From Letters to You: Genes & Proteins
Where we left off
For three tutorials I've been waving four letters at you — A, C, G, T — and pointedly refusing to say what they do. We built the twisted ladder, learned the two pairing rules, watched the copy machine run. But a book you can't read is just decoration. You've been staring at three billion letters with a completely fair question forming: so what? What is any of this actually FOR?
This tutorial is the answer, and it's the moment genetics stops being about molecules and starts being about you — your body, your hemoglobin, the color of your eyes.
One spine, this whole tutorial: buried in the three-billion-letter book are short meaningful passages called genes. Each gene is a recipe. What it builds is a protein. And proteins are the machines and building blocks that make and run every part of you. Letters → recipe → machine → you. That's the whole chain, and by the end it'll feel obvious.
First, the thing nobody tells you: most of the book is not instructions
Here's a fact that reorganizes everything. Of your three billion letters, only about 1 to 2 percent are actual recipes. The other ~98% is not genes — it's spacers, switches, old broken copies, and long stretches whose job we're still arguing about.
Sit with that. The "book of life" is mostly not recipes. It's like a 1,000-page cookbook where maybe 15 pages have recipes on them and the rest is... blank margins, coffee stains, notes, and pages torn out of other books and taped in. For decades biologists called the non-recipe part "junk DNA." That name is now a mild embarrassment — a lot of it turned out to be the switches that decide when and where each recipe gets used, which is arguably more interesting than the recipes. But the headline stands: a gene is a rare, special stretch. Most of your DNA is not one.
Why start here? Because it kills the beginner's mental picture — "DNA is a list of genes, back to back." It isn't. Genes are islands in a very large ocean of other stuff. Hold that image; it comes back hard when we get to how populations differ, because most of the differences between people sit outside the genes.
So what IS a gene?
A gene is a stretch of the letter sequence — could be a few hundred letters, could be a couple million — that spells out the instructions to build one specific product. For the huge majority of genes, that product is a protein.
That's it. A gene is a passage with meaning, the way a recipe is a meaningful passage inside a cookbook full of blank pages. ...TTACGCATGGCA... isn't obviously special until you know it's the start of the recipe for, say, the protein that carries oxygen in your blood.
And crucially: a gene doesn't DO anything on its own. It's inert text. It's a recipe card in a drawer. The doing is done by what the recipe builds.
The star of the show: proteins
If genes are the recipes, proteins are the food — except the "food" is the actual working substance of you. This is the most under-appreciated fact in all of biology, so let me say it plainly:
Almost everything your body IS, and almost everything your body DOES, is proteins.
- The stuff you're made of, structurally: keratin (hair, nails), collagen (skin, tendons, the scaffolding holding you together) — proteins.
- The machines that do the work: hemoglobin ferrying oxygen through your blood, the enzymes in your gut dismantling your lunch, the antibodies hunting invaders, the tiny motors hauling cargo around inside your cells — all proteins.
Muscle? Protein. The receptors in your eye that catch light? Proteins. When people say "you are what your genes make you," what they literally mean is: your genes are recipes for proteins, and proteins are what you're built from and run by.
What a protein actually is (this part is beautiful)
A protein is a chain of beads folded into a shape.
The beads are amino acids — there are 20 kinds, think of them as 20 colors. A protein starts life as a long floppy string of these colored beads, in a specific order: red-red-blue-green-yellow-blue-... anywhere from a few dozen to thousands of beads long.
Then the magic: the string doesn't stay floppy. Following nothing but the push and pull between its own beads, it folds up on itself into a precise 3D shape — a blob, a channel, a claw, a hinge. And here is the sentence to tattoo on your brain:
The shape is the job.
Hemoglobin folds into a shape with a little pocket that grips an oxygen molecule and lets go at the right moment — that shape is what "carrying oxygen" means. A digestive enzyme folds into a shape with a groove that a food molecule slots into so it can be snipped — that shape is "digestion." Change the folded shape and you change (or break) the job. Function follows form, all the way down.
So the full chain, once more, now with the mechanism filled in:
Gene (a passage of letters) → spells out → the bead order → the beads fold into → a shape → the shape does → a job → enough jobs add up to → a trait, a working body, you.
How do letters name beads? The genetic code
One gap left: how does a string of four letters (A/C/G/T) specify a string of 20-color beads? Four things can't name twenty things one-for-one. So the cell does the obvious workaround — it reads the letters three at a time.
Three letters, each of 4 kinds, gives 4 × 4 × 4 = 64 possible triplets — plenty to name 20 beads (with room to spare, so several triplets map to the same bead, and a few act as punctuation like "STOP here"). Each triplet is called a codon, and the lookup table — which triplet means which amino acid — is the genetic code. GCA → one particular bead. TGG → another. And this table is nearly identical in every living thing — you, a banana, the bacteria in your gut all read GCA the same way. That deep universality is a preview of a much bigger idea coming later: we're all running variations of the same ancient software.
You don't need to memorize the table. You need exactly one thing: letters are read in triplets, and each triplet names the next bead in the protein. That's the decoder ring that turns text into a machine.
(There's a gorgeous molecular machine, the ribosome, that physically trundles along the gene reading codons and clicking the named beads into a growing chain. We'll meet it properly later; for now just know the reading-and-building is done by dedicated machinery, not by wishful thinking.)
The payoff: why this unlocks the podcast
Here's where three tutorials of groundwork suddenly pays rent. When Reich and Dwarkesh talk — and they do, constantly — about "a gene for X," you now know exactly what that phrase means and doesn't mean:
- They discuss FOXP2, "the language gene" — a gene where just two letter-changes on the human lineage tweaked the protein it builds. Engineer those same changes into mice and the mice literally squeak differently. One recipe, edited slightly → a slightly different protein → a changed trait. (Reich's twist: Neanderthals had those same two changes, so FOXP2 can't be the thing that made us special — but you can't even follow that argument until you know a gene is a protein recipe.)
- They discuss the Tibetan high-altitude gene — a single gene variant, inherited from the archaic Denisovans, that lets Tibetans thrive on oxygen-thin mountain air. A different recipe → a differently-tuned protein → a body that handles altitude.
- They discuss the genes behind skin color, eye color, hair — ancient Europeans had dark skin but blue eyes because different pigment genes, building different pigment proteins, came from different ancestral populations that later fused.
Every one of those is the same chain you just learned: change the gene → change the protein → change the trait. That's the atom of the entire field. Without it, the podcast is people saying "gene" a hundred times and you nodding politely. With it, you're in the room.
And remember the ocean around the islands: because genes are only ~1–2% of the genome, most of the letter-differences between two people fall outside any gene — which is exactly why, many tutorials from now, those "silent" differences become the perfect molecular stopwatch for dating history. The junk isn't junk to a historian.
The challenge
Three questions, escalating:
Warm-up. A gene is a recipe. In one sentence, what's the product it builds, and what makes that product able to do a job? (If you can say "a protein, and its folded shape is its function," you've got the spine.)
The real one. Suppose a single letter in a gene gets miscopied — remember the "hairline crack" (mutation) from last tutorial — and it changes one codon, so one bead in the protein comes out the wrong color. Walk the chain forward: what could happen to the protein's shape, and therefore to its job, and therefore to the person? (This exact scenario is sickle-cell anemia: one letter, one wrong bead in hemoglobin, a warped shape, misshapen blood cells. One typo, echoing all the way up to a human being.)
Think ahead. If only ~1–2% of your DNA is genes, is a random letter-difference between you and me more likely to land inside a gene or outside one? And why might a difference that lands outside every gene be genetically "quieter" — easier for evolution to ignore? (You're reinventing the logic of the molecular clock. We'll build it for real soon.)
Next: your DNA is two metres long and only ~1–2% of it is recipes — so how is all of it crammed into a cell far too small to see? That's where **chromosomes* come in.*