Comment by WhompingWindows | Hacker News Reader

Comment by WhompingWindows | Hacker News Reader

WhompingWindows

I worked in Syn Bio from 2008-2010, long before CRISPR came onto the scene. I worked on a couple of iGEM projects, that's the international genetically engineered machines conference at MIT.

The basic premise was we would design E. Coli "strains" or "models" that had foreign DNA inserted via commonly-shared genetic tools. It is comparable to using Linux as a shared OS, lots of other FOSS developers helping each other... except that to get totally new DNA into your E. coli, you had to have it shipped in on dry ice from a willing lab halfway around the world. So, you can see why the field is much smaller and more niche than anything software-related...downloading a library is WAY faster and cheaper than shipping in a new genetic module.

CRISPR would've been a game changer for us. We could've much more easily edited gene sequences, though it still would've been hard to get foreign DNA from other labs...I don't believe CRISPR is good for writing huge new sequences... But small changes, small modifications, would've been MUCH faster with CRISPR. We could've gotten a foreign DNA sequence and then tried out 100 different mutations very simply, perhaps discovering an even better functionality than nature's original code.

Crispr is not necessary for e coli mutagenesis. For plasmids you really want to use Gibson assembly. For chromosomal modification lambda red is as easy as crispr.

Gibson assembly is basically foolproof. I trained an intern with almost no molecular biology wetlab experience and in three months he designed DNA for and made fifty mutants (I picked whuch mutations to try) and ran biochemistry on half of them:

https://pubmed.ncbi.nlm.nih.gov/24934472/

For yeast, you need neither. Just drop the DNA in and the yeast takes it up!

You're right that you don't need CRISPR for genome engineering of S. cerevisiae (baker's yeast) or E. coli.

However, having worked with both prokaryotes and eukaryotes it definitely makes life easier. During my PhD (about metabolic engineering of S. cerevisiae) CRISPR started to get traction. Before I could incorporate one change at a time. Using CRISPR/Cas9 I could do up to 6 (at different places). That's a big deal!

I also PoC'ed CRISPR in a bacterium (C. glutamicum). Also there it opened up new avenues.

I had a guest lecture from a couple of researchers working on bioproduction that mentioned crispr let them speed up their design process for new e coli strains, but yeah in general it's not too difficult to hack together a pathway with a Gibson (unless you want to do more than 5 parts in the assembly which gets finicky). Or a golden gate if you can get the sequences for the overhangs right. But I will say I've worked in a lab with at least one person who wasnt able to perform a relatively simple gibson somehow, so the principle that the universe will always find a greater fool still applies

OK, so gibson assembly is not completely foolproof, when I first started working with it I thought the whole thing was a joke because I spent a month trying to get even one clone, and it didn't take. Turned out the site we were trying to (TetR IIRC, it's been 10 years so I'm not sure) clone into had a strange tandem inversion sequence flanking it. If you stick to cloning into MCSes or assembling sequences you've checked/built yourself it should be fine (the liklihood of problematic random tandem inversions is low).

a 1:1 cloning is very much doable with terrible hands. You can reasonably do up to 8 pieces in the assembly, but as you get beyond 4 pieces it becomes important to follow the instruction exactly.

The craziest thing I did, though, was I cloned a (small) eukaryotic chromosome into an E Coli, by gibsoning against the telomere sequence.

At some point the difference becomes using very specially prepared competent cells and being very careful about how you get them in (the mechanism of electorporation is... not obvious), and of course one of the tricky things about gibson is that it's super salty, meaning that you have to be careful when electroporating.

For doing mutagenesis on plasmids, you don't even need gibson. Overlapped primers do some weird circularization during PCR, so for point mutations you don't even need gibson mix.

I'm personally a big fan of synthesis + GoldenGate for building genetic circuits. Gibson, IMO, destroys reproduciblity (how many different PCR kits are there?) and leaning towards Gibson assembly over enzymatic assembly methods (GoldenGate, maybe even BioBricks) makes biotech an Ivory-Tower exercise, since not everyone has access to oligos.

> since not everyone has access to oligos

I don't know about that. I've bought 50-mer oligos at $20 a pop to do a 6-part assembly in my garage (I made a strain of yeast that generates cannabinoids). DNA oligos "basically last forever" if you handle them relatively correctly, even in a terrible defrost-cycling consumer freezer. Enzymes, especially specialty ones, not rock solid ones like modern TAQ, have a very short halflife in your freezer, especially if your freezer is consumer-grade.

> how many different PCR kits are there?

Yeah, just buy NEB. Seriously. Not worth screwing around with companies that have lower QC standards.

NEB doesn’t ship residential, and there are many parts of the world that you can’t buy $20 oligos from. I like doing pretty complicated builds, and so I’d go bankrupt pretty much immediately because I don’t just build one thing.

I've gotten neb to ship residential...

How is gibson ivory tower? It's literally just based on overlapping sequences. And how is it hard to reproduce? You select out colonies and verify with sanger. These are not things that are inaccessible to most people, especially if you live near a DIY bio hacker space, and if you get involved in the biohacker community you'll get a sense of what providers ship to residential address, where to get certain equiptment, and so on. Just googling will get you information that can help you figure out how to get what you need.

https://medium.com/@ThatMrE/a-guide-to-diybio-updated-2019-a...

Reordering of oligos is expensive, which makes people not reproduce. Sanger sequencing is pretty expensive ($5 sure, but over 5kb on 3 different constructs becomes pricey). PCR machines also ain’t that cheap (when you’re a noob and don’t know what to buy), especially when you need to do more than 8 PCRs at a time.

In theory, works in practice, and in practice, works well in theory. Even for experienced DIYbio folks, cloning isn’t just 1,2,3.

Also, PCRs themselves aren’t very reproducible. It’s not about the overlapping sequence.

Reply on news.ycombinator.com