Data supporting "Transformer Model Generated Bacteriophage Genomes are Compositionally Distinct from Natural Sequences"

Sequence and composition data supporting doi: 10.1101/2024.03.19.585716. Uncompressed file size is ~5.8GB.

Data in zip files is organized by sequence provenance (generRNA, natural, or transformer (megaDNA)). Common file types between folders include:

Multi-record fasta file: Sequence data for all sequences of a given provenance. For generRNA sequences, these are found within the `seq` column of file "MFE_distribution_Fig4a.csv"

Composition files: Individual sequence level compositional metrics for sliding 120 bp windows. Only structural metrics were used in this study.

Genomad: Results from the genomad pipeline (https://portal.nersc.gov/genomad/)

Stats: Aggregate statistics for all sequences of a given provenance.

The natural folder also has a metadata file detailing the taxonomy for all natural sequences.Figure datasets are the cleaned (sometimes aggregated) datasets that underly specific figures in the manuscript. The figure designations are based on the order in: https://www.biorxiv.org/content/10.1101/2024.03.19.585716v1.