strain separation in low-complexity metagenomes using long reads, Nature Communications (2021). doi: 10.1038/s41467-021-24515-9