Preprint: https://www.medrxiv.org/content/10.1101/2023.09.08.23295250v1
- Install Nexstrain CLI following the instructions at https://docs.nextstrain.org/projects/cli/en/stable/installation/
- Setup a runtime, e.g. "docker" using
nextstrain setup docker - Check your installation with
nextstrain check-setup --set-default
The easiest way to make you have all the right software versions, you can use the docker image corneliusroemer/ba286 with nextstrain build.
To use that docker image, you need to add the --docker --image=corneliusroemer/ba286 flags to the nextstrain build command.
Note: If you don't use the docker image, you need to make sure you have treetime v0.11.1 (or higher) installed which is at the time of writing incompatible with augur. That's why it's easiest for now to use docker and not worry.
For the paper, we sampled the BA.2 background set based on an ncov-ingest curated GISAID data dump and downloaded BA.2.86 sequences from GISAID's web interface.
Unfortunately, we cannot share the ncov-ingest curated data due to GISAID's data sharing policy. However, to at least partially reproduce the subsampling part of the analysis, you can use ncov-ingest's "open" data, which is based on Genbank data and is freely available but less globally representative and balanced. To do so, you need to add --config data_source=open to the workflow invocation (or set that option in config/config_dict.yaml).
The other option is to use the exact BA.2 sequences from the paper. To do so, download the "Input for Augur pipeline" tarball for EPI_SET_230907cf from GISAID and place the archive at data/background.tar. Then, add --config data_source=frozen to the workflow invocation (or set that option in config/config_dict.yaml).
We generated the build using --config data_source=gisaid.
Besides the BA.2 data, you also need to download BA.2.86 sequences from GISAID. To reproduce the analysis in the paper with the exact BA.2.86 sequences, download EPI_SET_230909co and place the archive at data/BA286.tar. Alternatively, you can use the sequences from the paper by adding --config data_source=frozen to the workflow invocation (or set that option in config/config_dict.yaml).
Alternatively, you can download any set of BA.2.86 sequences (for example more current) and also use them for the analysis. Just download the "Input for Augur pipeline" tarball from GISAID and place the archive at data/BA286.tar.
Run the workflow with:
nextstrain build --docker --image corneliusroemer/ba286 .You can then view the resulting tree with Auspice using:
nextstrain view auspice/BA.2.86.json