Sure, you could do that.[1]

Or.... you could play with the full set of SNPs.

Imagine playing Top Trumps, only instead of 1-10, you only have the base pairs, ACGT to choose. Be simple and say that alphabetical order decides it - Adenine beats Cytosine beats Guanine beats Thiosine[2]. So each person starts with a full whack of identical cards representing themselves and has to choose a base pair to compare. If they win, they win a point, and so on til you run out of cards.

Boring, and impossible.

(I think I'm getting 600,000 base pairs back. That won't fit on a card.)

But then apply bioinformatics techniques to it. Instead of playing, develop an AI to play.

I mean, you can see corpuses for each of these base pairs - you know if you've got a good chance or not.

(Make it more interesting - restrict each person to picking each base pair only once. So there are > 600,000 moves)

So far so - compare your cards against the likelihood of winning, slowly work down that sorted list.

But then consider - each of these base pairs will be correlated with other base pairs. So you can start to take the correlation data into account.

(But of course, the game actually becomes very boring past where you see the other's card - at that point you share in the hidden knowledge, and that's all the fun of this game. Which is why we can't play with the others cards, or even see them. Past what has been revealed through play. In fact, maybe it would be more interesting if you can't see your card details either.)

And maybe you could start getting real crazy - start taking real world data and medical histories and use that as input to the game. Oh, your dad got cancer? Then let's see your pair #908709 - I've got G.

There you are - a game loosely based on Top Trumps that is actually a bioinformatics project.

Let's never make this.

[1] We should do this, it's a far better idea than I outline here.

[2] I am fucking astounded I remembered what the bases in DNA are, and how to spell them.