Automated design of multiple-class piecewise linear classifiers

A new method and a supporting theorem for designing multiple-class piecewise linear classifiers are described. The method involves the cutting of straight line segments joining pairs of opposed points (i.e., points from distinct classes) ind-dimensional space. We refer to such straight line segments aslinks. We show how nearly to minimize the number of hyperplanes required to cut all of these links, thereby yielding a near-Bayes-optimal decision surface regardless of the number of classes, and we describe the underlying theory. This method does not require parameters to be specified by users — an improvement over earlier methods. Experiments on multiple-class data obtained from ship images show that classifiers designed by this method yield approximately the same error rate as the bestk-nearest neighbor rule, while providing faster decisions.This research was supported in part by the Army Research Office under grant DAAG29-84-K-0208 and in part by the University of California MICRO Program. We thank R. W. Doucette of the U.S. Naval Weapons Center and R. D. Holben of Ford Aerospace Corporation for providing the ship images in our experiments.     

The afterglow of GRB 050709 and the nature of the short-hard gamma-ray bursts

The final chapter in the long-standing mystery of the gamma-ray bursts (GRBs) centres on the origin of the short-hard class of bursts, which are suspected on theoretical grounds to result from the coalescence of neutron-star or black-hole binary systems. Numerous searches for the afterglows of short-hard bursts have been made, galvanized by the revolution in our understanding of long-duration GRBs that followed the discovery in 1997 of their broadband (X-ray, optical and radio) afterglow emission. Here we present the discovery of the X-ray afterglow of a short-hard burst, GRB 050709, whose accurate position allows us to associate it unambiguously with a star-forming galaxy at redshift z = 0.160, and whose optical lightcurve definitively excludes a supernova association. Together with results from three other recent short-hard bursts, this suggests that short-hard bursts release much less energy than the long-duration GRBs. Models requiring young stellar populations, such as magnetars and collapsars, are ruled out, while coalescing degenerate binaries remain the most promising progenitor candidates.     

Aspects of gene polymorphisms in cerebral infarction: inflammatory cytokines

During the last decade, a growing corpus of evidence has indicated an important role of inflammatory cytokines in the pathogenesis of cerebral lesion following stroke. Recent data suggest that genetics may in turn contribute to modulating the effects of inflammatory cytokines on cerebral infarction (CI). This paper reviews the physiologic characteristics of major inflammatory cytokines and recent research developments related to cell biology and pathobiology in CI. In particular, this review focuses on the genetic aspects of inflammatory cytokines and their implications in CI.Received 22 June 2004; received after revision 11 November 2004; accepted 16 December 2004     

Sipa1 is a candidate for underlying the metastasis efficiency modifier locus Mtes1

We previously identified loci in the mouse genome that substantially influence the metastatic efficiency of mammary tumors. Here, we present data supporting the idea that the signal transduction molecule, Sipa1, is a candidate for underlying the metastasis efficiency modifier locus Mtes1. Analysis of candidate genes identified a nonsynonymous amino acid polymorphism in Sipa1 that affects the Sipa1 Rap-GAP function. Spontaneous metastasis assays using cells ectopically expressing Sipa1 or cells with knocked-down Sipa1 expression showed that metastatic capacity was correlated with cellular Sipa1 levels. We examined human expression data and found that they were consistent with the idea that Sipa1 concentration has a role in metastasis. Taken together, these data suggest that the Sipa1 polymorphism is one of the genetic polymorphisms underlying the Mtes1 locus. This report is also the first demonstration, to our knowledge, of a constitutional genetic polymorphism affecting tumor metastasis.     

Regulation of PDGF signalling and vascular remodelling by peroxiredoxin II

Platelet-derived growth factor (PDGF) is a potent mitogenic and migratory factor that regulates the tyrosine phosphorylation of a variety of signalling proteins via intracellular production of H2O2 (refs 1, 2-3). Mammalian 2-Cys peroxiredoxin type II (Prx II; gene symbol Prdx2) is a cellular peroxidase that eliminates endogenous H2O2 produced in response to growth factors such as PDGF and epidermal growth factor; however, its involvement in growth factor signalling is largely unknown. Here we show that Prx II is a negative regulator of PDGF signalling. Prx II deficiency results in increased production of H2O2, enhanced activation of PDGF receptor (PDGFR) and phospholipase Cgamma1, and subsequently increased cell proliferation and migration in response to PDGF. These responses are suppressed by expression of wild-type Prx II, but not an inactive mutant. Notably, Prx II is recruited to PDGFR upon PDGF stimulation, and suppresses protein tyrosine phosphatase inactivation. Prx II also leads to the suppression of PDGFR activation in primary culture and a murine restenosis model, including PDGF-dependent neointimal thickening of vascular smooth muscle cells. These results demonstrate a localized role for endogenous H2O2 in PDGF signalling, and indicate a biological function of Prx II in cardiovascular disease.     

The DNA sequence of human chromosome 21

Chromosome 21 is the smallest human autosome. An extra copy of chromosome 21 causes Down syndrome, the most frequent genetic cause of significant mental retardation, which affects up to 1 in 700 live births. Several anonymous loci for monogenic disorders and predispositions for common complex disorders have also been mapped to this chromosome, and loss of heterozygosity has been observed in regions associated with solid tumours. Here we report the sequence and gene catalogue of the long arm of chromosome 21. We have sequenced 33,546,361 base pairs (bp) of DNA with very high accuracy, the largest contig being 25,491,867 bp. Only three small clone gaps and seven sequencing gaps remain, comprising about 100 kilobases. Thus, we achieved 99.7% coverage of 21q. We also sequenced 281,116 bp from the short arm. The structural features identified include duplications that are probably involved in chromosomal abnormalities and repeat structures in the telomeric and pericentromeric regions. Analysis of the chromosome revealed 127 known genes, 98 predicted genes and 59 pseudogenes.     

Direct oxidation of hydrocarbons in a solid-oxide fuel cell

The direct electrochemical oxidation of dry hydrocarbon fuels to generate electrical power has the potential to accelerate substantially the use of fuel cells in transportation and distributed-power applications. Most fuel-cell research has involved the use of hydrogen as the fuel, although the practical generation and storage of hydrogen remains an important technological hurdle. Methane has been successfully oxidized electrochemically, but the susceptibility to carbon formation from other hydrocarbons that may be present or poor power densities have prevented the application of this simple fuel in practical applications. Here we report the direct, electrochemical oxidation of various hydrocarbons (methane, ethane, 1-butene, n-butane and toluene) using a solid-oxide fuel cell at 973 and 1,073 K with a composite anode of copper and ceria (or samaria-doped ceria). We demonstrate that the final products of the oxidation are CO2 and water, and that reasonable power densities can be achieved. The observation that a solid-oxide fuel cell can be operated on dry hydrocarbons, including liquid fuels, without reforming, suggests that this type of fuel cell could provide an alternative to hydrogen-based fuel-cell technologies.     

Methylation of the CDH1 promoter as the second genetic hit in hereditary diffuse gastric cancer

Aberrant promoter methylation and the associated loss of gene expression is a common accompaniment of human cancers. Nonetheless, it has been challenging to demonstrate in any given tumour that methylation of a specific gene was causal and not consequent to malignant transformation. In this regard, our attention was drawn to the genesis of gastric cancers in individuals with hereditary diffuse gastric cancer (HDGC). These individuals harbour germline mutations in the gene encoding E-cadherin, CDH1, but their cancers have consistently demonstrated absence of loss of heterozygosity at the CDH1 locus. These findings suggested the hypothesis that CDH1 promoter methylation might function as the 'second genetic hit' in the genesis of these cancers.