I'd never thought they would actually assemble this genome. But put together the father of string-graph assembly (Eugene Myers) and a shitload of money (PacBio reads), and you get an assembly of a 32gb genome.
3 billion bp in a haploid human genome, 6 billion in a full diploid one. If you store the masked/unknown/variable ranges as annotations on the side, you can get away with close to 2 bits per base, as you suggested.
The axolotl haploid genome is 32 Gbp, which, in 2-bit encoding and ignoring masked bases, takes 8GB to store.
Axolotl, like humans, is diploid, meaning it has 2 copies of each chromosome.
That’s only about 10 times the size of the human genome.
That being said, the complexity of a genome doesn’t scale with size. Many plant genomes are gargantuan but highly repetitive. The Kolmogorov complexity of a genome would better predict the complexity of the organism than a genome’s length would.
Flash memory has considerably better read/write speeds. It can also be read electronically instead of chemically. I don't think it makes sense to compare them.
