Baby is healed with first personalized gene-editing treatment

How does it know how to gps around? From what I know everything down there is a chemical reaction with some minimal physical motion, but how do you program it to know where to change and what and how.

It’s more like a “ctrl+F” for DNA. Hopefully there’s only 1 match (the target site).

Add gene has a great guide as to what goes on at the molecular level: https://www.addgene.org/guides/crispr/

Essentially you can design an rna molecular that contains a 20 nucleotide long sequence that can target your region of interest, with the caveat that there is a standard recognition sequence proximal to your sequence of interest (PAM sequence)

So you create a molecule that binds to a certain location in the dna, and then deploy a billion of them?

It doesn’t know anything about where it “needs” to go. One of the weirder and more unintuitive things about molecular biology is just how fast everything moves inside a cell. The CRISPR molecule diffuses from one side of the nucleus to the other in a couple seconds and probably bumps into the entirety of the genome in a matter of minutes or hours. It’s very, very crowded inside cells, proteins and DNA and metabolites are constantly bumping into each other and are tumbling around at frankly incomprehensible rates. So, nothing needs to “know” where it needs to go, it simply gets pushed and jostled around until arrives there and then the attraction between the CRISPR’s RNA and the DNA takes over

More or less, yes.

An interesting part of the study was determining what a clinical dose _should_ be. You need enough to edit enough liver cells. But don’t really want to completely overdo it to limit potentially negative side effects. Seems like they got it right enough here, with the first dose having some effect and the subsequent dose having more.

Well its more like search and replace, where you cross your fingers that it only replaces the words you are trying replace without impacting the rest of the text in the document.

You need billions to cover multiple cells, you don't need many for a cell.

The counterintuitive part is how fast thermal motion is relative to the size of dna.

In body temperature water, the thermal velocity of water molecules jostling around everything is ~600m/s. The nucleus of a human cell is ~6µm in diameter. That is, your average water molecule bounces around at a speed that makes it move from one end of the nucleus to another roughly 100 million times per second.

Larger molecules move more slowly, but they still zip around fast enough that nothing needs to "seek" to a specific position in a cell to get there, everything will touch everything just from thermal random walk in a very short time. So how biology works is that inside the cell there might be just one messenger, which will have to hit a specific piece of dna just right in order to do anything, but that's still nearly instantaneous from our perspective.

This sounds so much like "simulated annealing" with reactive components and almost no lack of energy in the system. Various energies/reactions occur, which unlock or lock out other possible reactions.