Xingzhu Chen

Copyright © 2016 is held by owner/author.In this paper we present optimizations that use DVFS mechanisms to reduce the total energy usage in scientific applications. Our main insight is that noise is intrinsic to large scale parallel executions and it appears whenever shared resources are contended. The presence of noise allows us to identify and manipulate any program regions amenable to DVFS. When compared to previous energy optimizations that make per core decisions using predictions of the running time, our scheme uses a qualitative approach to recognize the signature of executions amenable to DVFS. By recognizing the "shape of variability" we can optimize codes with highly dynamic behavior, which pose challenges to all existing DVFS techniques. We validate our approach using offline and online analyses for one-sided and two-sided communication paradigms. We have applied our methods to NWChem, and we show best case improvements in energy use of 12% at no loss in performance when using online optimizations running on 720 Haswell cores with one-sided communication. With NWChem on MPI two-sided and offline analysis, capturing the initialization, we find energy savings of up to 20%, with less than 1% performance cost.

Knowledge of three-dimensional structures of each individual particles of asymmetric and flexible proteins is essential in understanding those proteins' functions; but their structures are difficult to determine. Electron tomography provides a tool for imaging a single and unique biological object from a series of tilted angles, but it is challenging to image a single protein for three-dimensional reconstruction due to the imperfect mechanical control capability of the specimen goniometer under both a medium to high magnification and an optimized beam coherence condition. Here, we report a fully mechanical control method for automating ET data acquisition without using beam tilt/shift processes. This method could reduce the accumulation of beam tilt/shift that used to compensate the error from the mechanical control, but downgraded the beam coherence. Our method was developed by minimizing the error of the target object center during the tilting process through a closed-loop proportional-integral control algorithm. The validations by both negative staining and cryo-electron microscopy suggest that this method has a comparable capability to other ET methods in tracking target proteins while maintaining optimized beam coherence conditions for imaging.

The Yr5 gene confers resistance to all races of the wheat stripe rust pathogen [Puccinia striiformis Westend. f. sp. tritici Eriks. ] identified so far in the USA. Cosegregating resistance gene analog polymorphism markers for Yr5 are available but their use requires skills in polyacrylamide gel electrophoresis and may not be polymorphic across various varieties. To develop better markers to be used in marker-assisted selection for the Yr5 resistance, sequence tagged site primers were designed on the basis of the sequences of RGAP markers Xwgp-18 from the spring wheat 'Avocet Susceptible' and Xwgp-17 from the Yr5 near isogenic line in the AVS background carrying the Yr5 gene from T. aestivum subsp. spelta Thell. cv. Album. Three sets of STS markers were developed to amplify a region including a polymorphic 6-base pairs insertion-deletion. The cosegregation of the STS markers with Yr5 was confirmed with 114 BC 7:F3 lines developed from the cross between AVS and TSA. The STS markers worked well in five out of 17 non-Yr5 wheat varieties, but the remaining varieties had a similar size of fragment to the Yr5 marker. Because the codominant STS markers were based on a 6-bp indel, they could not be separated by agarose gel electrophoresis. Cleaved amplified polymorphic sequence markers were then developed on the basis of a DpnII restriction site that is present in all non-Yr5 varieties and absent in the Yr5 NIL. The CAPS markers for the Yr5 NIL and non-Yr5 varieties can be separated by agarose gel electrophoresis. The codominant STS markers are easier to score than the original RGAP markers. The CAPS markers are not only easier to score, but also can be used in crosses of an Yr5 donor with a much wider range of wheat germplasms. These markers should be valuable tools to accelerate the introgression of Yr5 into commercial cultivars and to combine Yr5 with other genes for durable resistance to stripe rust.

© 2015 Macmillan Publishers Limited All rights reserved. Genome-wide gene expression measurements have enabled comprehensive studies that integrate the changes of gene expression and phenotypic information to uncover their novel associations. Here we reported the association analysis between psychophysical phenotypes and genome-wide gene expression changes in human adaptation to one of the most extreme climates on Earth, the Antarctic Dome Argus. Dome A is the highest ice feature in Antarctica, and may be the coldest, driest and windiest location on earth. It is considered unapproachable due to its hostile environment. In 2007, a Chinese team of 17 male explorers made the expedition to Dome A for scientific investigation. Overall, 133 psychophysical phenotypes were recorded, and genome-wide gene expression profiles from the blood samples of the explorers were measured before their departure and upon their arrival at Dome A. We found that mood disturbances, including tension, depression, anger and fatigue, had a strong, positive, linear relationship with the level of a male sex hormone, testosterone, using the Pearson correlation coefficient analysis. We also demonstrated that significantly lowest-level Gene Ontology groups in changes of gene expression in blood cells with erythrocyte removal were consistent with the adaptation of the psychophysical characteristics. Interestingly, we discovered a list of genes that were strongly related to significant phenotypes using phenotype and gene expression PCC analysis. Importantly, among the 70 genes that were identified, most were significantly related to mood disturbances, where 42 genes have been reported in the literature mining, suggesting that the other 28 genes were likely novel genes involved in the mood disturbance mechanism. Taken together, our association analysis provides a reliable method to uncover novel genes and mechanisms related to phenotypes, although further studies are needed.