Old concept list & things to avoid
Bryan Bishop
http://diyhpl.us/~bryan/papers2/polymerase/
1) electronic control of polymerase
2) nucleotide gun made out of a nanotube pointed at the finger domain
of some DNA polymerase
3) single-polymerase water droplet & add in a single dNTP at a time
4) physical display of dNTP as template for current base addition
(i.e., on a stick) (not a protein template)
5) a protein that can undergo conformational changes that polymerase
thinks represents the template strand
6) pull/push a template through DNA polymerase to control which dNTP
it should be selecting for
7) protein-template DNA polymerase, where the polymerase itself has a
giant protein that enzymatically encodes dNTP information (protein
template, like in CCA-adding enzymes)
8) mechanical pressure on polymerase
9) some magical ultrasound method
Things I'd like to avoid in the best possible solution (I'd settle for
sucky solutions though):
(1) Any step that involves dissociating from the growing molecule.
Takes too long to re-associate.
(2) Mistakes.
(3) Wash steps. Takes too much time.
(4) Parallelism. Just one polymerase/enzyme, thank you very much, no
"law of big numbers" stuff going on here.
(5) Pause/step mode. Including a step-inducing state change would be
fantastic, but I think we can do continuous incorporation if we have
to?
(6) if possible: off-enzyme nucleotide selection. The polymerase
should be controlled, not the nucleotides - there's too many of those
to reliably, quickly control.
(7) if possible: light. Complicates the setup, but isn't a huge deal in the end.
davidad (David A. Dalrymple)
hope for controlling a single molecule using electronics. Using a
nanowire to flow electric current or mechanical pressure directly into
an enzyme is surely *more* expensive and technically challenging than
optical techniques. It's possible that other electromagnetic fields
than waves (electrostatics, magnetic moments) could be helpful, but
I'd definitely be thinking Maxwell's Equations if I were you, not
solid-state physics. I also take issue with (2). If you want to beat
conventional synthesis, we're talking about 10^5 base pairs and up. Do
you really think you can get error probability below 10^-5? I think
it's probably better to accept that mistakes will be made, and
incorporate a strategy for error correction.
Cheers,
- davidad
Nathan McCorkle
substrate (silicon, helical silicon (that might not exist, but I've
heard of metallofullerenes being helical with a very similar
twist/rotation as DNA http://www.jenlaurltd.com/Products.html))
If you could electronically control a "template", then you might not
need anything special other than ddNTPs and polymerase
Now that I think of it though, polymerase probably needs/wants (in
wild-type enzymes) to "grip" the template strand... which a solid
plane of silicon/etc wouldn't allow.
I definitely think mistakes need to be suppressed to the point of
natural in vivo levels... any error handling needs to be automatic and
on-the-fly.
I think optical modification of the enzyme state is, like David said,
the best way to interface. We can easily buy lots of different colored
lasers and they work through vacuums, etc...
On Fri, Feb 24, 2012 at 8:35 PM, Bryan Bishop <kan...@gmail.com> wrote:
--
Nathan McCorkle
Rochester Institute of Technology
College of Science, Biotechnology/Bioinformatics
Dan Bolser
> Here's an old list of concepts from..
> http://diyhpl.us/~bryan/papers2/polymerase/
>
> 1) electronic control of polymerase
> 2) nucleotide gun made out of a nanotube pointed at the finger domain
> of some DNA polymerase
> 3) single-polymerase water droplet & add in a single dNTP at a time
> 4) physical display of dNTP as template for current base addition
> (i.e., on a stick) (not a protein template)
> 5) a protein that can undergo conformational changes that polymerase
> thinks represents the template strand
> 6) pull/push a template through DNA polymerase to control which dNTP
> it should be selecting for
> 7) protein-template DNA polymerase, where the polymerase itself has a
> giant protein that enzymatically encodes dNTP information (protein
> template, like in CCA-adding enzymes)
> 8) mechanical pressure on polymerase
> 9) some magical ultrasound method
Nice ideas!
> Things I'd like to avoid in the best possible solution (I'd settle for
> sucky solutions though):
>
> (1) Any step that involves dissociating from the growing molecule.
> Takes too long to re-associate.
>
> (2) Mistakes.
Yup.
> (3) Wash steps. Takes too much time.
>
> (4) Parallelism. Just one polymerase/enzyme, thank you very much, no
> "law of big numbers" stuff going on here.
Actually, I think there is no other way. Even running at 1000 bp/s, it
would take weeks to print a human genome. I think cycling (computer
controlled washing) and parallelism are viable techniques.
> (5) Pause/step mode. Including a step-inducing state change would be
> fantastic, but I think we can do continuous incorporation if we have
> to?
Wont this be highly inaccurate? Hmm... I think I don't understand this point.
> (6) if possible: off-enzyme nucleotide selection. The polymerase
> should be controlled, not the nucleotides - there's too many of those
> to reliably, quickly control.
Unless you allow cycling with something like chain termination.
> (7) if possible: light. Complicates the setup, but isn't a huge deal in the end.
No opinion.
> - Bryan
> http://heybryan.org/
> 1 512 203 0507
Cheers Bryan!
Dan.
Nathan McCorkle
> On 25 February 2012 01:35, Bryan Bishop <kan...@gmail.com> wrote:
>> Things I'd like to avoid in the best possible solution (I'd settle for
>> sucky solutions though):
>>
>> (1) Any step that involves dissociating from the growing molecule.
>> Takes too long to re-associate.
>>
>> (2) Mistakes.
>
> Yup.
>
>
>> (3) Wash steps. Takes too much time.
>>
>> (4) Parallelism. Just one polymerase/enzyme, thank you very much, no
>> "law of big numbers" stuff going on here.
>
> Actually, I think there is no other way. Even running at 1000 bp/s, it
> would take weeks to print a human genome. I think cycling (computer
> controlled washing) and parallelism are viable techniques.
Dan, I think what Bryan is saying is that per reaction, only one
polymerase/growing-strand.... rather than with cloning/transformation
where you have 1000s of molecules and 1000s of cells where the outcome
may be only 10 successful reactions
Parallel reactions are definitely useful, especially because the
reaction centers and overhead microfluidics would be small
>
>
>> (5) Pause/step mode. Including a step-inducing state change would be
>> fantastic, but I think we can do continuous incorporation if we have
>> to?
>
> Wont this be highly inaccurate? Hmm... I think I don't understand this point.
>
>
>> (6) if possible: off-enzyme nucleotide selection. The polymerase
>> should be controlled, not the nucleotides - there's too many of those
>> to reliably, quickly control.
>
> Unless you allow cycling with something like chain termination.
>
>
>> (7) if possible: light. Complicates the setup, but isn't a huge deal in the end.
>
> No opinion.
>
>
>> - Bryan
>> http://heybryan.org/
>> 1 512 203 0507
>
>
> Cheers Bryan!
> Dan.
--
Jonathan Cline
On Friday, February 24, 2012 5:35:44 PM UTC-8, Bryan Bishop wrote:
Things I'd like to avoid in the best possible solution (I'd settle for
sucky solutions though):
(2) Mistakes.
Why avoid mistakes? Perhaps you can error correct them later. This is a yield tradeoff. Lower yield is sometimes a better solution.
## Jonathan Cline
## jcl...@ieee.org
## Mobile: +1-805-617-0223
########################
Bryan Bishop
Why avoid mistakes? Perhaps you can error correct them later. This is a yield tradeoff. Lower yield is sometimes a better solution.