This repository contains all the workflows and the scripts needed to reproduce the experiments about MALVIRUS accuracy.
The only software prerequisite is the open source package manager conda, needed to retrieve and install all the other software used in the experiments. Notice that the experiments have been run on Ubuntu 20.04 and we did not test them on different OSs (although MALVIRUS is cross-platform).
After you cloned (or downloaded this repository, in order to install all the software and retrieve the read sample data from SRA you should use the following command:
bash ./prepare-environment.shSince the terms of use of GISAID do not allow us to redistribute their data, you should also download them from GISAID and copy them into the repository.
In particular, you have to download the package containing all the sequences in FASTA format (the filename should be sequences_fasta_<date>.tar.xz), you should decompress it and extract the file sequences.fasta. Then you have to recompress in XZ format and place it in the directory catalogs/input/gisaid/<date>/ with the name sequences.fasta.xz.
Similarly, download the metadata package (the filename should be metadata_tsv_<date>.tar.xz), extract the file metadata.tsv, recompress it in GZIP format and place the resulting file in the same directory.
Then, update the file catalogs/Snakefile at line 5 by replacing "210407" with the date indicated in the files you previously downloaded.
Finally, download the sequences whose accession IDs are listed in file gisaid_accession_ids.txt and place each one of them in a file <GISAID_ACCESSION_ID>.fasta of directory experiments/input/gisaid. Replace all the FASTA headers with >sample.
After all the software and the data have been retrieved, you can run the workflow contained in directory catalogs with the command (replace 4 with the number of CPU cores you want to use for the execution):
snakemake --use-conda -j4When all the catalogs have been built, you can test MALVIRUS accuracy on the read samples using the workflow contained in directory experiments. You should use the following command to start the workflow on every catalog (from directory experiments):
ls vcf | ./multi-exec.shAt the end, you will have a set of CSV files experiments/output/<catalog name>/summary.k31.csv summarizing the results for each catalog.