4 ms·

Biology's Master Programmers

37 points by llambda 14y ago

grhino 14y ago
"We use genomics and informatics to find a gene cluster. But that's an information unit," Verdine says. "We have to get the molecule. Synthetic biology involves coaxing the cluster into biosynthetic factories, which then produce the molecules. If we don't have the molecule, the cluster is useless." Once you have a scalable process for producing chemicals that are potential drugs via DNA sequencing & analysis and bacterial synthesis, how do you figure out what disease process to test those chemicals against? Do you shotgun them against a large set of disease processes? Purify the product, analyze it, and store it in a database and do theoretical work to pick out candidate diseases for the chemical? Is there something about the gene cluster that gives clues as to what the chemical might be doing for the organism?
- refurb 14y ago
  The knowledge of what the gene cluster does (by creating animals where that gene is unnaturally over or underexpressed) gives insight into what the genes product might do. Not that different than what is currently done in biopharmaceutical research.
pdelgallego 14y ago
I will love to be able to switch career path at some point in the (far) future, but I don't see how to start doing the move.
username3 14y ago
Researchers today may alter, copy, and paste DNA with increasing ease, but they still struggle when it comes to actually composing DNA that does anything useful. They are still editing nature's code and learning from it. It turns out that for now, nature is still the best programmer. Sounds like Intelligent Design.
- Lambdanaut 14y ago
  I tend to see the evolutionary process as a sort of pseudo intelligence, constantly striving to make the best organism for the environment it's in.
- saraid216 14y ago
  No, intelligent design is when they take a metaphor like the one above and presume it's descriptive. Saying that your sky blue eyes are actually pieces of the sky, for instance.
j_s 14y ago
> Researchers today may alter, copy, and paste DNA with increasing ease, but they still struggle when it comes to actually composing DNA that does anything useful. Leo Laporte did a TWiT Special on this recently: http://twit.tv/show/twit-live-specials/126 http://twit.tv/show/twit-live-specials/126 The discussion centered around taking existing genomes and splicing in functionality. The example given was setting up a Kickstarter project for trees that glow in the dark. The government is aggressively pursuing this tech. [Edit] Direct links: dna editor: http://www.genomecompiler.com/ http://www.genomecompiler.com/ dna printer: http://angel.co/cambrian-genomics http://angel.co/cambrian-genomics
- corwinbad 14y ago
  Hi j_s, this is Omri from Genome Compiler - thanks for the mention! Anyone can download our free gene/genome design software and start ordering DNA today at genomecompiler.com Omri
  - lupatus 14y ago
    I checked-out your company. It seems really cool. And, I think that it is really awesome that you are commercializing this tech and advocating it so much on HN (seriously). But, the part I am having trouble understanding, as an amateur, is what I do with the DNA once I order it. With software, I double-click an executable or go to a web address, and BAM! I can share pictures of kittens with my friends. How do I turn that DNA into a bio-luminescent kitten, for example, that I can show off to my friends?
    - corwinbad 14y ago
      Hi Lupatus, thanks! Well, today you have to know some biology and have access to lab (or community lab like bio-curious or genspace) to "execute" the biology. We are trying to abstract biology to make it more accessible to people outside of the research labs then you can order the DNA and transform living things in your kitchen :-)
Xcelerate 14y ago
As a chemical engineer who loves programming just as much, I have had some thoughts about bioengineering for a while now. Part of the problem is that the immense complexity involved in living organisms is unpredictable, even with today's computing technology. Human-engineered circuits are easy to predict; they were designed in a way that made the most sense. With biology however, something that appears to be working may have a completely unrelated side effect that isn't noticed until decades later, or even something so bizarre that it isn't noticed until hundreds of years later in future generations. The way we currently find treatments (grossly oversimplifying) is to grab some substance that looks like it might work, test it in animals, test it in humans, wait a while to make sure nothing really bad happened, and then prescribe it as a drug. If computers get advanced enough, I would much prefer the case where all of these interactions are fully simulated. You can go through millions of variations of substances all while tracking every change they make to each part of the virtual organism. Obviously, this is a VERY long way off, but it would be nice because we wouldn't have to worry about unknown side effects. Of course, our current way is the best we've got, and it saves millions of lives each year, so while there may be some future risk, we're doing the best we can. *On a somewhat related note, I've always thought it would be cool to have programmable blood filters. In other words, you could essentially provide the type of molecule or cell you are looking to trap, and then something like a ccd would either redirect it or let it pass through.
- Daniel_Newby 14y ago
  > ... I've always thought it would be cool to have programmable blood filters. It is called flow cytometry. http://en.wikipedia.org/wiki/Flow_cytometry http://en.wikipedia.org/wiki/Flow_cytometry
  - Xcelerate 14y ago
    Wow, that's pretty cool! I didn't realize people were already working on this.

37 points by llambda 14y ago

"We use genomics and informatics to find a gene cluster. But that's an information unit," Verdine says. "We have to get the molecule. Synthetic biology involves coaxing the cluster into biosynthetic factories, which then produce the molecules. If we don't have the molecule, the cluster is useless." Once you have a scalable process for producing chemicals that are potential drugs via DNA sequencing & analysis and bacterial synthesis, how do you figure out what disease process to test those chemicals against? Do you shotgun them against a large set of disease processes? Purify the product, analyze it, and store it in a database and do theoretical work to pick out candidate diseases for the chemical? Is there something about the gene cluster that gives clues as to what the chemical might be doing for the organism?

refurb 14y ago
The knowledge of what the gene cluster does (by creating animals where that gene is unnaturally over or underexpressed) gives insight into what the genes product might do. Not that different than what is currently done in biopharmaceutical research.

pdelgallego 14y ago

I will love to be able to switch career path at some point in the (far) future, but I don't see how to start doing the move.

username3 14y ago

Researchers today may alter, copy, and paste DNA with increasing ease, but they still struggle when it comes to actually composing DNA that does anything useful. They are still editing nature's code and learning from it. It turns out that for now, nature is still the best programmer. Sounds like Intelligent Design.

Lambdanaut 14y ago
I tend to see the evolutionary process as a sort of pseudo intelligence, constantly striving to make the best organism for the environment it's in.
saraid216 14y ago
No, intelligent design is when they take a metaphor like the one above and presume it's descriptive. Saying that your sky blue eyes are actually pieces of the sky, for instance.

j_s 14y ago

> Researchers today may alter, copy, and paste DNA with increasing ease, but they still struggle when it comes to actually composing DNA that does anything useful. Leo Laporte did a TWiT Special on this recently: http://twit.tv/show/twit-live-specials/126 http://twit.tv/show/twit-live-specials/126 The discussion centered around taking existing genomes and splicing in functionality. The example given was setting up a Kickstarter project for trees that glow in the dark. The government is aggressively pursuing this tech. [Edit] Direct links: dna editor: http://www.genomecompiler.com/ http://www.genomecompiler.com/ dna printer: http://angel.co/cambrian-genomics http://angel.co/cambrian-genomics

corwinbad 14y ago
Hi j_s, this is Omri from Genome Compiler - thanks for the mention! Anyone can download our free gene/genome design software and start ordering DNA today at genomecompiler.com Omri
- lupatus 14y ago
  I checked-out your company. It seems really cool. And, I think that it is really awesome that you are commercializing this tech and advocating it so much on HN (seriously). But, the part I am having trouble understanding, as an amateur, is what I do with the DNA once I order it. With software, I double-click an executable or go to a web address, and BAM! I can share pictures of kittens with my friends. How do I turn that DNA into a bio-luminescent kitten, for example, that I can show off to my friends?
  - corwinbad 14y ago
    Hi Lupatus, thanks! Well, today you have to know some biology and have access to lab (or community lab like bio-curious or genspace) to "execute" the biology. We are trying to abstract biology to make it more accessible to people outside of the research labs then you can order the DNA and transform living things in your kitchen :-)

Xcelerate 14y ago

As a chemical engineer who loves programming just as much, I have had some thoughts about bioengineering for a while now. Part of the problem is that the immense complexity involved in living organisms is unpredictable, even with today's computing technology. Human-engineered circuits are easy to predict; they were designed in a way that made the most sense. With biology however, something that appears to be working may have a completely unrelated side effect that isn't noticed until decades later, or even something so bizarre that it isn't noticed until hundreds of years later in future generations. The way we currently find treatments (grossly oversimplifying) is to grab some substance that looks like it might work, test it in animals, test it in humans, wait a while to make sure nothing really bad happened, and then prescribe it as a drug. If computers get advanced enough, I would much prefer the case where all of these interactions are fully simulated. You can go through millions of variations of substances all while tracking every change they make to each part of the virtual organism. Obviously, this is a VERY long way off, but it would be nice because we wouldn't have to worry about unknown side effects. Of course, our current way is the best we've got, and it saves millions of lives each year, so while there may be some future risk, we're doing the best we can. *On a somewhat related note, I've always thought it would be cool to have programmable blood filters. In other words, you could essentially provide the type of molecule or cell you are looking to trap, and then something like a ccd would either redirect it or let it pass through.

Daniel_Newby 14y ago
> ... I've always thought it would be cool to have programmable blood filters. It is called flow cytometry. http://en.wikipedia.org/wiki/Flow_cytometry http://en.wikipedia.org/wiki/Flow_cytometry
- Xcelerate 14y ago
  Wow, that's pretty cool! I didn't realize people were already working on this.