GBADN

Back to the console

A genome study of Pokémon Emerald

What if a game cartridge were sequenced like an organism? These are the usual figures of a genome paper, measured on one game. The numbers are real, the biology is a joke: a chromosome here is a best guess at what a part of the ROM holds, and those labels can be wrong.

The figures read the decoded ROM at 4 bases per byte. The FASTA file itself stores compressed data, which looks like random sequence. The figures hold only measurements, never any part of the game.

67.1 Mbp in 803 contigs, N50 4.3 Mbp, L50 6
ChromosomeLengthGCContigs
chrData24.6 Mbp37.2%397
chrAudio17.5 Mbp19.5%18
chrEmpty11.9 Mbp0.0%18
chrGraphics6.6 Mbp31.7%277
chrCode6.5 Mbp35.3%92
chrHeader768 bp42.7%1

Curiosity: Emerald's genome twin

Front cover of Pokémon Emerald Version, showing the legendary Pokémon Rayquaza.
Pokémon Emerald. Cover art © Nintendo, Game Freak and Creatures, shown from Wikipedia. GBADN isn't affiliated with them.
Microscope photo of an adult Oikopleura dioica: a translucent rounded trunk with a golden gut and the start of its tail.
Oikopleura dioica. Photo by Mowgli, CC BY-SA 4.0, scaled down, from Wikimedia Commons.
Pokémon EmeraldOikopleura dioica
Genome67.1 Mbp65–70 Mbp
Chromosomes6 (best guesses)3 pairs
Life cycleUntil you stop playingAbout 4 days at 23 °C

Oikopleura dioica is a tiny tadpole-shaped sea creature that filters food through a mucus house it keeps abandoning and rebuilding. It is a tunicate, the sister group of vertebrates, and has one of the smallest genomes of any sequenced animal. The match is only in size: Emerald reaches 67.1 Mbp at 4 bases per byte, and estimates for the animal vary between populations and methods. Sources: Bliznina et al., 2021, larvaceans on Wikipedia.

Karyotype of six chromosomes ranked by size, from chrData at 24.6 Mbp to chrHeader at 768 bp. chrEmpty is solid dark and chrAudio ends in a long dark band.
Karyotype. Empty space stains solid black because padding has almost no G or C. The end of chrAudio stains the same way: it may be padding that the classifier filed as sound.
Circos plot with rings for chromosomes, GC content and byte entropy, and ribbons joining repeated stretches. The thickest ribbons join chrGraphics to itself and to chrAudio.
Circos plot. The search found 297 exact repeats of at least 2 kbp, covering 1.3 Mbp. 74 of them cross between chromosomes, so some of the same bytes received different labels.
Contig map along the ROM, an Nx curve with N50 at 4.3 Mbp, and cumulative length by chromosome. Code sits at the start of the ROM.
Assembly report. The contig map doubles as a map of the cartridge, with code first. chrData and chrGraphics are fragmented into hundreds of contigs, and six contigs hold half the genome.
Self dot plot of four chromosomes. Short marks cluster near the diagonal, and a few long lines sit in chrGraphics and between chrGraphics and chrAudio.
Self dot plot. Lines off the diagonal are long duplications, the same signal as the thick Circos ribbons.
Base composition by chromosome, GC content histograms per window, and a 21-mer spectrum on log scales falling steadily from 23 million unique 21-mers.
Sequence quality. Code has a narrow GC peak, while audio varies widely. Most 21-mers occur once, and one occurs about 8 million times, even with the empty chromosome left out.

Make these for your own cartridge with gbadn-study.