What is the highest data compression ratio achieved on a population of genomes measured to include the compressing algorithm itself? The size of the compressed database approximates its algorithmic information content*.
I've seen a claim of on the order of 1000:1 compression but it appears dependent on an external "reference" database. In my opinion, a reference data must be included in the size of the compressing algorithm.
*The Universal Turing Machine notion of Minimum Description Length is also called the Kolmogorov Complexity or Algorithmic Information measure of the data. I'm using the term "Algorithmic Information" as the most concise and accessible descriptive phrase for this measure.
The relevance of this measure is that according to the closely related theory of algorithmic probability, the algorithm that maximally compresses a dataset also provides the most optimal data-driven predictions.