COALA dataset from 'Transfer learning improves antibiotic resistance class prediction'

This includes all 3 different versions of the COALA dataset.
COALA100 dataset is a collection of antibiotic resistance genes from 15 databases along with metadata information from these databases which includes the respective antibiotic class.

The COALA70 dataset is the cd-hitted by 70% threshold version of the COALA100 dataset.

The COALA40 dataset is the cd-hitted by 40% threshold version of the COALA100 dataset.

All three datasets are in fasta format. The last section of the description line has the antibiotic label the gene confers resistance to. The second last section is the database name from where the gene was collected. All other sections convey information about the gene.