Slot Machine: A Compression Transform for Multi-FASTA Files

The large size of genome files and their current widespread use in many fields makes compression of these files very important for their efficient storage. However, many compression tools, e.g. gzip, require the entire file to be decompressed before any part can be accessed. With this problem in mind, we introduce a reference-free compression transform for genome files that support compressed pattern matching - namely, the capacity to search the compressed file for information contained in the original file. Our transform, Slot Machine, leverages the distribution of characters in multi-FASTA files to decrease their size by approximately 66% while allowing a user to search and decompress only the parts of the file needed at any given time. As an added bonus, the output of Slot Machine can be further compressed by pairing with existing compressors. In our experiments, we achieved compression-ratios of up to 5.25, or a compression rate of 80.96%. In that measure, our results improved upon NAF in every experiment and are comparable to MFCompress.