This should be re-titled something like: with 200x longer sequences and making products without culturing, dirt can make antibiotic gold.

The two prospects:

Erutacidin, disrupts bacterial membranes through an uncommon interaction with the lipid cardiolipin and is effective against even the most challenging drug-resistant bacteria.

trigintamicin, acts on a protein-unfolding motor known as ClpX, a rare antibacterial target

The difficulty with bacterial DNA is that they have common elements and actively share DNA to boot. Sequencing only short sections make genome assembly unreliable. 200x longer sequences makes much more accurate genomes.

Then even if you find genes, we can't usually culture enough bacteria to make the product (typically instead injecting the sequences into bacteria we can culture). So being able to make the product without culturing the organism is key.

Spoiler, I haven't read the article, but my understanding is cardiolipin targeting antibiotics have failed in the past because our mitochondria are enriched for it. (Which makes sense here because the mitochondria are derived from ancient bacteria). I'm sure there is potential for optimization for medical applications, but we will have to be very careful for adverse effects.

Edit: nvm, brain fart. OP is correct.

> So being able to make the product without culturing the organism is key.

No, it isn't. The article talks about using chemical synthesis, rather than using a biological platform to express the product via genes.

"To convert the newly uncovered sequences into bioactive molecules, the team applied a synthetic bioinformatic natural products (synBNP) approach. They bioinformatically predicted the chemical structures of natural products directly from the genome data and then chemically synthesized them in the lab. With the synBNP approach, Brady and colleagues managed to turn the genetic blueprints from uncultured bacteria into actual molecules—including two potent antibiotics."