[BBF Standards] systemic limitation of biobricks for combinatorial logic?
Deepak Chandran
deepakc at u.washington.edu
Tue May 20 11:24:42 EDT 2008
Hello,
A question concerning PoPS devices: do they truly resolve the cross-talk
problem? Three inverters of the same kind would interfere with one
another, so what aspect of this particular issue does it solve? My
understanding was that PoPS is a way of establishing compatible
input/output relationships. If there was a PoPS device that does not use
any protein to achieve its function, then it would evade the crosstalk
problem; but is it possible to build a genetic network that has no
proteins (setting aside RNA-based transcription factors)?
It might also be the case that the solution to the problem (for the case
of a single-cell system) may not be achieved by the way we think in
digital electronics. Evolution has probably built its networks through
swapping of regulatory regions, gene duplications, and gene (domain)
fusions, which are probably followed by changes in binding affinities.
Although this strategy may seem limiting in some ways (such as making
millions of copies of the same device), it probably has certain
advantages that we need to exploit. Since synthetic protein networks are
in the air, it may be worthwhile thinking how tasks can be distributed
between protein networks and genetic networks to achieve a goal.
--Deepak
Dr. Markus Schmidt wrote:
> Thanks JC for bringing the issues back into the disussion. The system-
> wide visibility of all gate outputs is a serious obstacle to the
> development of the standardized biopart concept. When I posted the
> message on lacking specifity in February there was hardly any
> reaction, which really surprised me. Sure as long as the experimental
> phase of biobricks now runs systems that contain only few parts, all
> these problems do not arise, but thinking about the future and about
> the prospect of this approach we should dedicate a susbstantial amount
> of time and energy to come up with solutions.
>
> If we take electronic integrated circuits as an example (and not as a
> metapher), than producing compartiments is the goal.
>
> Which ways are there to produce compartiments?
>
> To begin with I would say there are spacial, chemical, sematic or time-
> based compartiments.
>
> 1. Spacial:
> 1.1. new organelles. This is a nice idea but how many organelles can
> you engineer into a cell? Tens, hundreds, but certainly not millions.
> 1.2. cell-cell communications. of course this is an option but it is
> basically the same situation as in the organelles, although with the
> option to increase the number of differnet cells without the packing
> problems of organelles. Basically the way by wich the cells
> communicate is the bottleneck.
> 2. Chemical:
> 2.1. Number of molecules to be used as an information carrier is
> extremely large but at the cost of reduced specifity and increasing
> cross talk.
> 2.2. Quorum sensing. This is done by molecules and only because it
> involves the extracellular environment it doesn't mean it solves the
> problem of open logic gates.
> 3. Semantic:
> 3.1. I guess the zinc finger story comes in here, an approach that
> target the genetic code. The information you can store on a lets say
> x bp long DNA is 4^x (may be reduced for some mutation-robustness
> meassures) and could provide enough specifity to deal with ultra large
> scale circuits. This is actually a promising approach, programming an
> RNA computer
> 4. Time-based
> 4.1. PoPS, for Polymerase Per Second. This is a nice idea (and the
> comic is fun) and transcends/converts the problem of chemical
> specifity to a unambigous signal unit. OK, lets say you found a way to
> meassure the PoPS rigt on the DNA. Say you meassuered x PoPS. Then you
> have a subsequent PoPS analyzer that makes e.g. the following
> decision: if x<y output=0; if y<x<z output undefined; if z<x output=1.
> Problem solved, isn't it? Well it is but only in the case if the PoPS
> analyzer (counter) sits right after the PoPS relevant piece of DNA,
> otherwise you would have to transfer the result x into a chemical
> signal to transport it to another part of DNA or elsewhere and then
> you run into the same problem of open logic gates. So if you avoid
> that and realize a linear logic line (a Ford like assembly line) you
> are quite limited in running your software.
>
> However, what I think can be done is to combine all these approaches
> in order to push the limit of the maximum number (Nmax) of realizable
> "logic gates" or operations a little bit. Maybe each approach can help
> us to push the Nmax by a factor of 100 or 1000 (or maybe more).
> However, this is way a different story than with Moore's law, where
> basically the reduction of size ( and packaging) of logic gates was
> and is the main driving force to improve the number of transistor per
> chip.
>
> Cheers, Markus
>
>
>
>
>
>
>
> Am 20.05.2008 um 05:35 schrieb Drew Endy:
>
>
>> As background information relevant to your discussion:
>>
>> 1. there is a common signal carrier for gene expression devices. it
>> is called PoPS, for Polymerase Per Second. read about it here: http://openwetware.org/wiki/Adventures
>>
>> 2. biochemical cross talk within a self mixing volume can be handled
>> via the specificity of molecular interactions. look up zinc finger
>> DNA binding proteins. read Reshma Shetty's dissertation from MIT
>> (this year). Also read Zarrinpar A, Park SH, Lim WA., PMID: 14668868
>>
>> 3. new organelles could be created thereby providing engineered
>> additional spatial insulation. see the 2007 UCSF iGEM team's second
>> project (here: http://parts.mit.edu/igem07/index.php/UCSF/Organelle_Intro)
>> . Also read Barry Canton's dissertation from MIT (this year).
>>
>> 4. cell-cell signaling can be used to communicate across bacteria.
>> read Weiss and Knight (http://www.princeton.edu/~rweiss/papers/rweiss-dna6.pdf
>> ) and more recent papers from Ron's lab at Princeton (see http://weisswebserver.ee.princeton.edu/pubs.html)
>> .
>>
>> Drew
>>
>>
>>
>> On May 19, 2008, at 10:42 PM, James Lawson wrote:
>>
>>
>>> Hi folks,
>>>
>>> This might be a little from left field, and I don't have the
>>> solution, but I have had this idea since I became familiar with the
>>> gene-centric abstraction level that the entire Biobricks / standard
>>> parts systems is based on. I don't think the level that we're
>>> designing these systems at is helping us here, particularly with
>>> respect to this system-wide visibility of gate outputs. We have no
>>> 'encapsulation' mechanism. The non-nucleated cells we're using as
>>> chassis for these systems don't have any mechanism for membrane-
>>> based compartmentalisation, so any all the 'computation' is done in
>>> pretty much the same compartment. Nucleated, compartmentalised cells
>>> are able to get a step further by creating an encapsulation
>>> mechanism, where some kind of reaction or transformation of species
>>> may be contained within a delineated zone.
>>>
>>> Leveraging any of these membrane compartmentalisation (or membrane
>>> mircodomain) based systems of encapsulation would require some
>>> pretty extensive engineering and standardisation of the transport
>>> and cytoskeletal systems within a cell, so I guess it isn't really
>>> tractable for a while.
>>>
>>> To add to Herbert's comments:
>>> In unicellular organisms, one could say that quorum sensing systems
>>> are analogous to neural networks in multicellular organisms as a
>>> solution to this problem of encapsulation and input/output
>>> visibility management we're talking about. I have heard quite a lot
>>> about people doing engineering with quorum sensing systems and I
>>> think this is probably where the synthetic biology community will
>>> continue to look to. I don't really know how much we know about
>>> heterogeneous bacterial colonies - perhaps someone could enlighten
>>> me on the state of the field?
>>>
>>> Kind regards,
>>> James Lawson
>>>
>>> Herbert Sauro wrote:
>>>
>>>> Funny you should mention this solution:
>>>>
>>>>
>>>>> Can this be solved by using single cells for each circuit, then
>>>>> somehow providing inter-cell communication?
>>>>> (Also seems quite difficult, correct?)
>>>>>
>>>> Because this is how evolution got round the problem by inventing
>>>> neural circuits.
>>>>
>>>> Herbert Sauro
>>>>
>>>> J C wrote:
>>>>
>>>>
>>>>> Hi all,
>>>>> I have been reading about biobricks for some time,
>>>>> I am a computer engineer.
>>>>>
>>>>> For combinatorial logic, it seems the limitation of biobricks
>>>>> is the system-wide visibility of all gate outputs, is this correct?
>>>>> In circuits with multiple gates, each gate must employ a unique
>>>>> input signal and output signal, otherwise the multiple outputs
>>>>> will conflict with the inputs.
>>>>>
>>>>> How is this being addressed other than by finding unique
>>>>> signals to feed to each gate?
>>>>>
>>>>> This is not a scalable solution for larger circuits, i.e. adders
>>>>> or latches (RAM storage for more than 1 bit).
>>>>>
>>>>> Can this be solved by using single cells for each circuit, then
>>>>> somehow providing inter-cell communication?
>>>>> (Also seems quite difficult, correct?)
>>>>>
>>>>> There was a similar comment on this recently, though I didn't
>>>>> see any solution discussion?
>>>>> http://biobricks.org/pipermail/standards_biobricks.org/2008-February/000034.html
>>>>> "The lack of physical separation of signals, as is the case in
>>>>> microelectronics, could be one of the biggest limitations to the
>>>>> standardized bioparts concept. "
>>>>>
>>>>>
>>>>> The idea offered in that thread,
>>>>> "Actually this could lead to a design process where you operate
>>>>> e.g. on a
>>>>> system level and design your nice circuit, but depending on the
>>>>> circuit the
>>>>> design computer programme chooses one of different devices (and
>>>>> finally
>>>>> parts) that interact in the way you like."
>>>>>
>>>>> This is actually quite impossible for multi-gate designs. A multi-
>>>>> bit adder
>>>>> or memory storage would require far too much complexity. The
>>>>> components
>>>>> (i.e. "transistors") need to have completely compatible outputs so
>>>>> that
>>>>> any output can connect to any input. Otherwise they are not
>>>>> "generic parts".
>>>>> Similar issues led to much complexity in the early days of
>>>>> semiconductor
>>>>> fab.
>>>>>
>>>>> --
>>>>> Cheers!
>>>>>
>>>>> _______________________________________________
>>>>> Standards mailing list
>>>>> Standards at biobricks.org
>>>>> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>>>>>
>>>>>
>>>> _______________________________________________
>>>> Standards mailing list
>>>> Standards at biobricks.org
>>>> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>>>>
>>>>
>>> <j_lawson.vcf>_______________________________________________
>>> Standards mailing list
>>> Standards at biobricks.org
>>> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>>>
>> _______________________________________________
>> Standards mailing list
>> Standards at biobricks.org
>> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>>
>
>
>
> *----------------------------------------------------------*
> Dr. Markus Schmidt
> International Dialogue and Conflict Management
> Abt-Karlg. 19/21, 1180 Vienna, Austria
> phone: +43(0)19900811
> mobile: +43 660 6856623
> www.idialog.eu
> email: markus.schmidt at idialog.eu
> *-----------------------------------------------------------*
>
>
>
>
>
> _______________________________________________
> Standards mailing list
> Standards at biobricks.org
> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>
More information about the Standards
mailing list