Wednesday, December 15, 2010

Every program has bugs and every bug has a program!

Here's another Nature paper being embraced as though it's new news. Single celled organisms, in this case E coli, can be 'programmed' like computers.  As one report puts it,
A team at the University of California has successfully implanted E coli bacteria with the key molecular circuitry to act as computers.  They've given the cells the same sort of logic gates, and created a method to build circuits by 'rewiring' communications between cells. It means cells could be turned into miniature computers, they say.
Not electronic IO devices, these 'logic gates' are built of genes.  Naturally enough.  Since that's exactly how communication happens naturally within and between cells.  But the story is being reported so metaphorically, in robotic terms, that the beauty of how genetic signaling has worked for 4 billion years is getting lost in translation.

The idea that life works by a kind of logic was a theme in our earlier 2004 book with that title:  Genetics and the Logic of Evolution, and in Mermaid's Tale.  We didn't invent the ideas, of course, but have tried to show their ubiquity as characteristics of life: presence or absence, combinations of factors, and so on, are how life works. It's a combination of 'boolean' phenomena and others, and isn't a computer program the same way Excel (or blogspot.com) is.  But if evolution accounts for aspects of how life gets that way, is by no means the only story in life.

Here's one example of the computer metaphor invoked by the media:
"The result is that bacteria can be enslaved to become part of a hive mind computer, performing the will of a central controller."  
The Nature paper's senior author does the same in an interview: 
"Here, we've taken a colony of bacteria that are receiving two chemical signals from their neighbors, and have created the same logic gates that form the basis of silicon computing."

Staphylococcus biofilm (photo from
Wikimedia Commons)
In fact, the authors of the paper do appropriately recognize that this is how gene signaling actually works, since it was the basis of their work.  What they did was take advantage of the quorum sensing ability of bacteria, a trait that allows them to collect into large groups called biofilms by sending and receiving signals among themselves when times are tough, and group living would be more beneficial than trying to make it alone.  The researchers engineered two gene promoters, segments of DNA that are involved in 'turning on' a gene, to signal to a repressor, a gene that represses another gene's expression by blocking its promoter, to study the patterns of cell to cell communication between 2 different strains of E coli.  
The repressor inactivates a promoter that serves as the output. Individual colonies of E. coli carry the same NOR gate, but the inputs and outputs are wired to different orthogonal quorum-sensing ‘sender’ and ‘receiver’ devices. The quorum molecules form the wires between gates. By arranging the colonies in different spatial configurations, all possible two-input gates are produced, including the difficult XOR and EQUALS functions. The response is strong and robust, with 5- to greater than 300-fold changes between the ‘on’ and ‘off’ states. This work helps elucidate the design rules by which simple logic can be harnessed to produce diverse and complex calculations by rewiring communication between cells.
Everything in a digital computer is dependent on highly ordered on-off states (that can be represented as true/false or 1's and 0's). Continuing with the computer program metaphor, what the experiment shows is that a bacterium can be transgenically altered to switch genes on or off under logic-gate types of conditions.  The idea isn't new in any way, since it's how gene expression works, but the engineering aspect shows that progress is being made towards highly controlled manipulability of genome usage and, at least initially, the engineering of microbes to be the next beasts of burden for human society:  to do work for us.

There is, however, a danger in using metaphor uncritically or in the media. In that sense this is just more media hype, just like the arsenic story last week and so many Big Stories in genetics every week.  It's interesting in itself without that, even if it's being misrepresented in terms of the impression that this is conceptually new.

Genomes are not computer programs. They work very differently in many ways, and one of them is that they involve quantity as well as quality (they are not strictly 'Boolean'). The timing and concentration of components of a logical system are important in life, not just its digital aspects.  So may be their 3-dimensional concentrations within cells, or their 2-dimensional patterning on exposed cell surfaces, as well as their 1-dimensional pattern along chromosomes.  There are other dimensional aspects of gene control, too.  These quantitative aspects would perhaps correspond to what are called 'analog' computing.  Life uses both.  But in so many ways it is misleading to think of life in terms that would be familiar to C++ (or even Perl!) scribblers.

What this study shows is that there are controllable aspects of gene usage that go beyond individual gene knock-out or knock-in modifications such as are so often done in experimental systems like fruit flies or inbred mice.  Complex communication can be engineered: in some circumstance, bacteria start making something and secreting it, and other bacteria detect and respond to it in engineered ways.  The media are touting this as if, in a way, it will be an equivalent to a new electronics era, of an enormous array of useful products designed and produced.  Maybe it will, but to what extent is the story being taken beyond a technical achievement to boost stock prices?

The company with which the investigators collaborated, Life Tech, does have big plans, as reported here:
The company plans to commercialize the technology as genetic programming software, said Kevin Clancy, senior staff scientist of bioinformatics.
The software would "look just like programming language for a computer or a robot," Clancy said. It would convert the instructions into a DNA sequence that could be made by a DNA synthesis company. The DNA would then be inserted into the cell, such as a bacterial, yeast or mammal cell.
But, cells came first, so let's turn this on its head.  Digital computers, restricted to simple 0, 1 states as they are, even when the 1's and 0's are arrayed with as much complexity needed as to, say, connect the world through Facebook, would only make pretty primitive cells.  The complex combinatorial 'Boolean' aspects of a cell dwarf even the simple life of bacteria that are being manipulated.  But if all other things in the bacteria can be controlled well enough, the hope would be that the changes the engineers would like to add, would be seen through the rest of the complexity.  In spite of the media hype, clearly if bacteria really can be harnessed, (and this seems likely), this could have many useful applications.

1 comment:

peterfirefly said...

"Genomes are not computer programs. They work very differently in many ways, and one of them is that they involve quantity as well as quality (they are not strictly 'Boolean')."

" But in so many ways it is misleading to think of life in terms that would be familiar to C++ (or even Perl!) scribblers."

Actually, they are totally programs and they totally do computations. Sure, they may not look like iPads, the computation is not in silico, and the code doesn't look like C++.

It's still computation, though.

Sure, the more exotic forms of computation and programming are not found in first year comp. sci. classes. That doesn't mean they don't exist or that they are not studied.

In fact, computation, autocalibration and regulation are very important in biology and are things any biologist should know about. Few do, of course.

A good place to start would be control theory on an intuitive level (i.e. before the complex numbers and the linear algebra). Talk to an engineer or check the wikipedia page.