Prediction of hosts susceptible to SARS-CoV infection using ACE2 protein sequences

Code and data used in Mollentze N, Keen D, Munkhbayar U, Biek R, Streicker DG. (2022). Variation in the ACE2 receptor has limited utility for SARS-CoV-2 host prediction. eLife 11:e80329. doi:10.7554/eLife.80329.

Requirements

• Install the mamba package manager
• All other requirements can then be installed using:

mamba env create -f environment.yml

Repeating published analyses

To repeat all analyses and recreate the figures used in the manuscript, run:

conda activate sars_susceptibles
make

Usage notes

1. Most required external data will be downloaded automatically, with two exceptions:

   • ACE2 sequences. The Makefile contains code to download these, but any sequence updates by NCBI will mean sequences matching the earlier accession number used here cannot be dowloaded (easily). The sequences used here are therefore included in data/external/ace2_protein_sequences.fasta.
   • IUCN range data (required to recreate map figures). Downloads require a log in, see data/iucn_range_maps/ for instructions.

2. This pipeline was designed for workstations, and may need some minor modifications to run on a regular PC:

   • By default, 20 parallel threads will be used. Edit the --n_threads argument throughout the Makefile to change this
   • Creating ensemble models may use large amounts of memory. See scripts/train_ensemble.R for instructions if this becomes an issue.