Computer Simulation: Monte Carlo races

Phys. Rev. Lett. 94, 170201 (2005)

Researchers don't like to admit defeat, but physicists grappling with a certain class of Monte Carlo computer simulations should be told that there's no easy way to speed them up.

Monte Carlo simulations are often used to look for equilibrium states in systems with many interacting components. But when those components are fermionic particles — electrons, for example — the time taken to find reliable solutions increases exponentially with the number of particles. This hampers simulations for calculating electronic structures, especially for systems (such as high-temperature superconductors) in which the electrons interact strongly.

Now, Swiss physicists Matthias Troyer and Uwe-Jens Wiese have proved that searches for a faster Monte Carlo method for fermions are doomed. They show that the problem belongs to the class of ‘nondeterministic-polynomial’ hard computational problems, for which no general short cuts exist.

Materials: Stiff stuff

Nature Mater. doi:10.1038/nmat1392 (2005)

Cells resting on stiff surfaces are better at taking up genes delivered by non-viral carriers; they also express these genes more readily.

David Mooney's team at Harvard University delivered DNA strands to mouse cells using polyethyleneimine as a vector, monitoring the gene's delivery by a change in fluorescence as it detached from the vector. The cells were attached to various hydrogels, whose rigidity could be varied easily.

Although gene expression was not affected by the gels' surface chemistry, it increased markedly for more rigid gels. Exploiting this effect might boost the rate of gene transfection and expression from non-viral vectors.

Cell Biology: Achilles' heel for HIV

Immunity 22, 607–619 (2005)

RNA silencing — already used by plants and invertebrates to fend off viral infection — is a tool proposed for treating HIV-1 infection in humans. At first, short interfering RNAs can inhibit HIV-1 replication in cell culture. But over time this inhibitory effect wanes, probably because HIV-1 mutates so rapidly. In new work, however, Kuan-Teh Jeang and colleagues, at the National Institute of Allergy and Infectious Diseases, identify an HIV-1 sequence that is protected from short interfering RNA attacks by a protein. This suggests that it cannot be altered for functional reasons. If researchers can find a way to evade the protector, the sequence may be a viable target for RNA interference therapy.

Biology: Stress ploys

PLoS Biol. doi: 10.1371/journal.pbio.0030176 (2005)

Bacteria induce mutations in their genome when put under stress. This finding might offer a way to combat the emergence of antibiotic resistance.

Research led by Floyd Romesberg at the Scripps Research Institute in La Jolla, California, revealed that the SOS response triggered by some antibiotics actively induces mutations in the bacterial genome. When the team interfered with one of the enzymes involved in this process, Escherichia coli did not evolve resistance to important quinolone and rifamycin antibiotics.

Biotechnology: Souped-up vectors

Nature Biotechnol. doi:10.1038/nbt1094 (2005)

Plants can be useful factories for manufacturing foreign proteins. But permanently modified crops carry environmental risks, and transient infection with a transgene is too inefficient to be of industrial use. Or so it was thought.

Researchers from Icon Genetics in Germany have shown how to modify a viral vector carrying the appropriate transgene to make it more potent. They made the vector, derived from the tobacco mosaic virus, more like the plant RNA by adding non-coding segments called introns. And they deleted segments that they suspected the plant nucleus would find hard to process. This increased export of transgenic RNA from infected nuclei, cutting 1,000-fold the number of vectors needed to make a cell produce the protein. The technique worked in a variety of different plants, including petunias, cucumbers and spinach.

Cell Biology: Pore relation

Genes Dev. 19, 1188–1198 (2005)

A controversial theory suggesting that genes relocate to the periphery of the nucleus when they are transcribed is supported by studies in yeast cells. A team led by Pamela Silver at Harvard Medical School tracked the location of the many genes whose expression is enhanced when yeast cells are exposed to a mating pheromone. The researchers show that most of the activated genes cluster around nuclear pores — openings through which molecules pass between the nucleus and cytoplasm. They also use high-resolution mapping to investigate how the chromosome hosting the genes changes its spatial orientation to put the genes in place.

Nanotechnology: Into the groove

Nano Lett. doi:1021/nl050405n (2005)

Nanoscale ice sculptures could provide templates for making transistors and miniature machines, suggests a team at Harvard University. The researchers used electron-beam lithography to carve grooves through a 20-nanometre-thick layer of ice on a silicon wafer, and then applied a coating of chromium. A rinse in isopropyl alcohol removed the icy template and its metal overlayer, leaving strips of chromium less than 20 nanometres wide.

Unlike the polymer mask materials used to pattern silicon today, ice doesn't need spinning or baking on to the substrate. And it can be removed by sublimation, eliminating the need for solvents, although this process leaves stray chromium flakes.

Cancer: Healing hijacked

Cell 121, 335–348 (2005)

The body's wound-healing process is subverted by breast-cancer tumours to help them grow and build a network of blood vessels, according to a study led by Robert Weinberg of the Whitehead Institute in Cambridge, Massachusetts.

Credit: VVG/SPL

Tumours are already known to contain fibroblasts — cells usually found only in inflamed or regenerating tissue that help build connective tissue. The team found that in invasive breast tumours, these fibroblasts make a protein called SDF-1 that encourages cancerous cells to grow. The protein also summons cells to generate the blood vessels that tumours need.

Hydrodynamics: Skipping stones

Phys. Rev. Lett. 94, 174501 (2005)

Credit: THE IMAGE BANK/GETTY IMAGES

Stones, as many time-wasters know, skim across water when thrown the right way. But how to get the most bounces? Experiments reported in Nature (427, 29; 2004) last year showed that a flat stone should be tilted at 20° to the water's surface for optimal skipping. Now, Shin-ichiro Nagahiro and Yoshinori Hayakawa of Tohuku University in Japan back up this finding with theory. They model the impact of a stone on water using smoothed particle hydrodynamics, and calculate the angle where the velocity required for an onward bounce is minimized. Happily, for the peaceful coexistence of theorists and experimentalists at the waterside, they find it is always around 20°.

Chemistry: Complex proteins

J. Am. Chem. Soc. doi: 10.1021/ja050304r (2005)

The functions of some proteins involved in cell signalling and the immune response are altered when small molecules, such as lipids and carbohydrates, are attached to an amino-acid side chain. Obtaining large amounts of these modified proteins for study has proved difficult because bacteria cannot make them.

Credit: OXFORD SCIENTIFIC

A possible solution, offered by chemists from the University of Illinois in Urbana, involves an artificial amino acid containing aziridine, a three-membered ring. Thiols, the sulphur equivalent of alcohols, selectively react with this ring, forming a new bond. So far, the team has attached small thiol-containing molecules to peptides, but other methods may extend the technique's scope to longer proteins.

Climate change: Fish scarper up north

Science doi:10.1126/science.1111322 (2005)

Climate change is forcing fish in the North Sea to shift to cooler waters, concludes a team of researchers led by Allison Perry and John Reynolds of the University of East Anglia, UK. The group looked at how the distribution of 36 bottom-dwelling species has changed over the past 25 years. Nearly two-thirds of them — including commercially exploited fish such as the Atlantic cod (Gadus morhua, pictured) and species not targeted by fisheries, such as the scaldfish (Arnogossus laterna) — have moved north or sunk to deeper waters.

Journal Club

Peter Dayan

University College London, UK

The director of UCL's Gatsby Computational Neuroscience Unit recommends some gainful reading.

Who of us lacking an anorak could admire an article with 60 equations and 15 instances of “piecewise linear”? But skip this paper — in the press at the International Journal of Bifurcation and Chaos — and you'll miss an important, if obscure, treat. It addresses a venerable question, which has long troubled me, in the neuroscience and psychology of decision-making: the role of neuromodulators.

Neuromodulators are an important class of transmitter. From a biophysical perspective, they seem to regulate general properties such as the excitability or ‘gain’ of their target neurons. Thus, many see them merely as widespread irrigants (or irritants, as our less charitable colleagues would have it).

But behavioural neuroscience studies suggest that neuromodulators can play a key role at specific times in decision-making tasks. The transmitters regulate competition between populations of neurons that represent choices. This allows a subject to integrate noisy sensory information with past experience about rewards to make almost optimal decisions.

Previous attempts to reconcile the general function with the specific were a bit too heuristic to convince me. I buried my head in the sand or muttered jeremiads — complaining that the necessary tests were impossible.

However, this paper squares the biophysics with the behaviour. The researchers use a model to show that changing the gain of neurons is precisely the right thing to do to achieve optimal decisions. Along the way, they link classical and modern suggestions about the mechanisms of decision-making, an area under intense behavioural and electrophysiological scrutiny.

So buy your anorak, and let this theory reign.

http://www.math.nyu.edu/~ebrown/papers/simple_networks.pdf