From Jones and Bartlett, a book on Stem Cells from Dr. Ann A. Kiessling and Scott C. Anderson:

Selected Articles:

July 26, 2012

God in the Details: the Whole Cell Simulator

Researchers simulate a living cell down to the last molecule

What I cannot create, I do not understand.

-- Richard P. Feynman

In the July 20 issue of Cell, Markus Covert and researchers from Stanford and the J. Craig Venter Institute describe a computer program that simulates every aspect of a living cell, down to the last molecule. Their signal achievement is to stitch together all the models of cellular activity, including protein generation, metabolism and cell division, into a single executable. This has been a holy grail of computational biology, and it is now in our hands. Literally. You can download the program:

If you have MATLAB, you're home free. Just install and simulate. This is not a tool reserved for elite cell-hackers at Stanford; it's an open-source platform for researchers everywhere. Because it is freely available, refinements will pour in, the model will expand, and one bright day we'll be simulating entire human systems. This is an exciting first shot at the ultimate targets: disease modeling and drug discovery. Simulations have the potential to be faster and easier than exhaustive laboratory experiments, and as a bonus, no test animals are required.
This is not the first attempt at a cellular simulator. Variations have been around almost since the beginning of computers. From their inception, computers seemed almost conscious, and therefore oddly lifelike. The newfangled machines were replacing human beings whose job was to perform complex calculations for science and the military. These people were called computers. To extract their highly-trained mathematical skills – arguably the pinnacle of human intellect – and inject it into a machine was literally a heady endeavor. Theories of computing and theories of life have been intertwined ever since. John Conway wrote the Game of Life in 1970, and it was a wonderful demonstration of how local rules can lead to global action. Simulations have become richer and more realistic ever since. In 1997, Craig Venter and company announced the availability of E-CELL, a cell simulator based on an idealized proto-bacteria with 127 genes – just enough to metabolize and clone itself. It showed what might be possible with real bacteria.

Mycoplasma GenitaliumSo, what goes on inside a cell? In particular, what goes on in Mycoplasma genitalium? First off, let's expose this tiny bug as parasite that thrives in primate genital tracts. Yes, the first creature chosen for simulation is a sexually transmitted disease. It's a perfect candidate, however, because it's tiny, has very few genes and it's familiar: the same bug was used as a resource for E-CELL and it was the template for the first synthetic creature ever designed, by researchers at J. Craig Venter Institute, who also assisted with this simulation project.

In a cell, DNA is the master blueprint that codes for everything. Accordingly, the simulator starts with DNA, and from that humble thread it weaves the complete tapestry of a living cell. Or, dispensing with metaphor: the phenotype is derived from the genotype. DNA codes for proteins which replicate the DNA, which codes for proteins, etc. That leaves out a few details, though. Okay, a few thousand details. But for people who thought that it was impossible to reduce this exquisite dance to mere molecules, here is a revelation: it now seems possible to simulate life with mere molecules. Here are some of those details.

For a bacteria, the primary job is replication. It's all about splitting heirs. If you aren't good at cloning, you won't be popular – and it is a popularity contest. Whoever clones the most owns the most. Real estate accrues to the prolific. And so the big job is replicating the master DNA, which is a correspondingly big part of the software. The replication module simulates a large protein scuttling along the tracks of the DNA, unzipping it and spinning out a copy as it goes.

But we're getting ahead of ourselves. Replication is the grand finale, just before the cell splits and the credits roll. Before then, we need to cast all the supporting players. For that, we need the DNA to build the proteins, which are the real actors in this play. If the DNA is the master program, a gene is a subroutine, containing all the code needed to assemble a protein. Making those proteins involves other large proteins that also chug along the busy track of DNA. These proletarian proteins make an RNA copy of each gene. This is called transcription and another module of the program simulates that as well.

The RNA cruises toward a hookup with a ribosome,  an ancient cellular factory that reads the RNA transcripts and manufactures proteins to spec. Ribosomes are a mix of RNA and proteins, and they are essential characters in our story. Every living cell on the planet has ribosomes. You can't make proteins without them. There is a module of the simulator for that, too.

Like a train through a tunnel, the RNA slides through the ribosome. Inside, the transcript is decoded, base by base, into amino acids. The ribosome then acts as an enzyme and links those amino acids together into a growing ribbon of protein. Those proteins go on to help with metabolism, cell structure and back, of course, to job number one, replicating the master DNA. All of these hectic activities are simulated by different modules of the Stanford code.

The constant shuttling along the DNA is reminiscent of a busy train station, and potentially just as dangerous. Alarmingly, the replication proteins and the transcription proteins can collide. When that happens, you may get a bad protein. Or far worse, you could screw up the master DNA replication. According to the simulation, this happens surprisingly often, over 30,000 times per cell cycle, messing up about one protein per second. It happens often enough that nature provides a backup protein that sweeps in and restarts the replication process whenever a replication protein gets knocked off. Nevertheless, this is a built-in source of mistakes that could seriously affect a cell or its progeny.

How do the simulations work to represent all this? They are painfully assembled from hundreds of exacting experiments that document just how fast each reaction is. A reaction can be represented by a differential equation that is solved numerically by a program like MATLAB. Given appropriate starting conditions, you just run each equation out for a short period (one second in the Stanford simulator) and then store the answers for the next period. The researchers grouped the number of basic cellular reactions into 28 modules. One deals with replication, one deals with metabolism, another with translation, etc. These modules each operate according to quite different rules, and so they are run as independent simulations.

After a second of run-time, the variables are collected and fed into any other module wherever there is an overlap. With these new starting values, the simulation is then run for another second. The process continues, second after second, until the cell splits in two, at which point the whole process starts all over again.

How well does the simulation work? Very well. It predicts reactions that are known to take place in the cell, and it even predicted unknown actions that were later verified experimentally. Given the DNA, it predicts how the cell reacts with its environment. Currently it takes a supercomputer with 128 processors over ten hours to take the cell from birth to division. That, coincidentally, is how long it takes the mycoplasma to divide in the real world.

The Stanford team ran thousands of simulations, deleting various genes, just to see what was absolutely essential. They found 284 genes that the cell couldn't live without. The rest, it seems, are for fine-tuning or improving efficiency. Considering the primacy of master gene cloning, it was odd to note that metabolic genes, not replicator genes, were most important to the smooth functioning of this simulated cell.

Here's a short video of the read-out from the simulator. It shows how the cell slowly grows while metabolic chemicals provide energy, DNA is processed into proteins and finally, the big kahuna, cell division itself.

God is an ever receding pocket of scientific ignorance.
― Neil deGrasse Tyson

This research seems to settle a contentious argument about something called strong emergence. This is the theory that a complex process can't simply be reduced to its constituent bits, because the whole is greater than the sum of its parts. Something happens in the assembly that is mysteriously non-reductive. Accordingly, biology can't really be derived from its underlying chemistry. According to Philip Warren Anderson, "The constructionist hypothesis breaks down when confronted with the twin difficulties of scale and complexity. At each level of complexity entirely new properties appear. Psychology is not applied biology, nor is biology applied chemistry. We can now see that the whole becomes not merely more, but very different from the sum of its parts."

Unfortunately for that philosophy, the simulation created by Covert et al. needs no mysterious forces to go from chemistry to biology. Given the right starting materials, life simply falls out. As to those starting materials: the simulation is a masterful recreation of the life cycle of a bug, but it doesn't start de-novo. There are initial values that need to be set before the simulator can start. It doesn't attempt to create those basic building blocks, which would be a different simulation: the simulation of the origins of life itself. Give them time. Instead the simulator starts with certain givens that reflect the end products of the previous simulation. The cell has split in two, and each one has inherited the products of the original cell. Properly primed, the entire life-cycle of the cell can be represented by basic chemistry. One reaction leads to another and before you know it, you're churning out copy after copy of the same master plan. This is life from chemistry, running in silicon.

As awesome as the simulation is for scientists and life-hackers everywhere, it is this defeat of mysterious life-forces that may leave the most lasting legacy. The simulation establishes with finality that science, with its relentless reductionism, is still the best way to understand the world. That is not meant to diminish the enormity of their accomplishment, but it may cause a frisson of humility as we are reminded once again that we are just stardust, a straightforward (if gnarly) extension of physics and chemistry. The magic, if that is the right word, is that we could create tools like math and computers that have allowed us to model a living creature.

Have we squashed the spirit and banished the soul? Only in a frivolous reading of those words. Do you feel diminished by knowing that there is a real world of physics and chemistry underlying your life? As this work makes clear, just because you understand everything down to the last molecule doesn't mean the beauty is diminished. That which we call the "essence of life" has been exposed as an amazing choreography of proteins, sugars, fats and DNA. Once it got started in some ancient incubator, it couldn't be stopped. Far from prosaic, that runaway chemical reaction has managed to coat the entire planet with a breathtaking array of creatures.

Philosophers may quibble that we surely shouldn't have expected to find a soul in a bacteria in the first place. And it may be true that bacteria are so different from us that they don't really count. But the sobering reality is that humans are little more than a large collection of bacteria with a few mammalian cells tossed in. There is surely a higher order of organization in multicellular creatures, and the brain is still safe with its secrets, but it's hard to believe that we aren't – at root – running by the same rules as all other forms of life. It's chemistry, all the way down.

So what's next with Sim Cell? One idea is to scale it up to bigger bacteria, such as E. Coli, which is a more interesting and ubiquitous bug. E coli has about ten times more genes and proteins, so the challenge is grand. Another idea is to scale up to eukaryotes, true nucleated cells. That is a jump to a cell some 10,000 times larger with almost a hundred times as many genes. That is a huge challenge. The authors don't mention it, but another simulation might be designed to study the creation of life from scratch, starting with all inorganic molecules and building from there.

One thing seems sure: having a simulation of life itself is pretty cool. And there's probably no going back.



Karr JR, Sanghvi JC, Macklin DN, Gutschow MV, Jacobs JM, Bolival B, Assad-Garcia N, Glass JI, Covert MW. A Whole-Cell Computational Model Predicts Phenotype from Genotype, Cell - 20 July 2012 (Vol. 150, Issue 2, pp. 389-401)

Anderson, P.W. (1972), "More is Different: Broken Symmetry and the Nature of the Hierarchical Structure of Science", Science 177 (4047): 393-396, Bibcode 1972Sci...177..393A, DOI:10.1126/science. 177.4047.393, PMID 17796623

M.M. Hanczyc et al., "Experimental models of primitive cellular compartments encapsulation,
growth and division," Science, 302:618?22, 2003.

D. Deamer, "A giant step towards artificial life?" Trends Biotechnol, 23:336?8, 2005.

V. Noireaux, A. Libchaber, "A vesicle bioreactor as a step toward an artificial cell assembly," Proc
Natl Acad Sci, 101:17669?74, 2004.

S. Rasmussen et al., "Bridging nonliving and living matter," Artif Life, 9:269?316, 2003.

K. Ishikawa et al., "Expression of a cascading genetic network within liposomes," FEBS Lett,
576:387?90, 2004.

W.K. Johnston et al., "RNA-catalyzed RNA polymerization: accurate and general RNA-templated
primer extension," Science, 292:1319?25, 2001.

K. Kobayashi et al., "Essential Bacillus subtilis genes," Proc Natl Acad Sci, 100:4678?83, 2003.

T. Oberholzer et al, "Protein expression in liposomes," Biochem Biophys Res Commun, 261:238?
41, 1999.

Tomita M, Hashimoto K, Takahashi K, Shimizu T, Matsuzaki Y, Miyoshi F, Saito K, Tanida S, Yugi K, Venter JC, Hutchison CA. E-CELL: Software Environment for Whole Cell Simulation. Genome Inform Ser Workshop Genome Inform. 1997;8:147-155.

Claire M. Fraser, Jeannine D. Gocayne, Owen White, Mark D. Adams, Rebecca A. Clayton, Robert D. Fleischmann, Carol J. Bult, Anthony R. Kerlavage, Granger Sutton, Jenny M. Kelley, Janice L. Fritchman, Janice F. Weidman, Keith V. Small, Mina Sandusky, Joyce Fuhrmann, David Nguyen, Teresa R. Utterback, Deborah M. Saudek, Cheryl A. Phillips, Joseph M. Merrick, Jean-Francois Tomb, Brian A. Dougherty, Kenneth F. Bott, Ping-Chuan Hu, Thomas S. Lucier, Scott N. Peterson, Hamilton O. Smith, Clyde A. Hutchison III, J. Craig Venter. The Minimal Gene Complement of Mycoplasma genitalium. Science 20 October 1995: Vol. 270 no. 5235 pp. 397-404 DOI: 10.1126/science.270.5235.397.

Meyers S, Friedland P. Knowledge based simulation of genetic regulation in bacteriophage lambda. Nucleic Acids Res., 12.1-9.

Copyright © 2000-2014 by Scott Anderson
For reprint rights, email the author:

Here are some other suggested readings in computer simulation: