Getting Started with CRISPR Experiments – Part 1 | Hacker News Reader

Getting Started with CRISPR Experiments – Part 1

genomics.quiltdata.com

58 pointsakarve9 comments

Hi HN. I'm one of the founders of Quilt Data. We've started a blog series on how to run CRISPR experiments. Let us know what you'd like to see in future parts of the series.

What is the iteration latency today? Eg I do this test described, learn something (maybe something as small as what I tried didn't work) and now I want to try something else.

It looks like each run requires a new order with suppliers. They may take time to produce and send. So how long from "I want to try this other approach requiring nee supplies" to "new approach is laid down for testing"? And when something gets laid down for testing, how long might a test take?

(FYI I'm personally very excited about these developments and trying to grok the real world experience a bit more. Haven't done a day in a wet lab since high school! )

Great question. In almost all cases, you have an actionable lead after doing a screen experiment. The problem is typically that you have learned too many, not too few things. And those things need to be backed with more experimental evidence.

But more generally, the iteration latency is about a week to order & receive reagents for another targeted screen. Or if you want to do a single gene CRISPR experiment, you can get the single oligo reagents the next day.

Then the cloning, virus production, cell infection, etc. will be about another week if you are rushing. We in genomics are still a far cry from the iteration latency in software dev, but given that many of our scientific forefathers/mothers in the previous generation counted on making a whole genetically engineered mouse (which could take a year+) for an experiment, we're making solid progress.

More questions about the real world lab experience are always welcome!

love it thank you so much. A back of the envelope calculations suggest 8 - 14 days per cycle so 25-50 cycles per year. Assuming no vacation. I can see how we just face an incredible amount of work to do. Parallelism of course will help. But it must be so amazing to be like a blind man wandering around in a dark room with a sense that there might be something warm in the distance but that's about it. At least you're doing real work.

What would be more powerful instead of "comment out" debugging is if you could have println debugging.

Something like this is possible with CRISPR by tagging endogenous genes with fluorescent proteins. In this way, the output of that particular genetic line of code can be traced. It's not a perfect metaphor, and the method doesn't scale genome-wide, but its a clever thought.

Also, iirc, there are several studies under way to permit in vivo activation/deactivation of specific genes. CRISPR the switch in place and subject it to the stimulus (magnetic, light, sound). Changing the phenotype on the fly.

println is in the spirit of transcriptomics

there is also CRISPRa (one of two non-cutting CRISPRs), which can be used to activate and enhance output (incl. transcription) from a select region of the genome = a form of println.

Reply on news.ycombinator.com