Editors' Choice

Science  14 Oct 2016:
Vol. 354, Issue 6309, pp. 193
  1. Cell Division

    Crossing over to let go

    1. Guy Riddihough

    Crossovers help ensure that chromosomes segregate properly during meiosis.

    CREDIT: TIM VERNON/SCIENCE SOURCE

    Eukaryotic cells package their genomes into chromosomes, and when they divide, they deliver a complete set of chromosomes to each daughter cell. During meiosis (the cell division that creates germ cells), “DNA crossovers” hold homologous chromosomes together, which helps to ensure that they segregate correctly. Studying worms, Machovina et al. show that cells carefully monitor how these crossovers form. The presence of a crossover strengthens the protein complex that holds the two homologous chromosomes together. Failure to form a crossover weakens the protein complex until a crossover can occur.

    Curr. Biol. 10.1016/j.cub.2016.09.007 (2016).

  2. Brain Research

    Encoding false memories

    1. Peter Stern

    A unique human ability is the use of concepts that confer meaning in an abstract way. Despite the importance of conceptual knowledge for human cognition, scientists know little about the underlying neural mechanisms and structures. To better understand this process, Chadwick et al. scanned people's brains while they performed tasks known to cause false memories in a statistically predictable manner. They found that activity in the temporal pole, a region known as the “semantic hub” of the brain, could predict false memories. Moreover, individuals had distinct patterns of activity in this region, allowing the authors to predict specific memory errors in different subjects.

    Proc. Natl. Acad. Sci. U.S.A. 113, 10180 (2016).

  3. Autism Genomics

    Mosaic mutations in autism

    1. Laura M. Zahn

    Spontaneously arising de novo mutations are likely important contributors to the development of autism spectrum disorder (ASD) in affected individuals. Genetic mutation is an ongoing process. Although people inherit deleterious mutations from their parents, some can also arise postzygotically. Freed and Pevsner examined the degree to which postzygotic mutations contribute to ASD and found mosaic mutations, which occur postzygotically and only in a subset of a person's cells, in multiple genes implicated in ASD. On the basis of these observations, the authors estimated that mosaic mutations contribute to ∼5% of ASD diagnoses.

    PLOS Genet. 12, e1006245 (2016).

  4. Materials Science

    Push both ways to avoid a jam

    1. Marc S. Lavine

    When subject to shear flow, dense solutions of particles can jam together, dramatically increasing their viscosity. This can be helpful, such as in traction control systems, or it can be harmful, such as when a pipe becomes blocked during industrial processing. During flow, when the stress exceeds a critical value, particle interactions transform from lubricated to frictional, leading to transient force chains. Lin et al. show that this particle jamming transition is not fixed by the suspension formulation, but rather can be dynamically tuned: The trick is to introduce perturbations in an orthogonal direction to break up the force chains before they fully form. This enables control of the thickening viscosity over two orders of magnitude.

    Proc. Nat. Acad. Sci. 10.1073/pnas.1608348113 (2016).

  5. Carbon Footprint

    Mapping the impact of nations

    1. Nicholas S. Wigginton
    CREDIT: KANEMOTO ET AL., ENVIRONMENTAL SCIENCE & TECHNOLOGY (1 SEPTEMBER 2016) ©ACS

    The consumption and production of goods have a much wider impact than simply on the country or region where they are used or made. For example, the carbon footprint of global supply chains varies within and between countries, mostly according to economics and trade. Using spatial modeling, Kanemoto et al.linked carbon emissions maps to industrial activity across 187 countries from 1970 to 2008. The carbon footprints of developed countries such as the United States have become more globally diffuse over that time span, whereas hotspots of carbon footprints in rapidly developing economies such as India and China are located in their expanding urban areas.

    Environ. Sci. Technol. 10.1021/acs.est.6b03227 (2016).

  6. Structural Biology

    Dynamics of a protein knot

    1. Valda Vinson

    A small fraction of proteins have an unusual conformation in which the backbone forms a knot. An example is a bacterial enzyme, TrmD, that transfers a methyl group from S-adenosyl methionine (AdoMet) to a guanine nucleotide that is conserved in many transfer RNAs (tRNAs). This methyl transfer ensures accurate protein synthesis. Christian et al. combine structural, mutagenesis, and computational studies to examine the role of the protein knot in catalysis. They show that the knot binds the AdoMet in a bent conformation oriented for methyl transfer. Despite its constrained topology, the knot undergoes complex dynamics that couple AdoMet binding to tRNA binding and facilitate catalysis.

    Knotted proteins such as TrmD can undergo complex internal movements.

    CREDIT: CHRISTIAN ET AL., NATURE STRUCTURAL & MOLECULAR BIOLOGY (29 AUGUST 2016) ©NATURE

    Nat. Struct. Mol. Biol. 10.1038/nsmb.3282 (2016).

  7. Machine Learning

    A quantum leap for machine learning

    1. Ian S. Osborne

    Machine learning is a field of artificial intelligence in which a set of algorithms and training sets can teach a computer to learn and then to independently perform a particular set of tasks. These thinking machines are now at the stage of pitting their wits against human opponents, and can, in recent instances, outperform and beat the very best opponents in complex gaming scenarios. So far, the implementations have been classical algorithms running on classical computers. Dunjko et al. show that certain aspects of quantum information processing, when applied to machine learning, could provide a quantum advantage. They identify certain traits in machine learning for which quantum processing algorithms can improve learning efficiency, and they show that quantum processing could provide faster solutions when applied to particular categories of machine learning problems.

    Phys. Rev. Lett. 117, 130501 (2016).

Navigate This Article