hauls enormous soapbox out, clambers on An embarrassment of question
riches!
Synthetic biology is, to sciatrix's question, definitely
optogenetics-y! I think of it as using biological mechanisms for
engineering ends.
That might be to produce chemicals we make anyway more
efficiently—Ginkgo Bioworks is a company that bioengineers yeast
to produce specific chemicals (they also have an insanely cool laboratory
space and do extremely nifty automation things).
There's also genetic circuits, wherein you use proteins and RNA to turn a
cell into a logical circuit of on/off switches. The circuits are
nominally-independent1 of each other, so the cell is a tiny
modular human-designed computer, but with all the super specific amazing
things evolution has given us. So like —if we figure out the system
that birds use on a cellular level to detect/navigate via magnetic
fields2, we could then lift those genes powered it and put them
into our cells/organism of choice, and have, I dunno, slime mold that
always pointed north. Or you can use them as detector: IF there's high
levels of lead, THEN produce green protein, or IF you sense [x toxin in the
gut], THEN do [something] to let us know.
The applications I'm MOST excited about though, and comparatively the least
leery of, is using biology to report on itself. Because the cost &
difficulty of synthesizing, sequencing, and modifying DNA has gone waaayyy
down (source), DNA can act as a non-invasive, long term, and organism-
or cell-comprehensive information storage mechanism. A favorite example of
this idea does come out of the Boyden lab, actually, and it's this
proposal for a "molecular ticker tape" for recording neural signalling.
Here's the problem: we want to record as much information from as many
neurons as possible about when they talk to each other. Ideally, we would
know when every neuron was firing, and how much, and what other neurons
they were synapsing to, and at the single cell level. Most methods rely on
jamming some glass needles in there and recording from a group of 100
neurons at a time (temporally accurate, but non-comprehensive and not
single cell) or fMRI-like things (temporally slow, also non-comprehensive
and not single cell i have such a chip on my shoulder about fMRIs
jeeez) or recording from a single cell in a dish. All really cool;
all obviously limited in time and throughput. BUT WHAT IF.
WHAT IF. What if instead, you had a mouse where in each neuron of that
mouse there was a very specific and engineered DNA polymerase. Mostly, this
polymerase just chugs along, synthesizing some known and inert sequence.
But then, the neuron gets activated and there's a massive ion flux where
the voltage inside the cell goes from -70 mV to +40 mV, and our special
polymerase's synthesis error rate is sensitive to ion changes (i.e. more
positive ions, more errors!). And what you end up with3 is: DNA
produced over the entirety of the cell's lifetime, that theoretically
records every time that cell has fired. And you have that in every single
one of the 71 million cells of the mouse, so you haven't been bottlenecked
by your sample, and it isn't limited by "wow putting a rig on a mouse is a
PAIN", and it isn't limited by even needing to pick a specific time: you
have the whole record from birth to death. It's. So Cool. SO COOL.
So00ooo00ooo cool.
And um. That's. Some information about synthetic biology! I am most
familiar with the Boston biotech scene (hi, neighbor!) and also this is of
course my particular lens, which is very much "relevant to my [former] lab
or people in it" rather than "a review paper of the field."
1. I mean, though, ARE THEY REALLY probably not because nothing
operates independently anywhoooo
2. Although, even saying this, I do think magnetic sensors are
really cool, but I quail at the idea of reducing the extremely complex
navigation system of a pigeon to "it's got magnetoparticles all balled up
on one side that's it." I basically don't like genetic circuits
conceptually because I think biology-as-is is...way better at making things
than we are
3. if you solve like 100000000 technical issues with implementation
shh we're talking concepts here!
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
timeasmymeasure
no subject
Date: 2019-02-07 05:56 pm (UTC)hauls enormous soapbox out, clambers on An embarrassment of question riches!
Synthetic biology is, to![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
sciatrix's question, definitely
optogenetics-y! I think of it as using biological mechanisms for
engineering ends.
That might be to produce chemicals we make anyway more
efficiently—Ginkgo Bioworks is a company that bioengineers yeast
to produce specific chemicals (they also have an insanely cool laboratory
space and do extremely nifty automation things).
There's also genetic circuits, wherein you use proteins and RNA to turn a cell into a logical circuit of on/off switches. The circuits are nominally-independent1 of each other, so the cell is a tiny modular human-designed computer, but with all the super specific amazing things evolution has given us. So like —if we figure out the system that birds use on a cellular level to detect/navigate via magnetic fields2, we could then lift those genes powered it and put them into our cells/organism of choice, and have, I dunno, slime mold that always pointed north. Or you can use them as detector: IF there's high levels of lead, THEN produce green protein, or IF you sense [x toxin in the gut], THEN do [something] to let us know.
The applications I'm MOST excited about though, and comparatively the least leery of, is using biology to report on itself. Because the cost & difficulty of synthesizing, sequencing, and modifying DNA has gone waaayyy down (source), DNA can act as a non-invasive, long term, and organism- or cell-comprehensive information storage mechanism. A favorite example of this idea does come out of the Boyden lab, actually, and it's this proposal for a "molecular ticker tape" for recording neural signalling. Here's the problem: we want to record as much information from as many neurons as possible about when they talk to each other. Ideally, we would know when every neuron was firing, and how much, and what other neurons they were synapsing to, and at the single cell level. Most methods rely on jamming some glass needles in there and recording from a group of 100 neurons at a time (temporally accurate, but non-comprehensive and not single cell) or fMRI-like things (temporally slow, also non-comprehensive and not single cell i have such a chip on my shoulder about fMRIs jeeez) or recording from a single cell in a dish. All really cool; all obviously limited in time and throughput. BUT WHAT IF.
WHAT IF. What if instead, you had a mouse where in each neuron of that mouse there was a very specific and engineered DNA polymerase. Mostly, this polymerase just chugs along, synthesizing some known and inert sequence. But then, the neuron gets activated and there's a massive ion flux where the voltage inside the cell goes from -70 mV to +40 mV, and our special polymerase's synthesis error rate is sensitive to ion changes (i.e. more positive ions, more errors!). And what you end up with3 is: DNA produced over the entirety of the cell's lifetime, that theoretically records every time that cell has fired. And you have that in every single one of the 71 million cells of the mouse, so you haven't been bottlenecked by your sample, and it isn't limited by "wow putting a rig on a mouse is a PAIN", and it isn't limited by even needing to pick a specific time: you have the whole record from birth to death. It's. So Cool. SO COOL. So00ooo00ooo cool.
And um. That's. Some information about synthetic biology! I am most familiar with the Boston biotech scene (hi, neighbor!) and also this is of course my particular lens, which is very much "relevant to my [former] lab or people in it" rather than "a review paper of the field."
1. I mean, though, ARE THEY REALLY probably not because nothing operates independently anywhoooo
2. Although, even saying this, I do think magnetic sensors are really cool, but I quail at the idea of reducing the extremely complex navigation system of a pigeon to "it's got magnetoparticles all balled up on one side that's it." I basically don't like genetic circuits conceptually because I think biology-as-is is...way better at making things than we are
3. if you solve like 100000000 technical issues with implementation shh we're talking concepts here!