Avoiding backtracking enables spectral analysis of sparse networks
Spectral density of the nonbacktracking matrix used for clustering.
Researchers who study networks, whether social, biological, or technological, analyze datasets for clustering to detect communities of related nodes or boundaries between dissimilar groups. One common, computationally efficient approach uses a class of techniques known as spectral algorithms, but applying the standard versions of these algorithms to sparse networks often yields suboptimal results. To address this problem, Florent Krzakala et al. (pp. 20935–20940) present a technique that uses a so-called “nonbacktracking” matrix to encode sparse data prior to analysis and thus redeems the efficacy of spectral algorithms for many popular models of network behavior. Supported by theoretical arguments and analogies from probability theory, statistical physics, and the theory of random matrices, the authors demonstrate the advantages of the nonbacktracking matrix by applying it to a number of real-world networks, including several established benchmarks for community detection. Furthermore, the technique was found to be applicable to a recently discovered phase transition for community detectability in cluster models. In addition to network theory, the nonbacktracking matrix can be generalized and applied to other types of sparse data, according to the authors. — T.J.
Effect of oxytocin on children with autism
Researchers have shown that oxytocin, administered via a nasal spray, enhances social aptitudes in children with autism. The brain mechanisms underlying this effect, however, have remained unclear. To illuminate the effect of the hormone, Ilanit Gordon et al. (pp. 20953–20958) monitored the brain activity of children with autism by using fMRI. After receiving a dose of intranasal oxytocin, the children performed social and nonsocial tasks, namely classifying a person’s mental state based solely on an image of their eyes (a social judgment), and categorized images of automobiles (a nonsocial judgment). Brain centers associated with reward, social perception, and emotional awareness showed greater activity during social tasks with oxytocin than without the hormone. The authors also observed a decrease in activity in some brain regions during nonsocial tasks, suggesting that oxytocin may attune the brain to the difference between social and non-social judgments. Further, salivary oxytocin concentrations during the experiment were positively correlated with increased social brain activity. The authors suggest that oxytocin may help make social interactions more rewarding, because the brain regions activated during the experiment have been frequently identified as hypoactive in people with autism. The findings may inform the future use of oxytocin in autism treatment, according to the authors. — P.G.
Software system growth models evolution of mammalian body size
Common patterns in the dynamics of software and mammal body size.
In software development, “packages” are bundled files that comprise the whole system. Marco Gherardi et al. (pp. 21054–21058) investigated the dynamics of package growth in Ubuntu, an open source software-based Linux operating system that has grown from a few hundred to tens of thousands of packages, as a model for how large and complex systems evolve. According to the authors, two elementary laws can describe package size evolution in a Linux operating system. Growth, the authors report, is not driven by simple addition and subtraction but follows a pattern dominated by large changes in scale. Secondly, the authors report, increases in sizes between consecutive releases are limited by a so-called “soft bound,” which depends on the initial size of a package. The authors also demonstrate that a simple, stochastic model implementing these two laws accurately predicts the package sizes of the latest Ubuntu release. According to the authors, the analytical approach shows that an evolutionary mechanism similar to the growth of the software system might underlie the distribution of body sizes in mammals. — T.J.
Post-9/11 impacts on political behavior of victims’ family and friends
Following a sweeping tragic event, victims’ families and friends often find their lives changed in many ways; some become more politically active than before, and may sustain their political activity long after the original tragedy. To understand the political impact of the September 11 attacks, Eitan Hersh (pp. 20959–20963) investigated public records to assess how the political behavior of victims’ families and neighbors changed following the attack. The author compared the affected population to a group of families and friends of so-called control victims, residents of the New York metropolitan area whose demographics and prior political activity matched those of September 11 victims. The author examined publicly available records of campaign contributions, election participation, and political party affiliation, and found that victims’ families increased campaign contributions and participated more in primary and general elections in the decade after September 11 relative to the comparison group, an effect seen to a lesser extent in victims’ neighbors. The authors report that 17% of the affected population changed their political party affiliation after September 11, on average shifting away from Democratic and toward Republican affiliations. The results suggest that a large-scale traumatic event such as September 11 may spur long-lasting political involvement and a shift of political ideologies by those most deeply affected, according to the author. — P.G.
Visible light alters yeast metabolic rhythms
Sodium azide (Top), an inhibitor of electron transport, and visible light (Bottom) shorten the period and reduce the amplitude of the YRO.
The budding yeast Saccharomyces cerevisiae lacks the photoreceptors such as cryptochromes, phytochromes, and rhodopsins that enable many organisms to sense light and prepare for its effects via oscillatory timekeeping mechanisms. However, under certain conditions, S. cerevisiae exhibit respiratory oscillations characterized by 1–6-hour rhythms of oxygen consumption, metabolite production, cell division, and gene expression. James Robertson et al. (pp. 21130–21135) investigated the effects of visible light on yeast respiratory oscillations (YRO) by exposing yeast grown in total darkness to 12-hour intervals of visible light. The authors found that blue and green wavelengths shortened the period and reduced the amplitude of the YRO, whereas red light had little or no effect. The wavelengths of the blue light treatments corresponded to the peak absorption spectra for cytochrome oxidase, an integral part of the mitochondrial electron transport chain, the authors report. The authors also found that inhibition of electron transport mimicked the effects of visible light on YRO, suggesting that light may modulate YRO by inhibiting the activity of light-absorbing cytochromes involved in electron transport. The findings demonstrate that visible light modulates cellular rhythmicity and metabolism in S. cerevisiae, and suggest that exposure of cells to light in nature or in fluorescence microscopy should not be assumed to be innocuous, according to the authors. — N.Z.
3D culture system for large-scale expansion and differentiation of pluripotent stem cells
hPSCs cultured in the 3D hydrogel culture system for 35 passages uniformly express Nanog and Oct4, two markers of pluripotency.
Human pluripotent stem cells (hPSCs), which can self-renew indefinitely and develop into many different cell types, have promising biomedical applications, such as cell replacement therapies, tissue and organ engineering, and high throughput drug and toxicity screening. However, such applications require large numbers of high-quality cells, and the expansion and differentiation of hPSCs on a large scale remains challenging. Yuguo Lei and David Schaffer (pp. E5039–E5048) developed an efficient and scalable 3D cell culture system for hPSC production. The authors added hPSCs to a synthetic polymeric solution that is liquid at low temperatures but solidifies into an elastic hydrogel when warmed, providing a 3D environment that promoted cell growth and prevented cell aggregation. After optimizing the culture conditions, the authors found that the cells displayed a high expansion rate of approximately 7–20-fold after 4–6 days in culture and 1072-fold after 280 days in culture, and the majority of cells continued to express molecular markers of pluripotency. By replacing the expansion medium with differentiation medium, the authors differentiated the hPSCs into neural progenitors, endoderm progenitors, beating cardiomyocytes, and midbrain dopaminergic progenitors. According to the authors, the 3D hydrogel culture system may aid efficient, economical, and reproducible cell production for future applications of hPSCs. — N.Z.