Beyond the $1K Genome: DNA ‘Writing’ Comes Next | Hacker News Reader

Beyond the $1K Genome: DNA ‘Writing’ Comes Next

techcrunch.com

33 pointsjonbaer16 comments

> Illumina hit the $1,000 mark in January of 2014.

Everyone I know (admittedly small sample size) is still paying around $2000-3000 for 30x WGS. Does the $1000 genome actually exist, or is this just pure PR on Illumina's part? What is BGI charging nowadays?

If you own your own Illumina, you can arguably get everything done for ~$1200-1400 with current chemistry, which gets better and cheaper with each iteration. We've also tested Oxford Nanopore's minION, which has probably the best chances of truly cheap sequencing bit that comes at the expense of accuracy.

At 10¢/bp, that's <$100/gene for most averaged-sized proteins. That price range makes producing novel proteins competitively priced over the tedious effort of stitching up chimeras from preexisting templates. Scientifically there's a lot that can be done if you can order proteins from scratch rather than having to scrounge around and copy/paste from what you already have access to. However, that de novo synthesis is currently only economically accessible to very few scientists/companies.

This new cheap synthesis will effectively make economical the 'writing of new code' rather than the current state of 'copy and paste from github, with a couple of minor modifications', and will bring commensurate advances to the industry. This will drop a significant barrier to the progress of synthetic biology and its downstream uses.

Twist just raised their series C at $37M (~80M total) [1]. Emily (the author) is their CEO.

[1] http://boston.citybizlist.com/article/275318/fidelity-manage...

How does that math near the end work out?

If there are 3B bp in the human genome, then the write the whole thing for $1000 it should be $3.3e-7, which is a whole lot less than the quoted $0.02.

The comparison is rather confusing because this article is not about synthesizing genomes, but individual genes. The $0.02 seems to be just a somewhat arbitrary number at which DNA synthesis isn't a large cost factor anymore.

The genes that are typically ordered are a few thousand nucleotides long, synthesizing the whole genome would be an absolutely monumental effort and probably out of reach for now. The largest artificially synthesized genome was done by Craig Venter and that had only about a million base pairs.

Interesting, thanks!

Writing is only a small step, the real goal is virtualization. Also, with tools like crispr, writing is not as big of a problem, per se.

Though really, if parallelization is the method maybe a cool benchmark would be sequencing a whole genome and then reprinting it, all while you're out for lunch or coffee.

I don't think you know what a CRISPR is. It doesn't have anything to do with DNA synthesis.

Lol and the GP is a cofounder of Experiment (YC W13), "an online community and a crowdfunding platform for scientific research." You really have to wonder how much due diligence YC does on its biotech founders.

I actually ask that question myself all the time :)

Sorry, I meant that CRISPR is helping people design much faster. Hopefully it will open the door for more people to design and use it, but I don't think DNA synthesis is the real technical bottleneck for new ideas here. Current cloning techniques seem to be fine tools, but I'm unaware of immediate research questions blocked by printing huge amounts of DNA any faster than current methods.

Call me when there's a Docker for DNA applications. All those cells are expensive overhead.

I doubt we'll still be making calls at that point.

You can check this blog post I wrote a while ago: https://thewinnower.com/papers/72-heroku-for-science

Haha, that's really cool. I meant something a little more literal though - being able to run several "applications" within individual cells. In bacteria you can often do that with plasmids, though if you don't modulate their promoters they'll easily exhaust the resources of the cell... kind of like setting bad ulimits on shared hosting.

That said, I would like to see some remote heroku like science platform. It seems like anything that's just pure DNA is a bit easier to do that with. Dealing with proteins sounds pretty hard though, there's so much variation in what they do and their physical properties.

Reply on news.ycombinator.com