Setting the record straight on ejaculate

[elfs]

So, apparently there's this meme going around that a single ejaculation containes something like 1.5 terabytes of data. However, after de-duplication and BWT-based compression, you'd only end up with something like 15KB of Shannon information.

See, this is what comes (ahem) of me encountering this meme while I'm in the midst of a crash course on bioinformatic visualization techniques.

Comments

[atheorist.livejournal.com]

Note that if a sufficiently advanced entity were trying to communicate something with ejaculate, they could use protein configurations and populations, organelle positions, all kinds of extra-dna channels. Terabytes seems reasonable if you confine your attention to just DNA sequences.

I can't replicate the 15kb figure, though; even if you're compressing relative to some reference genome, the best relative compressor of genomes in wide use today gets something like 3MB (http://sun.aei.polsl.pl/gdc/).

[en-ki.livejournal.com]

...and that requires a reference genome, so it's less "compression" and more "diff".