[BBF Standards] Two separate standards

[Deepak Chandran]

Hello Raik and rest,

Reshma's description on devices is very clear (on the link you had 
provided). My discussions were targeting basic biological parts vs 
composite parts (instead of basic parts vs devices). Turning a part into 
a device is a matter of adding the correct interface and identifying the 
function (correct?).

Nonetheless, any composite part is a circuit made from basic parts. This 
circuit wiring diagram is very important to include in our standard 
representation of any composite part, whether it is an complex 
oscillator or a protein generator. Such a circuit wiring diagram does 
not exist for basic biological parts (correct?).

This does not have anything to do with the Registry's hierarchy or what 
family a part belongs under. I am just saying that in order to fully 
describe a composite part, I would need different descriptors than when 
describing a basic part.

Additionally, I do not understand why the sequence information is needed 
for composite parts -- isn't this redundant information which makes room 
for errors in a database?

If there are two representations (for basic and composite parts), the I 
agree with Ralf that they should not be allowed to wander completely 
independent of one another.

--Deepak

Raik Gruenberg wrote:
> Hi Deepak,
>
> the starting point of your interesting discussion may be a mix-up of 
> the terms 'device' and 'part'. Which is kind of not surprising because 
> our use of the terms is often fuzzy. Please have a look at our current 
> Biobrick, Part, Device definition (and comment if there is anything 
> unclear or inconsistent):
>
> http://openwetware.org/wiki/The_BioBricks_Foundation:Standards/Technical/Exchange#What_is_a_Biobrick.3F 
>
>
> So parts/Biobricks are the physical pieces of DNA that are passed 
> around -- something you can put into a plasmid, run on a gel, 
> sequence, etc. Composite Biobricks are still the very same, their 
> sequence (which is still unique) just happens to be a concatenation of 
> other Biobricks.
>
> As Drew and Josh pointed out, there is no single 'atomic' detail level 
> in biology but modularity is emerging in hierarchical layers. So what 
> you would consider a basic part may be constructed as composite part 
> by somebody else.
>
> By contrast, our definition of a device better approaches what you 
> were talking about:
>
> *  Devices are combinations of one or more parts that have a 
> human-defined function.
> * Some devices can be encoded in a single stretch of DNA (a basic or 
> composite part), others encompass disconnected parts (e.g. encoded in 
> two different locations, possibly even cells).
> * (suggestion Reshma) Devices expose specified interfaces for their 
> functional connection with other devices (example: PoPS)
> * (suggestion Raik) A Biobrick device is defined by a unique 
> combination of unique Biobricks
>
> So this is the level where we can start to create 'black boxes' with 
> standardized interfaces.
>
> Following this definition I would re-formulate your question ;-) Do we 
> need two different standards -- one for parts, and one for devices?
>
> IMO, since we already have two different names for it, we can keep 
> them in the same standard. We just have to keep the standard *open* 
> for the definition of interfaces and functions at different levels (of 
> modularity).
>
> Greetings,
> Raik
>
> Deepak Chandran wrote:
>> Hello Josh, Drew, and rest,
>>
>>    While it seems logical to first have standard definitions for 
>> promoter, rbs, coding regions, terminators, etc. before working on 
>> standards for devices and systems, I think the two are somewhat 
>> independent. People have been building biological circuits 
>> successfully (iGEM students for example). What I am proposing is for 
>> some of us (myself included) to write out a format that can represent 
>> all of the iGEM projects. This format can allow others to take iGEM 
>> projects (and any other such projects) and connect them together to 
>> form larger circuits without having to understand the details of the 
>> project itself. All these circuits will be represented in the 
>> standard language so that one does not need to reconstruct an entire 
>> device to reuse it. I think this is very much possible while the 
>> standards on promoters/rbs/etc are still pending. It is also a bit 
>> easier maybe, because many concepts can be borrowed from engineering 
>> disciplines.
>>
>> --Deepak
>>
>> Josh Perfetto wrote:
>>> I think you are making some very insightful remarks without 
>>> realizing it :)
>>> The only way to engineer large-scale systems is by breaking the 
>>> system down
>>> into modular units with well-defined interfaces, otherwise it just 
>>> gets too
>>> complex.  The circuit diagram analogy is perfect: inputs, black 
>>> boxes, and
>>> outputs, and I think it is a reasonable representation at this 
>>> level.  If we
>>> are defining things at this level, and have well-defined interfaces 
>>> like
>>> PoPs for inputs and outputs, then I agree with you that it is not 
>>> necessary
>>> to specify the DNA sequence of the entire circuit, but rather just 
>>> define
>>> the order of the black boxes, which can define their own DNA 
>>> sequences.  You
>>> can replace one black box of the circuit diagram with another and 
>>> predict
>>> what will happen.  Once we standardize how to specify devices at 
>>> this level,
>>> including devices with multiple inputs, it will be very powerful.
>>>
>>> When we drill down into other levels, like the upstream and core 
>>> portions of
>>> a promoter, obviously PoPs is a totally useless interface for 
>>> defining how
>>> these parts fit together, and so a circuit diagram is not 
>>> appropriate for
>>> specifying the workings of a promoter as you said.  We need to 
>>> decide at
>>> which levels standardization would be useful and how to define 
>>> interfaces
>>> between parts at these useful levels.  However I don't think it's 
>>> essential,
>>> at this stage, to model the world.  For example, if we want to 
>>> standardize
>>> how to define promoter behavior, we can leave the internals of promoter
>>> working as a black art, or at least out of the scope of standardized 
>>> parts,
>>> and only define the characteristics of the promoter at that level, 
>>> at least
>>> for now.
>>>
>>> I have not been in this community very long and am unsure about the 
>>> exact
>>> definition that others are attributing to the term "devices" (and I 
>>> think
>>> this does vary a bit, perhaps because it is not yet standardized :) 
>>> ).  But
>>> I don't see any problem with defining devices formally in terms of 
>>> A) PoPS
>>> in/out, and B) other modifiers like UV/LuxR.  I think a key question 
>>> is how
>>> much these other modifiers from B are standardized in their 
>>> definition vs.
>>> how much this is an ad-hoc definition.  I think it's probably best 
>>> to follow
>>> the 80/20 rule and standardize the 20% of properties that have are 
>>> used 80%
>>> of the time, like chemical concentration, where the chemical in 
>>> question is
>>> defined by something like CAS registry number, while leaving the other
>>> specialized properties that some devices might measure, like 
>>> magnetic field
>>> orientation, unstandardized.
>>>
>>> -Josh
>>>
>>>  
>>>
>>> -----Original Message-----
>>> From: Deepak Chandran [mailto:deepakc at u.washington.edu] Sent: 
>>> Friday, March 14, 2008 7:24 PM
>>> To: Josh Perfetto
>>> Cc: 'Drew Endy'; standards at biobricks.org
>>> Subject: Re: [BBF Standards] Two separate standards
>>>
>>> Hello Josh and Drew and rest,
>>>
>>> The Polkadots system is a great example. Basically what I am saying 
>>> is that the sequence doesn't need to be specified in order to define 
>>> the polkadots system. As long as I know the parts and the order in 
>>> which they are placed on the plasmid(s), I can reconstruct the 
>>> system. By "interaction between parts", what I mean is that in order 
>>> to understand how polkadots is working, I need to see the circuit 
>>> that is presented on the polkadots website. It would be nice if the 
>>> circuit diagram was our standard representation, but that is not 
>>> reasonable.
>>>
>>> I think an operon is a similar case -- it is a very simple circuit. 
>>> One can draw a diagram that explains how an operon functions. So, I 
>>> am looking at operons as engineered objects (though humans didn't 
>>> engineer them first).
>>>
>>> What I am proposing is that circuits such as polkadots or even 
>>> operons should be defined in such a way that a synthetic biologist 
>>> can cut out one of the components of the circuit and replace it with 
>>> another device/part. If we cannot do that, then these biological 
>>> circuits are not engineerable.
>>>
>>> On the contrary, a promoter is different. I don't know how one can 
>>> explain the inside of a promoter using a circuit diagram. The fact 
>>> that promoters can be broken down into smaller components is true. 
>>> But can you take the first half of one promoter and stick it to the 
>>> second half of another promoter (I don't know)? Would the resulting 
>>> promoter have predictable behavior? Similarly, can I take pieces of 
>>> two RBS and piece them together and be able to predict the function 
>>> of the resulting RBS? In constast, replacing the promoter or RBS 
>>> inside an operon has predictable behavior, and the affect of 
>>> replacing parts inside the polkadots can also be predicted (just 
>>> like I can predict the result of changing my computer's CPU). I 
>>> suppose one could say replacing protein domains has predicable 
>>> behavior.
>>>
>>> By "interaction between parts", I did not mean any particular type 
>>> of interaction. But I was not clear on this: is PoPS the standard 
>>> input/output for BioBrick parts? Does this mean that I cannot have a 
>>> device that uses UV as an input and/or produces LuxR as an output?
>>>
>>> --Deepak
>>>
>>> Drew Endy wrote:
>>>
>>>  
>>>> Not true.  A promoter can be disassembled, for example in bacteria, 
>>>> into -10 and -35 regions.
>>>>
>>>> You may find it helpful to stop thinking about natural objects such 
>>>> as operons and start thinking about devices as *engineered* objects.
>>>>
>>>> For example, consider the material produced in 2004 by the Polkadorks:
>>>>
>>>> http://parts.mit.edu/wiki/index.php/IAP2004:Polkadorks
>>>>
>>>> In the Polkadorks work you can see that they clearly defined device 
>>>> boundaries based on PoPS, including PoPS pass-through boundaries 
>>>> leading to multi-device mRNA.     
>>> Josh Perfetto wrote:
>>>  
>>>> Hi Deepak,
>>>>
>>>> I think that the view that a promoter cannot be sub-divided reflects a
>>>> particular level of interest.  From another point of interest, it 
>>>> would be
>>>> very interesting to break promoters down into UPE, spacers, core
>>>>     
>>> promoters,
>>>  
>>>> boxes, etc., and look at the "interactions within the part" with 
>>>> factors
>>>> like UBF1.
>>>>
>>>> You are bringing up an interesting point with your operon.  There 
>>>> are a
>>>> couple ways you could define an operon.  For example, you could 
>>>> create an
>>>> operon sensitive to a specific repressor, and define it in part by a
>>>> function that relates the concentration of that repressor to the 
>>>> level of
>>>> polymerase that will flow out.  Or, you could define it in such a 
>>>> way as
>>>>     
>>> it
>>>  
>>>> has an interchangeable repressor sub-part, and define the operon in 
>>>> part
>>>>     
>>> by
>>>  
>>>> a function which is dependent on another function defined by the 
>>>> sub-part
>>>> part definition.
>>>>
>>>> If we were to take the first approach, it doesn't mean that the operon
>>>>     
>>> can't
>>>  
>>>> be decomposed which obviously it can, it just means that such
>>>>     
>>> decomposition
>>>  
>>>> is not an interface of the part, but an internal detail of the 
>>>> part.  The
>>>> second approach doesn't represent the ability to arbitrary combine 
>>>> parts
>>>>     
>>> and
>>>  
>>>> be able to predict what will happen, but rather a part which is
>>>>     
>>> specifically
>>>  
>>>> designed to take another part of a specific type and as a parameter 
>>>> and
>>>> predict what will happen.
>>>>
>>>> But stepping back for a moment, aren't we now really discussing 
>>>> devices?
>>>> I.e. the first approach is a single device with a PoPs out that's
>>>>     
>>> dependent
>>>  
>>>> on a specific repressor, while the second approach is two devices, the
>>>>     
>>> first
>>>  
>>>> with a PoPs out, and the second that relates its PoPs in and the 
>>>> level of
>>>>     
>>> a
>>>  
>>>> specific repressor to produce a PoPs out value?  When you say "the 
>>>> interaction between the parts is the crucial information", it seems 
>>>> that
>>>> you're really either talking about an interface in terms of PoPs 
>>>> (in which
>>>> case these parts could be called devices at least in my 
>>>> understanding of
>>>>     
>>> how
>>>  
>>>> the term is being used), or are proposing that another interface 
>>>> besides
>>>> PoPs be standardized in the definition of parts.
>>>>
>>>> -Josh
>>>>   -----Original Message-----
>>>> From: standards-bounces at biobricks.org
>>>> [mailto:standards-bounces at biobricks.org] On Behalf Of Deepak Chandran
>>>> Sent: Friday, March 14, 2008 2:31 PM
>>>> To: Drew Endy
>>>> Cc: standards at biobricks.org
>>>> Subject: Re: [BBF Standards] Two separate standards
>>>>
>>>> Drew,
>>>>
>>>> The parts,devices,systems hierarchy is fine. What is confusing to 
>>>> call an entire operon (possibly more than one promoter, rbs, gene) 
>>>> a "part" and a simple promoter a "part" as well. I can take an 
>>>> operon apart and replace one of the rbs, whereas the promoter 
>>>> cannot be taken apart. There is a difference between these two from 
>>>> an engineering perspective.
>>>>
>>>> I think that the operon should be defined in terms of the parts 
>>>> that it is made from and how they are interacting. Each individual 
>>>> part would then contain the sequence information and other 
>>>> characteristics. For anything made from multiple parts (say a 
>>>> bistable switch), the interaction between the parts is the crucial 
>>>> information. But if you take a basic part like a promoter, there is 
>>>> no interaction within the part. This is why I think that the 
>>>> standard definitions for basic parts and parts made from more than 
>>>> one part should be different.
>>>>
>>>> Again, I am completely open to disagreement here. It would be 
>>>> sufficient to explain how you can apply the same standards to 
>>>> describe a bistable switch and a promoter.
>>>>
>>>> --Deepak
>>>>
>>>> Drew Endy wrote:
>>>>      
>>>>> Parts are parts.  I'd use the language basic parts if you need a 
>>>>> modifier.
>>>>>
>>>>> Devices are composite parts.  But, not all combinations of parts 
>>>>> are devices.  Devices are limited by the requirements bounding 
>>>>> device specification.  For example, polymerase per second (PoPS) 
>>>>> serves to define a common signal carrier for inputs and outputs 
>>>>> when dealing with gene expression devices.
>>>>>
>>>>> Note that there is no absolute physical definition of a part.  You 
>>>>> can keep drilling down into particles and sub-atomics.  So, I 
>>>>> believe that the standards should be linked to the human-invented 
>>>>> abstraction hierarchy (i.e., parts, devices, systems), which was 
>>>>> invented for practical reasons (i.e., it is good for something).  
>>>>> Support for reliable physical and functional composition is 
>>>>> definitely one of the first and best goods that we can have.
>>>>>
>>>>> For the small amount of protein engineering work going on.  
>>>>> Docking and phosphorylation motifs are typically though of as 
>>>>> parts.  For the small amount of RNA engineering work going on, 
>>>>> aptamers and ribozyme active sites are also thought of as parts.
>>>>>
>>>>>
>>>>>
>>>>> On Mar 14, 2008, at 12:14 PM, Deepak Chandran wrote:
>>>>>
>>>>>          
>>>>>> Hello standards group,
>>>>>>
>>>>>> From all the standards discussion, I think that there are two 
>>>>>> separate
>>>>>> standards for parts. If this is the case, then it should be made 
>>>>>> explicit.
>>>>>>
>>>>>> The first type of standard is for atomic parts (better word needed?)
>>>>>> such as promoters, rbs, etc. (that cannot be broken down 
>>>>>> further). The
>>>>>> "standards" for these parts is simply a list of characteristics. For
>>>>>> example, a promoter would have its sequence and Jason-units as its
>>>>>> characteristic feature. RBS would have sequence and some other 
>>>>>> units.
>>>>>> The list of characteristic features are probably best if 
>>>>>> developed by
>>>>>> experimentalists -- the question to ask is: what is it that makes 
>>>>>> this
>>>>>> promoter unique?
>>>>>> Sequence information is definitely needed for atomic parts.
>>>>>>
>>>>>> The second type of standard is for composite parts (better word?) 
>>>>>> such
>>>>>> as iGEM projects. I do not think that the sequence for composite 
>>>>>> parts
>>>>>> is needed -- if we know the atomic parts that make up this composite
>>>>>> part, then we can easily determine the sequence. What is needed 
>>>>>> is the
>>>>>> order in which the atomic parts are arranged on the plasmid and the
>>>>>> circuit-diagram that explains the mechanism of the part. Someone 
>>>>>> should
>>>>>> be able to take a composite parts and replace some of the atomic 
>>>>>> parts
>>>>>> just like upgrading a computer by replacing the RAM.
>>>>>>
>>>>>> One can wonder whether things like protein domains or fusion 
>>>>>> proteins
>>>>>> are atomic or composites. I think the question is whether the 
>>>>>> part is
>>>>>> modular (i.e. whether it can be taken apart and reconstructed using
>>>>>> slightly different components).
>>>>>>
>>>>>> In summary:
>>>>>> A language is needed for composite parts.  (sequence not needed)
>>>>>> A list of characteristics is needed for atomic parts. (sequence 
>>>>>> needed)
>>>>>>
>>>>>> If there is disagreement on this, please let me know why, so that 
>>>>>> I can
>>>>>> eliminate my confusion. If there is agreement, then perhaps we 
>>>>>> should
>>>>>> make this fact explicit and categorize our proposals to one of 
>>>>>> the two.
>>>>>> We can make more directed progress that way.
>>>>>>
>>>>>> By the way, this categorization is not meant to disturb the
>>>>>> part/device/system hierarchy.
>>>>>>
>>>>>> Ralf, I think that you language makes more sense if it is for 
>>>>>> composite
>>>>>> parts, because you would need more descriptions for an atomic 
>>>>>> part. I
>>>>>> will comment on that later.
>>>>>>
>>>>>> --Deepak
>>>>>>
>>>>>> _______________________________________________
>>>>>> 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
>>>>
>>>>       
>>>   
>>
>> _______________________________________________
>> Standards mailing list
>> Standards at biobricks.org
>> http://biobricks.org/mailman/listinfo/standards_biobricks.org
>>
>>
>