Working group roadmap: Technology and infrastructure development

Jeffery A. Schloss

Let's try to get started. We have until 2 pm. That should be plenty of time to solve all the problems.

Jeff Schloss. No longer working at NIH since December.

What to get out of the working group?

What are the current bottolenecks and what would make the biggest difference if we could change those if we could improve? What are the key bottlenecks? Many different ways to achieve some things, sometimes only a single critical path. Which technologies need improvement?

I am coming from a model where at NIH we had some money to decide to devote to solving that set of problems. That's not the case here for GP-write; we don't yet have the money, but what could we do with that?

Up to sequencer verification? Up to delivery? What about other technology if you had working DNA synthesis? Are we missing importnat scope and discussion?

"So what"- why make DNA? Many demonstration projects will happen.

Working group-- should we pick demonstrations and say what are the key barriers? Get some test beds and gives a focal point into the trnasformation issue, depending on organism or application.

$1/genome target goal

Would be nice if synthesis would work entirely no matter what sequence is requested.

Data storage -- oligo pool method, not necessarily base-by-base. But that is due to current technological limitations in DNA synthesis.

Fairly clear endpoint-- synthesize 3B bp of human genome. We can do that with current tech. We need next-gen tech for design and synthesis to get over the line in a reasonable timeline at a reasonable price point.

Is the goal to be able to synthesize a human genome? Being able to write any large genome is interesting.

Human genome project singular focus. Let's focus on the obvious thing, the original vision of this, writing the human genome. There will be stepping stones, model systems.

yeast is 12M bp, factor of 4 difference.. We could bruteforce.

HGP spent $3B on sequencing. Should we do it bruteforce for writing? Is that okay? Or should we focus on next generation technology?

"Human" just gets people excited. It's the size that matters. "Moonshot".

Iteration time is also important. Short iterative time. Turnaround time of 10-20 kb time. Have to move away from chemistry, to enzymatic approaches.

How do write a human genome, or any genome at that scale, or bgiger things? How do we do this fast, accurate and cheap?

Only craig venter has made a synthetic bacterial genome from scratch. That was 7 years ago.

Centralized repoistory for the intermediate pieces... Reagents, DNA molecules, constructs. Organized repository with tracking, metadata, physical repository. Conditionings for joining consortium, includes sharing xyz materials? Has that been decided?

yeast to store megabase chromosomes, and then split up human chromosomes, 48 yeast clones..

What is a critical step that needs at least an order of magnitude improvement in one of its aspects?

  • 1000x cost reduction in DNA synthesis technology (machine development)
  • 1000x reduction in genome synthesis cost
  • 10x longer oligos that are error-free
  • 1000mers
  • DNA assembly is a huge problem
  • assembly
  • error-free large gene synthesis, more than 10kb
  • representation of standards - for database storage
  • multiplexing and parallelization of synthesis
  • enzymatic synthesis
  • cell delivery mechanisms, host can deal with up to 10 Mb right now, cell fusions can do this but where do you put it when you synthesize it? miniprep a chromosome nope. 500kb electroporation
  • large construct amplification technology
  • host organism or host cell, is it human, do you have to deconstruct a human cell or come up with a platform
  • break the genome down into smaller chunks, swap in to build the yeast chromsome, etc...
  • what is the minimum human genome
  • scifi book character that was intron-free, Altered Carbon
  • microfluidics
  • foundry on a chip
  • large dna transfer into cells
  • universal chromsome that can be added into any cell simply, like C24 technology, like on-off switch, get rid of it when you don't need it; or whole library that you could activate
  • better phosphoramidite chemistry
  • better enzymatic DNA synthesis
  • better inkjet arrayers
  • single molecule 10 kb synthesis, get around repetitive sequence problems
  • single molecule techniques

Next steps?

  • DNA synthesis survey whitepaper/review
  • call for proposals to meet certain goals
  • call for pilot projects meeting certain DNA synthesis goals
  • survey the community for speculative DNA synthesis technology for long-term development projects
  • GP-write machine shop to develop custom DNA synthesis technology?

I am going to report back on the technology and infrastructure group. More than 20 people joined the discussion. I think everyone volunteered to participate in an ongoing basis.

For those of you in the room, feel free to amplify.

We started out by talking about what are the goals? What are the goals of this project? I think there was agreement that GP-write needs to agree on a grand challenge. It needs to not only direct the progress but to be able to state when it has succeeded. The human genome project succeeded several times.

There was reasonable agreement that this should be to synthesize a human genome that is maintained in a host cell, which seemed to be an important part of that. To achieve this in 10 years, and there was some discussion about whether this should be a copy or whether a new human genome. There seemed to be agreement that copying is okay. Experimental genomes might be more dififcult.

There was some unresolved discussion on whether we need to have in addition to that singular goal some testbed objectives to answer the "so what" question, so what if you can copy a human genome? So if we do this, we need to ensure that those examples don't limit the technology directions and obviate the development of technologies.

It's probably true that one could bruteforce copying a human genome using existing technologies and get that done in 10 years, but this would not be very desirable. The project would not be much of a success, in that case. We need a bigger goal than that. Maybe the goal would be to bruteforce it today for X cost, then our goal would be to be albe to do it for 1/y cost or some small fraction or large orders of magnitude.

We need robust sharing of data and materials, we need technology to support the archiving of large constructs. We need standards for representation of the construct to make that sharing most effective.

That's something of a wishlist- we didn't get to prioritization- we didn't have a complete wishlist.

The $1/genome synthesis. It wasn't meant as a joke. Synthesizing any sequence including the ones that today are difficult like with repeats. Significant enthusiasm for short turn around time. Someone said the compile time is even weeks for small constructs. Computational tools to make it more flexible and faster to design and synthesize genomes, with like the paradigm of the 3d printer. We need at least 10x longer error-free oligos and it would be great to have 1000mers that were routinely error free. We should shoot for 10 kb or larger. There were many votes for DNA assembly improvements, especially for fragments above 3k bp. People thought it was important to develop multiplexed parallelized synthesis. Large construct packaging and transfer of large DNA. Automation of DNA assembly. Getting DNA into cells, getting and maintaining some of this. There is some redundancy here so I am trying to make sure it gets through here. 10 Mbp pieces into cells. Certainly want to develop in the interim ... capacities.

There was agreement that during the course of the project we need to understand which if any repeats are going to be needed, or if we should define a minimum human genome that differs. Universal chromosomes that can be turned on and off at will.

Another dream was DNA synthesis on a chip and DNA foundry on a chip with microfabrication and microfluidics.

That is what I was able to capture from the many ideas bouncing around the idea.

I will send this summary around for those of you who were in the room so that you can annotate and add to the summary.

We will put all the comments on the website.

Thank you.


Q: I am confused. I thought you said you wanted us to make wild type, whoever that is. But then you said that wouldn't count.... a funding agency was not happy with wild type in Sc2, they said do something new if you're going ot synthesize it.

A: If you demand that you do a reduced genome, then that requires huge amount more work. Maybe that's the right goal. But even today we can't come close to synthesizing a human genome. And then you get into this discussion- if we bruteforce it, we could probably do it in 10 years. But let's not only bruteforce it with today's methods and costs, but say we're going to synthesize some normal human genome but at considerably reduced costs from today's costs, or even actually reduced from following the current projections of cost reductions. If you take the current cost projections, we want to reduce 10, 100x, 1000x the cost. For those 20 people, they had different ideas about--...

Q: Nuance to that... from the discussion. I don't think we were trying to set the goal for GP-write. We were just wanted to set for ourselves as a working group some idea of a focus so that we could then think about what technologies do we need for that focus. We weren't trying to make a decision about the goal for GP-write should be to do a wildtype human or a non-wildtype human.