Soil Formation of "Atlantic Rankers" from NW Spain—A High Resolution Aluminium and Iron Fractionation Study

Atlantic rankers belong to the group of ＂cryptopodzolic rankers＂, which are ubiquitous in the mountainous cool/temperate humid regions of Western Europe. The rankers of Galicia （NW Spain） formed by thousands of years of colluviation. The preponderance of Al-stabilised organic matter （OM） masks the horizonation and polycyclic character （i.e., stratification） of these soils. Cryptopodzolic rankers are generally thought to be the outcome of podzolisation. This soil type is part of the recent discussion on how to classify soils developed from nonvolcanic parent material having andic properties. To better understand the formation processes of these soils, the Al and Fe fractionation of four typical Atlantic rankers were studied by selective dissolution in acid NH4-oxalate, Na-pyrophosphate and the chlorides of K, La and Cu. A high-resolution sampling approach allowed us to investigate the soils in greater detail than simply sampling by horizon. The rankers studied display a distribution of Fe- and AI-OM complexes that is typical of cryptopodzolic soils. However, these organomineral associations were probably immobile due to the high Al saturation. We argue that the soils owe their characteristic chemical status to external factors rather than to translocation of organomineral associations： variations in AI-OM concentrations could be linked to changes in weathering/leaching intensity and colluviation rates caused by anthropogenic disturbances or changes in regional climate regime.

基于Nano LC-MS/MS的板蓝根内源性肽鉴定及活性虚拟筛选研究

目的:研究板蓝根内源性肽的氨基酸序列及潜在生物活性。方法:采用固相萃取法分离纯化板蓝根的内源性肽,结合Nano LC-MS/MS技术,以de novo法进行氨基酸序列的表征,最后通过Peptide ranker平台,结合DBAASP和CAMPR3平台进行肽段活性虚拟筛选。结果:ALC>50%的肽段共鉴别出195个,>90%的有5个,分别是:Leu-Phe-Asp-Phe、Phe-Thr-Leu-Ser-Arg、Leu-Val-Lys-Pro-Glu-Pro、Acetylation-Lys-Phe-Asp-Phe、Leu-Pro-Glu-Pro-Glu-Pro。选择ALC>60%的131条肽段进行活性预测评价,Peptide Ranker得分>0.80的共有17条肽,与DBAASP和CAMPR3筛选结果具有较高的重合性。结论:板蓝根内源性肽有一定的潜在的抗菌活性,还需进一步验证确认。

基于BigQuant大数据平台的股票投资策略开发

文中基于BigQuant平台股票投顾系统,利用StockRanker算法和回测机制,对中国股市在整个样本期2010年1月1日至2019年2月5日具有正常交易的全部A股中剔除沪深300指数成份股后的1848只股票特征数据进行分析,给出最有投资价值的股票排序,从而为具有不同风险偏好的投资者提供智能化、个性化的资产配置建议。文中基于标准指数基金中证500指数,通过策略判断,用业绩优异的非成份股代替业绩较差的成份股,开发出一款D产品,它具有超越标准指数基金更好、更稳定的投资收益。

关系数据库关键字查询结果排序方法

为了提高传统排序方法的排序准确率及效率,本文提出一种列表级的并行学习排序方法,用于关系数据库关键字查询结果的排序。采用一种列表级学习排序算法并在此基础上引入并行框架,使训练过程能够在多个节点上并行执行,有效地减少训练时间;并且提出一种基于贪婪搜索算法和排序性能的弱排序器分层构建策略,提高排序模型的训练效率和有效性;分别在公开数据集IMDB和Wikipedia上进行实验。实验结果表明:与传统的排序算法相比,本文方法具有较高的训练效率及排序准确率。

造船新工法在10.5万载重吨原油船上的应用

阐述了上海外高桥造船有限公司在10.5万dwt原油船的设计策划、建造策划以及生产设计过程中推行的一些设计理念和建造工法,旨在合理利用有限资源,实现船厂绩效的全面提升。