May 21, 2010

About Life

Mycoplasma mycoides - AP Photo/J. Craig Venter Institute

I usually don't like to be repeating other people's post, specially because people probably read about that everywhere else before reading here. :( But I was thinking about writing a post about life for some time and then I will use the opportunity to do so.

As news run fast, everybody by now must know about the creation of an "artificial" cell by the J. Craig Venter's Insitute. The paper was published in Science Express:

[1] Gibson et al., Creation of a Bacterial Cell Controlled by a Chemically Synthesized Genome, / 20 May 2010 / Page 4 / 10.1126/science.1190719

As is well reminded in this discussion in Nature (you can also find another nice discussion in the Edge that includes the opinion of Freeman Dyson), what they've done was to construct a pre-designed DNA from parts extracted from a unicellular organism called Mycoplasma mycoides (the one in the photo above), reassemble it and implant it into another of these organisms from which the original DNA was previously extracted. There is not synthetic components in the usual sense of the word, what was "synthetic" was the design of the new DNA, which did not exist before. They gave the new organism temporarily the beautiful and creative name Mycoplasma mycoides JCVI-syn1.0 (the guy is already going to win a Nobel and still wanted to imortalise his initials in the poor cell).

The feat is not small though. It has important implications and open many doors. One of the important things is that it serves to test some of our present knowledge about how DNA works. DNA is a program and what they are doing is writing test programs to check how much we really know about the programming language. Given that the created cell seems to function and reproduce normally, it shows that we at least understand the basics about writing these programs.

Think about the cell as a very delicate computer that will interpret a program encoded in the form of DNA. Mathematically, its okay to see the cell as a Turing machine with the DNA being the input tape (to see what I am talking about, look at the first YouTube video in the sidebar to your left). Consequently, one of the doors their work opens is that it is possible to vary the program and see what happens.

Before I continue to talk about life, let me just list here some articles and blog posts about JCVI's team feat:

I have been in some conferences on complex systems lately and many people has been trying to create artificial organisms really from scratch. It is a different kind of strategy where instead of using already DNA, the idea is to discover what exactly are the minimum possible assemblage of parts that is capable of creating a life-form when put together. Well, everybody knows that the definition of life is problematic, but in this case the idea is to create a self-replicating thing that is autonomous, in the sense that once created, it does not need our assistance to survive and reproduce anymore. I believe many groups are working on that, but I cannot find the references here right now. If somebody would like to point some of them, I would be glad to include here.

In any case, you see that life may have different meanings and definitions. You may have been convinced by someone that in order for an organism to be alive it must reproduce right? Wrong. This is a species-based definition. Reproduction is important for natural selection, but a sterile man is still alive, even not being able to reproduce itself. An individual-based definition for life would be a way to take any "individual" in the universe and just looking at it decide if it is alive or not. Not an easy task, of course.

When a definition is difficult to attain, the way around is to identify some properties that this definition must include. This is equivalent to say that if we cannot find sufficient and necessary conditions for something, we can concentrate either on sufficient or on necessary ones. I have been discussing it with some colleagues and one condition I really consider necessary for any individual to be alive is the capacity of processing information. In a more mathematical way, I would say that any living organism must be a Turing machine, of course not necessarily universal. Just to make it clear, this is a necessary, not sufficient condition. Not all Turing machines are alive, although I would never consider anything that cannot change its states according to some input information alive.

So, the one point I want to make here is that life requires information processing. Unfortunately, this is the only necessary condition in my list, although I think something can already be explored by using it. For instance, I have read many comments in Cosmic Variance about life and the second law. This is a quite interesting discussion actually: does life is dependent or invariably linked to the increase of entropy? A way to try to attack this problem is by looking at the necessary conditions for life and see if they need it. Now, I must draw attention to something which is called reversible computation. Rolf Landauer, some time ago, proposed the famous idea that the erasure of one bit of information increases the entropy of the environment by an amount of $k_B \ln 2$, where $k_B$ is Boltzmann constant. Usual computing gates, like the AND or OR gates, take two bits and give one in return. Modern computation is base on the use of these gates and, according to the so-called Landauer Principle, they necessarily are dissipative and increase the entropy of the environment.

However, there is another kind of gate which takes 2 inputs and give back other 2 in such a way that there is a one-to-one correspondence between inputs and outputs, by which I mean that given the outputs and the knowledge of which gate was applied, you can recover the inputs. An example is the CNOT gate. CNOT is reversible and do not erase information. According to Landauer Principle, CNOT do not necessarily increase entropy. Now, if there is any way, at least fundamentally, in which an organism could process information by using reversible instead of dissipative gates, it would not need to increase entropy at least for these process.

Although this is not such a big result, it demonstrates that formalising and analysing some fundamental aspects of life is possible. Although you may say I am biased, and I completely agree, these concepts seem to be nicely described in the framework of information theory (well, as well as thermodynamics/statistical physics, which in the end are all related). For instance, note that even reproduction is related to transmitting information from one individual to another. But I will deal with that in another post. 


Anonymous said...

Are all forms of life processing information? Even the most basic ones?

To me, the mystery of "life" resides in the strategies that are imbedded to sustain itself. DNA provides the code and the memory, but underneath this information are the rules that process this information and govern reproduction, mutation and natural selection.

Daniel Yokomizo said...

I would rather say that life is the computation, not the computer. In lambda calculus terms life is the expression, not the evaluator. The computer is the medium and physics is the operational semantics. Replication is just quines.

Roberto C. Alamino said...

Dear c0mplexity,

Everything you said may be important for a species-based definition, but the problem I was trying to address is a bit different, namely: imagine you are in a space exploration mission and you find just one specimen of "something" and want to decide if it is alive or not. How would you do that?

However, notice also that reproduction, mutation and natural selection are all process that involve information transmission in some way. But, as I said, they must be analysed for more than one individual.

Roberto C. Alamino said...


I tend to disagree slightly. Although I think that the computation is important, I would not consider a virus or a prion really alive. Of course, in the same way, I would not consider alive a cell without the DNA. I agree although that replication is really not important for an individual based definition.