BIO 202 — Lesson 21: Mutation target size and the parallel evolution it produces

Some genes are harder to change. We'll see those — like Hox genes. They're going to vary very little across all living things, all animals. On the other hand, things like microsatellites don't matter. Microsatellites are like the shitposting memes you make. Hox genes are like the word "mother." — 202_lec01_09

A — Two genes, two mutation rates

The word "mother" varies less across languages than the word "cup." Changing the word "cup" is easy — if I use a different word for cup, that's not a big deal. If, on the other hand, a lost child asks for their mother in a different way so that the person they're asking doesn't understand the kid is looking for their mother — that's a problem. Some things are important, so they change less. — 202_lec01_10

Predict first. A hundred separate populations, watched for the same time. A trait that many different mutations could produce will turn up:

In many of them, each arriving on its own In one of them, and then spread to the rest In none — any single mutation is far too unlikely

Two traits. The first needs one specific change. The second can be reached by any of several. Every individual change is exactly as likely as every other, and that never moves.

ways in to the second trait = 10 generations you watch = 300

reached trait one (one way in): —
reached trait two: —
chance of any one particular change: 1.0e-4 — identical for both, at every setting

B — Run it again and see what repeats

Vertebrate paralogs subfunctionalized: each picked up a piece of what the ancestral single Hox gene used to do. They aren't redundant. They aren't doing the same job in parallel. They're each doing a part of the job, and you need them both. That's what made the four-cluster duplication evolutionarily useful — it gave vertebrates more knobs to turn. — 336_lec11_06

TODO: parallel-evolution drill. Five traits classified by mutational target size; predict number of independent origins. SPEC: When a phenotype can be reached by mutating any of many genes, it will be reached many times independently. When only one gene can do it, the phenotype evolves once or never. Show five traits and predict how many independent origins each has.

C — The Alu in the regulator

At some point in the past, an Alu transposon jumped into a regulator of a Hox gene that activates the posterior region of the trunk. That region got broken because an Alu jumped into it. It took our tail. That is why apes don't have tails. The Alu jumped into a transcription factor binding site on the T-locus — that's the causal mechanism by which we lost the long tail. — 202_lec15_02

TODO: target-size diagnostic. Count of independent origins as a function of target size; a resampled range. SPEC: A specific case: an Alu transposon jumped into a regulator of a Hox gene that activates the posterior region of the trunk. That one event took our tail. Specify the test that distinguishes "big-target rapid evolution" from "small-target lucky single event."

D — Five phenotypes, five effective mutation rates

Your Hox genes are more than half a billion years old. As things that have made successful copies of themselves, they're doing amazing — with almost no changes. Nearly zero mutations in these genes for over half a billion years. That's impressive making more copies of yourself. — 202_lec31_03

TODO: five-phenotype classification panel. Stretch: connect to S17 scaffold; show that the "number of independent origins" pattern follows the target-size hierarchy. SPEC: Classify webbed feet, cave eye loss, lactase persistence, the placenta, hinged jaws by their effective mutational target size (scaffold S17 already drills the intuition; this lesson formalizes it).