Genetically encoded memory performance requirements

Back to Jerome's notebook

Note

 * A biological memory device have to fill some specifics requirements, some of them dictated by specific applications (see below).


 * So, maybe a good start would be to say: "we want our memory device to useable in the wider range of applications possibles".

We will probably have to make compromise between different requirements, but let's try to make the most performant system possible (of course).

= General requirements=

All these requirements are interdependent

the memory device must be reliable, so that writing, encoding and reading are efficients and information is not modified or lost and reading is easy.
 * Reliability

an important point is to be able to upscale from a first bit of information stored to built a system able to record multiple imputs, ideally using the same initial technology.
 * Scalability

The memory system should be modular, so for example the writing/storage mechanism could be combined with different imputs/reading mechanisms in a "plug and play" manner. We want to be able to record fast processes like c.elegans early embryonic cell divisions for example (20 min/cell cycle). This is a relative exigence that we can choose to adjust from a low to an absolute requirement. Ideally, the system should be useable in all kind or organisms from bacteria to Eukaryotes including plants, mammals etc....with no or minimal changes. So the less specie specific co-factors, the better the portability. However, we must keep in mind that if absolute, this requirement would exclude the use a number of mechanisms to construct the memory device. '''
 * Modularity
 * Reactivity
 * Portability: "trans-kingdom device"
 * '''Whatyouwant-ity

=Writing, storing, reading state=

The choice of the writing process(ese) and storage medium(s) are probably the more difficult and key parts in the project.

These two parts are intimately linked, however some specific requirements can be reached separately. Defining our needs will help orientate our choice(s)...

Writing
Speed


 * The quickest writing mechanism compatible with Hi-fidelity information processing/conservation.


 * Also depends on the application: for example, for a cell cycle counter, writing must be fast, as some cells divide each 20 mins...


 * so the faster the mechanism, the more the applications.

Hi-Fidelity, No interference


 * Information must be written non-ambiguously and in adequation with the signal received.


 * Writing must be activated only when the correct signal is present. This means that we probably will have to tune threshold levels for activating the "writer".

For example,we could use different enzymes from a same family, or coming from a same "ancestor" by directed evolution; in this case, we absolutly need to have no cross- talk between these different "writer".
 * Also, when new data is added, no change should be done to old data, except if desired.

Storage
Stability


 * Of course, the storage must be safe and stable during time and against environmental variation, ultimately organism death.

Structural organisation of information


 * The system should work in an incremental way, i.e. at each round of division for example we add information.

So how do we store data?

We can imagine different ways, but we will mostly use on using N switches that can have each at least 2 to n states. so we have also to think about how have combinatorial patterns, or imagine totally new mechanisms specifically designed for biological systems...

Heritability


 * State must be transmitted during cell division to the daughters cells.


 * That's why storing the state into DNA sequence, structure, or into DNA-structural protein is an attractive possibility.


 * If we choose to use other kinds of storage, we have to ensure that the storage molecules are equally segregated between each daugther cells at a sufficient concentration to maintain the state (see Ajo-Franklin et al., 2008)

Accessibility


 * The state should be easily accessible and readable.

This means that storage mechanisms that can lead to confusion should be avoided.

One could also imagine that the state is stored in different forms, which can be read by different procedures. For example, state could stored in DNA sequence, which could be viewed as an hard "back-up, and then this DNA arrangement determine different patterns of gene expression. So the state can be read by PCR, sequencing, and fluorescent proteins expression, morphology...(see Ham et al., 2008)

Reading
Reliability

The state must be read without any ambiguity.So, discrete "state values" should be preferred.

Speed

Here again, the faster we are able to read the state, the more applications we can have as we are able to implement/analyse more dynamic processes.

Non-destructive

Ideally,we should able to read the state without destructing the organism, so that we can follow different processes and sort living organisms according to their state to perform different experiments. Obviously, fluorescent reporters are excellent candidates.

=Summary note=

emergent requirements


 * Speed

to quickly write or read the state we can imagine to use post-translational modification, which have a response time-scale faster than transcriptionnal network for example. We could use signaling pathways, and engineered proteins or RNAs which switch conformation and write state when imput is present.


 * Stability, heritability...

With respect to these issues, DNA seems the most obvious candidate as a very stable molecule, support of genetic information transmitted to each daugther cell and the variety and great specifity of DNA editing enzymes. However, let's try to think about RNA or proteins...maybe with a DNA "hard back-up"?