Justin here, cofounder of Serotiny. We've built a web-app to make the design and organization of synthetic genetic constructs efficient, cheap and effective. We've built an abstraction layer to enable scientists to build novel genetic designs from functional units without worrying about the actual underlying DNA sequence or how the sequence gets manufactured. Once designed, we help you place an order for DNA from a synthesizer of your choice.
The app is free to use - go ahead and register. We charge 15% of the manufacturing cost once an order is placed. Design single protein constructs, or high-throughput combinatorial sets of proteins or mutation sets. [1]
Our genetic management infrastructure makes it straightforward to see where particular designs came from and how they've been used. It straightforwardly keeps track of the functions and restrictions of each design. For groups/labs we have an API that can respond to queries relating various protein constructs by things like function, sequence, or usage. [2]
Check it out, I'd be curious your thoughts. I'm happy to answer any questions.
We built it all with Go and Ember - a huge thanks to those in the community working on those tools.
Yes. We have a curated database of protein functions - so we have some idea of the function of new designs. Any designs submitted for manufacture are further screened - both the customer and the design - by both us and the DNA synthesizer.
Interesting... I will need to figure that out. I've not seen that before.
The HTML for the page is just a formatted response from our API server formatted by Go's internal templating engine. The back button seems to request something differently somehow.
We have built the infrastructure to design out entire plasmids - within Pinecone you can build 'non-coding Cassettes' that include all sorts of expression components. You can build designs that include all the standard DNA expression components - promoters, terminators, recombinase sites, etc. as well as incorporate the proteins to be expressed. We have not focused as much on building out variation in the design of the expression systems, but it is something we are actively working on making nicer.
In general, at least so far, our customers have had their own plasmids they're used to working with, or the generic expression systems offered by the DNA synthesizers have been sufficient.
We have a lot of friends building genetic design software. Each have slightly different focuses. Ours is primarily, and I think uniquely, focused on an abstraction above the DNA itself (protein construct design). We focus on biological function of the output of the DNA rather than the 'assembly code' of DNA itself. For better or worse, we're a 'C IDE' rather than an assembly editor. If you are skilled enough to read the genetic matrix, some of the other software permits more direct manipulation of DNA. Pinecone is useful if you need to work or communicate your designs at a higher level of abstraction.
SnapGene is another widely-liked native app for working with plasmid DNA: http://www.snapgene.com/
Pinecone chooses to focus on guiding designs in order to produce genetic tools that have particular functions rather than focusing on how to construct DNA. We leave the construction details to the DNA synthesizers.
Pricing: design and individual use is free, we charge a percentage of the manufacturing cost for designs submitted through us. We also make money building custom infrastructure (of which Pinecone is an example) for companies needing to keep track and analyze their genetic designs at a functional level.
Speed: The idea is because we use an abstraction above the DNA, certain kinds of high-throughput designs becomes VERY fast with Pinecone. "I want all proteins made with these 5 things up front, these 7 linkers, these 4 fluorescent probes - and all 140 combinations" - would take about 2 minutes to design with Pinecone, but would be days of error-prone work if done manually.
Reliability: We can't guarantee novel designs will work - biology is hard. But we can help give novel designs the best chance of working. Pinecone showcases what has worked - and makes it easy to riff off of well-worn designs. And because you're buying de-novo synthesized DNA (not copy/pasting other's code), your sequences are exactly what you asked for, not just 'good enough'.
You still have to figure out what "intelligence" is, and how to transfer, manipulate, or otherwise encode for it. I don't think we have the tools to actually transplant intelligence yet. We are in the early stages of understanding it at all. The field of "optogenetics" https://en.wikipedia.org/wiki/Optogenetics has been a powerful genetic way to help start along that path to understanding how brains work, to what extent a mouse is intelligent, and how to affect neural processes.
What's the ELI5 for this in terms of who customers are and what kinds of problems it is solving? Sounds terribly interesting but I have absolutely no knowledge of this space :-/
Customers: Researchers and those they communicate with. Specifically those doing early development of novel proteins - antibodies, biologics, CARs, CRISPR, enzymes, bio-materials, bio-sensors, optogenetics, or basic research. We help them organize and intelligently manage their libraries of constructs based on the constructs' capabilities.
Problem: Communicating genetic designs to yourself, to others in your field, to others in your company - your boss or your technicians, to manufacturers and suppliers. And communicate without error, with higher-level abstraction, and with functional rather than technical detail. High-throughput design and analysis of designs naturally falls out of those capabilities.
DNA is the 'blueprint' for the "protein" machines. If you want to build a new or novel biological machine, you must construct a DNA blueprint for it, that blueprint is ingested, and the machine is built to spec. Our software is essentially a 1-dimensional CAD program that lets you focus on, manipulate and organize the material properties of the biological nanomachines you are building, rather than focus on the manufacturing process. Think the difference between producing a high-level CAD file vs G-code for the design of a sub-10nm 3d object.
Historically, you have to build the blueprint by hand. The challenges of building the DNA blueprint itself were immense, and have slowly become more and more routine. Simply obtaining a close-enough blueprint to what you wanted was sufficient to develop synthetic insulin, synthetic HGH and a host of other billion-dollar biologic therapies you see on TV commercials every night. This tool is a break-point - it allows you to build biological machines based on what you want the machine to do, and leave the construction of the blue-print itself entirely behind the scenes. It compiles down the high-level design into a synthesizable blueprint without the user needing to intervene. Construction of DNA is fraught with all sorts of syntax rules that this helps to entirely obviate. With this software a researcher can focus on the properties of their desired product 'fluoresces green', 'binds to Gold', 'more soluble' rather than nuanced genetic construction rules.
Many useful protein machines can be deconstructed into component parts (each part itself encoded by DNA). Pinecone lets you drag and drop those component parts together, press buy, and get shipped the DNA that encodes those parts. Historically you'd have to parse a string of thousands of A, T, G and Cs (literally in Excel or Word) - where a single error would result in failure of the machine.
These proteins are useful therapeutically, economically, and socially - they are biology's nanotechnology. They are a few orders of magnitude more precise than Intel's new i9 processor's features, are 3D in nature, and work in wet, room-temperature environments.
DNA needs to be compiled into a protein in order to 'do' anything. DNA is the source code, proteins are the molecular machines built by the code. And every organism uses a similar compiler. So the DNA has to be put inside an organism before the DNA source code can be 'compiled' into a biological machine (a protein). Interestingly at the level of the compiler, almost every organism on Earth is capable of compiling most others' particular DNA into a protein (with a lot of exceptions).
Most purchased DNA that encodes a protein comes in the form of a bacterial 'virus' called a plasmid that can very easily be given to e coli - and it makes billions of copies of that DNA with very high fidelity in a few hours (https://en.wikipedia.org/wiki/Plasmid). This DNA can then be purified from that e coli in physically appreciable amounts and then be put into other organisms for ultimate usage. If you're purifying a chemical or a therapeutic the DNA is often put into yeast or e coli. If you're doing research, there are a number of 'model organisms' the DNA can be put into to supplement the genes already in the organism you're studying - including human cancer cells.
There are certain kinds of 'gene therapies' where the DNA is actually put into living human cells, often that have been harvested, and then put back into the person. This enables the genetic code for the new tools/proteins to be incorporated as a therapy.
The physical insertion of DNA into an organism is called "Transfection"
https://en.wikipedia.org/wiki/Transfection (or transduction, or transformation for various particulars).
Wonderful, thanks so much for the explanation. I think I knew that you had to get the DNA into an organism, but I had no idea how that could be done. The fact that the purchased DNA comes in the form of a virus that's ready to make lots more of that DNA is amazing.
I don't think so. But we are certainly trying to lower the barrier to entry for building useful biological tools. We hope to help 'smaller than billion-dollar-blockbuster drugs' be built by companies and researchers who are inventing all sorts of socially useful biological technologies. We are working with scientists to enable novel uses of biotechnologies for things like genetically targeted immunotherapies, synthetic biosensors, spider silk clothing, vegan gelatins meats and milks, oil-free biofuel production, and of course just much more rapid biological research.
Looks cool! Can you link to any experimental data/publications that show the functionality of enzymes that were designed this way?
It would be very interesting to get a feel for the success rate of such an approach.
You mention several research groups that use your tool, so I assume there is some in vitro/vivo data available.
As it's a 'ShowHN' - we've only had the software public for a short time. Biology takes a while, so many of our clients who have used the software have chosen to not yet make their designs public - they're still working with them.
What we can do is reverse engineer already published data into our system to see how it would work. If you're logged in you can see a synthetic Knoevenagel Catalyst like KN.1: https://serotiny.bio/pinecone/part/11291
And from that page you can see the mutations that were made to create that new enzyme, as well as trace the history of the synthetic design from its current sequence all the way back to its wild-type ancestor through years of research.
Our background is not in computational design of proteins from their atomic structure (like Rosetta). We enable someone who does have that expertise - who has such a design in mind, to actually go about producing their libraries, getting the material delivered, evaluating the effectiveness of what they've produced, and sharing that information with their colleagues in a straightforward and actionable way. And if they're picking variants by a screen or directed evolution, Pinecone would be useful in describing the results of the screen in order to either move forward, or put the results to work.
Similarly, some of the fluorophores have great 'histories' to them - and with our software you can see how various fluorophores were designed, where they came from, and how they've been used. See Dronpa: https://serotiny.bio/pinecone/part/9036 or some of the pH-sensitive fluorophores like ArcLight: https://serotiny.bio/pinecone/part/10533
We'd like to think this enables a more straightforward 'porting' of existing designs into new scaffolds - if the mutations to GFP made it pH-sensitive, similar mutations to YFP will likely make it pH-sensitive. Swap the fluorophore entirely, or pull in a natural variant of Cas9 and likely the same sites that produced a nickase from spCas9 will work on other cas9 constructs.
This sounds ripe for DEA and FDA regulation. If you cannot account for possibly dangerous synthesys or mutations, and im not reading that you fully can, then you have to ensure the end users can. Are you screening customers? Can anyone order? Is it possible for the synthesis to alter the plasmid, or am I not understanding the DNA packaging mechanism?
28 comments
[ 3.3 ms ] story [ 59.8 ms ] threadCheck it out, I'd be curious your thoughts. I'm happy to answer any questions.
We built it all with Go and Ember - a huge thanks to those in the community working on those tools.
[1] https://serotiny.bio/notes/support/tutorials/ [2] https://serotiny.bio/notes/applications/
And a few write-ups/dissections of proteins of interest to HN: https://serotiny.bio/notes/proteins/
Do you have any biosecurity-related protection mechanisms in place?
That said, you'll probably want to fix the spelling for "vendors" on the main page (it's misspelled 'venders').
( e.g. go to https://serotiny.bio/api/parts/11236, click 'edit in Pinecone', press cmd + [ ).
The HTML for the page is just a formatted response from our API server formatted by Go's internal templating engine. The back button seems to request something differently somehow.
Thanks for catching that. More work :)
In general, at least so far, our customers have had their own plasmids they're used to working with, or the generic expression systems offered by the DNA synthesizers have been sufficient.
If you want cheap and dirty way to manipulate DNA, APE is great: http://biologylabs.utah.edu/jorgensen/wayned/ape/
And YC's own Benchling has fantastic web-based software for designing, sharing and keeping track of plasmids: https://benchling.com/
Genome Compiler is impressive as well: http://www.genomecompiler.com/
SnapGene is another widely-liked native app for working with plasmid DNA: http://www.snapgene.com/
Pinecone chooses to focus on guiding designs in order to produce genetic tools that have particular functions rather than focusing on how to construct DNA. We leave the construction details to the DNA synthesizers.
Pricing: design and individual use is free, we charge a percentage of the manufacturing cost for designs submitted through us. We also make money building custom infrastructure (of which Pinecone is an example) for companies needing to keep track and analyze their genetic designs at a functional level.
Speed: The idea is because we use an abstraction above the DNA, certain kinds of high-throughput designs becomes VERY fast with Pinecone. "I want all proteins made with these 5 things up front, these 7 linkers, these 4 fluorescent probes - and all 140 combinations" - would take about 2 minutes to design with Pinecone, but would be days of error-prone work if done manually.
Reliability: We can't guarantee novel designs will work - biology is hard. But we can help give novel designs the best chance of working. Pinecone showcases what has worked - and makes it easy to riff off of well-worn designs. And because you're buying de-novo synthesized DNA (not copy/pasting other's code), your sequences are exactly what you asked for, not just 'good enough'.
[0]: https://serotiny.bio/notes/applications/car/
Problem: Communicating genetic designs to yourself, to others in your field, to others in your company - your boss or your technicians, to manufacturers and suppliers. And communicate without error, with higher-level abstraction, and with functional rather than technical detail. High-throughput design and analysis of designs naturally falls out of those capabilities.
DNA is the 'blueprint' for the "protein" machines. If you want to build a new or novel biological machine, you must construct a DNA blueprint for it, that blueprint is ingested, and the machine is built to spec. Our software is essentially a 1-dimensional CAD program that lets you focus on, manipulate and organize the material properties of the biological nanomachines you are building, rather than focus on the manufacturing process. Think the difference between producing a high-level CAD file vs G-code for the design of a sub-10nm 3d object.
Historically, you have to build the blueprint by hand. The challenges of building the DNA blueprint itself were immense, and have slowly become more and more routine. Simply obtaining a close-enough blueprint to what you wanted was sufficient to develop synthetic insulin, synthetic HGH and a host of other billion-dollar biologic therapies you see on TV commercials every night. This tool is a break-point - it allows you to build biological machines based on what you want the machine to do, and leave the construction of the blue-print itself entirely behind the scenes. It compiles down the high-level design into a synthesizable blueprint without the user needing to intervene. Construction of DNA is fraught with all sorts of syntax rules that this helps to entirely obviate. With this software a researcher can focus on the properties of their desired product 'fluoresces green', 'binds to Gold', 'more soluble' rather than nuanced genetic construction rules.
Many useful protein machines can be deconstructed into component parts (each part itself encoded by DNA). Pinecone lets you drag and drop those component parts together, press buy, and get shipped the DNA that encodes those parts. Historically you'd have to parse a string of thousands of A, T, G and Cs (literally in Excel or Word) - where a single error would result in failure of the machine.
These proteins are useful therapeutically, economically, and socially - they are biology's nanotechnology. They are a few orders of magnitude more precise than Intel's new i9 processor's features, are 3D in nature, and work in wet, room-temperature environments.
Most purchased DNA that encodes a protein comes in the form of a bacterial 'virus' called a plasmid that can very easily be given to e coli - and it makes billions of copies of that DNA with very high fidelity in a few hours (https://en.wikipedia.org/wiki/Plasmid). This DNA can then be purified from that e coli in physically appreciable amounts and then be put into other organisms for ultimate usage. If you're purifying a chemical or a therapeutic the DNA is often put into yeast or e coli. If you're doing research, there are a number of 'model organisms' the DNA can be put into to supplement the genes already in the organism you're studying - including human cancer cells.
There are certain kinds of 'gene therapies' where the DNA is actually put into living human cells, often that have been harvested, and then put back into the person. This enables the genetic code for the new tools/proteins to be incorporated as a therapy.
The physical insertion of DNA into an organism is called "Transfection" https://en.wikipedia.org/wiki/Transfection (or transduction, or transformation for various particulars).
The general concept when applied to human health is called "Gene Therapy". https://en.wikipedia.org/wiki/Gene_therapy
The manipulation of DNA as a tool to understand the mechanisms of biology can be termed "Molecular Biology". https://en.wikipedia.org/wiki/Molecular_biology
-Justin
Some of the uses for these proteins we've talked about already here on HN: https://serotiny.bio/notes/proteins/
What we can do is reverse engineer already published data into our system to see how it would work. If you're logged in you can see a synthetic Knoevenagel Catalyst like KN.1: https://serotiny.bio/pinecone/part/11291
or the Retroaldolase RA95.5-8: https://serotiny.bio/pinecone/part/11290
And from that page you can see the mutations that were made to create that new enzyme, as well as trace the history of the synthetic design from its current sequence all the way back to its wild-type ancestor through years of research.
Our background is not in computational design of proteins from their atomic structure (like Rosetta). We enable someone who does have that expertise - who has such a design in mind, to actually go about producing their libraries, getting the material delivered, evaluating the effectiveness of what they've produced, and sharing that information with their colleagues in a straightforward and actionable way. And if they're picking variants by a screen or directed evolution, Pinecone would be useful in describing the results of the screen in order to either move forward, or put the results to work.
Similarly, some of the fluorophores have great 'histories' to them - and with our software you can see how various fluorophores were designed, where they came from, and how they've been used. See Dronpa: https://serotiny.bio/pinecone/part/9036 or some of the pH-sensitive fluorophores like ArcLight: https://serotiny.bio/pinecone/part/10533
We'd like to think this enables a more straightforward 'porting' of existing designs into new scaffolds - if the mutations to GFP made it pH-sensitive, similar mutations to YFP will likely make it pH-sensitive. Swap the fluorophore entirely, or pull in a natural variant of Cas9 and likely the same sites that produced a nickase from spCas9 will work on other cas9 constructs.