Modeling data with structural and temporal correlation using lower level and higher level multilevel models

Novel imaging techniques are playing an increasingly important role in drug development, providing insight into the mechanism of action of new chemical entities. The data sets obtained by these methods can be large with complex inter-relationships, but the most appropriate statistical analysis for handling this data is often uncertain - precisely because of the exploratory nature of the way the data are collected. We present an example from a clinical trial using magnetic resonance imaging to assess changes in atherosclerotic plaques following treatment with a tool compound with established clinical benefit. We compared two specific approaches to handle the correlations due to physical location and repeated measurements: two-level and four-level multilevel models. The two methods identified similar structural variables, but higher level multilevel models had the advantage of explaining a greater proportion of variation, and the modeling assumptions appeared to be better satisfied.

Correlation shifts and real estate portfolio management

The success of any diversification strategy depends upon the quality of the estimated correlation between assets. It is well known, however, that there is a tendency for the average correlation among assets to increase when the market falls and vice-versa. Thus, assuming that the correlation between assets is a constant over time seems unrealistic. Nonetheless, these changes in the correlation structure as a consequence of changes in the market’s return suggests that correlation shifts can be modelled as a function of the market return. This is the idea behind the model of Spurgin et al (2000), which models the beta or systematic risk, of the asset as a function of the returns in the market. This is an approach that offers particular attractions to fund managers as it suggest ways by which they can adjust their portfolios to benefit from changes in overall market conditions. In this paper the Spurgin et al (2000) model is applied to 31 real estate market segments in the UK using monthly data over the period 1987:1 to 2000:12. The results show that a number of market segments display significant negative correlation shifts, while others show significantly positive correlation shifts. Using this information fund managers can make strategic and tactical portfolio allocation decisions based on expectations of market volatility alone and so help them achieve greater portfolio performance overall and especially during different phases of the real estate cycle.

Asset allocation, cross-class correlation and the structure of property returns

Practical applications of portfolio optimisation tend to proceed on a “top down” basis where funds are allocated first at asset class level (between, say, bonds, cash, equities and real estate) and then, progressively, at sub-class level (within property to sectors, office, retail, industrial for example). While there are organisational benefits from such an approach, it can potentially lead to sub-optimal allocations when compared to a “global” or “side-by-side” optimisation. This will occur where there are correlations between sub-classes across the asset divide that are masked in aggregation – between, for instance, City offices and the performance of financial services stocks. This paper explores such sub-class linkages using UK monthly stock and property data. Exploratory analysis using clustering procedures and factor analysis suggests that property performance and equity performance are distinctive: there is little persuasive evidence of contemporaneous or lagged sub-class linkages. Formal tests of the equivalence of optimised portfolios using top-down and global approaches failed to demonstrate significant differences, whether or not allocations were constrained. While the results may be a function of measurement of market returns, it is those returns that are used to assess fund performance. Accordingly, the treatment of real estate as a distinct asset class with diversification potential seems justified.

Automated categorisation of e-journals by synonym analysis of n-grams

Automatic keyword or keyphrase extraction is concerned with assigning keyphrases to documents based on words from within the document. Previous studies have shown that in a significant number of cases author-supplied keywords are not appropriate for the document to which they are attached. This can either be because they represent what the author believes a paper is about not what it actually is, or because they include keyphrases which are more classificatory than explanatory e.g., “University of Poppleton” instead of “Knowledge Discovery in Databases”. Thus, there is a need for a system that can generate an appropriate and diverse range of keyphrases that reflect the document. This paper proposes two possible solutions that examine the synonyms of words and phrases in the document to find the underlying themes, and presents these as appropriate keyphrases. Using three different freely available thesauri, the work undertaken examines two different methods of producing keywords and compares the outcomes across multiple strands in the timeline. The primary method explores taking n-grams of the source document phrases, and examining the synonyms of these, while the secondary considers grouping outputs by their synonyms. The experiments undertaken show the primary method produces good results and that the secondary method produces both good results and potential for future work. In addition, the different qualities of the thesauri are examined and it is concluded that the more entries in a thesaurus, the better it is likely to perform. The age of the thesaurus or the size of each entry does not correlate to performance.

Aerosol-induced correlation between visibility and atmospheric electricity.

Atmospheric aerosol acts to both reduce the background concentration of natural cluster ions, and to attenuate optical propagation. Hence, the presence of aerosol has two consequences, the reduction of the air’s electrical conductivity and the visual range. Ion-aerosol theory and Koschmieder’s visibility theory are combined here to derive the related non-linear variation of the atmospheric electric potential gradient with visual range. A substantial sensitivity is found under poor visual range conditions, but, for good visual range conditions the sensitivity diminishes and little influence of local aerosol on the fair weather potential gradient occurs. This allows visual range measurements, made simply and routinely at many meteorological sites, to provide inference about the local air’s electrical properties.

Comparative genomics of Brachyspira pilosicoli strains: genome rearrangements, reductions and correlation of genetic compliment with phenotypic diversity.

Background. The anaerobic spirochaete Brachyspira pilosicoli causes enteric disease in avian, porcine and human hosts, amongst others. To date, the only available genome sequence of B. pilosicoli is that of strain 95/1000, a porcine isolate. In the first intra-species genome comparison within the Brachyspira genus, we report the whole genome sequence of B. pilosicoli B2904, an avian isolate, the incomplete genome sequence of B. pilosicoli WesB, a human isolate, and the comparisons with B. pilosicoli 95/1000. We also draw on incomplete genome sequences from three other Brachyspira species. Finally we report the first application of the high-throughput Biolog phenotype screening tool on the B. pilosicoli strains for detailed comparisons between genotype and phenotype. Results. Feature and sequence genome comparisons revealed a high degree of similarity between the three B. pilosicoli strains, although the genomes of B2904 and WesB were larger than that of 95/1000 (~2,765, 2.890 and 2.596 Mb, respectively). Genome rearrangements were observed which correlated largely with the positions of mobile genetic elements. Through comparison of the B2904 and WesB genomes with the 95/1000 genome, features that we propose are non-essential due to their absence from 95/1000 include a peptidase, glycine reductase complex components and transposases. Novel bacteriophages were detected in the newly-sequenced genomes, which appeared to have involvement in intra- and inter-species horizontal gene transfer. Phenotypic differences predicted from genome analysis, such as the lack of genes for glucuronate catabolism in 95/1000, were confirmed by phenotyping. Conclusions. The availability of multiple B. pilosicoli genome sequences has allowed us to demonstrate the substantial genomic variation that exists between these strains, and provides an insight into genetic events that are shaping the species. In addition, phenotype screening allowed determination of how genotypic differences translated to phenotype. Further application of such comparisons will improve understanding of the metabolic capabilities of Brachyspira species.

Correlation between Fourier series expansion and Hückel orbital theory

The Fourier series can be used to describe periodic phenomena such as the one-dimensional crystal wave function. By the trigonometric treatements in Hückel theory it is shown that Hückel theory is a special case of Fourier series theory. Thus, the conjugated π system is in fact a periodic system. Therefore, it can be explained why such a simple theorem as Hückel theory can be so powerful in organic chemistry. Although it only considers the immediate neighboring interactions, it implicitly takes account of the periodicity in the complete picture where all the interactions are considered. Furthermore, the success of the trigonometric methods in Hückel theory is not accidental, as it based on the fact that Hückel theory is a specific example of the more general method of Fourier series expansion. It is also important for education purposes to expand a specific approach such as Hückel theory into a more general method such as Fourier series expansion.

A comparison of automated keyphrase extraction techniques and of automatic evaluation vs. human evaluation

Keyphrases are added to documents to help identify the areas of interest they contain. However, in a significant proportion of papers author selected keyphrases are not appropriate for the document they accompany: for instance, they can be classificatory rather than explanatory, or they are not updated when the focus of the paper changes. As such, automated methods for improving the use of keyphrases are needed, and various methods have been published. However, each method was evaluated using a different corpus, typically one relevant to the field of study of the method’s authors. This not only makes it difficult to incorporate the useful elements of algorithms in future work, but also makes comparing the results of each method inefficient and ineffective. This paper describes the work undertaken to compare five methods across a common baseline of corpora. The methods chosen were Term Frequency, Inverse Document Frequency, the C-Value, the NC-Value, and a Synonym based approach. These methods were analysed to evaluate performance and quality of results, and to provide a future benchmark. It is shown that Term Frequency and Inverse Document Frequency were the best algorithms, with the Synonym approach following them. Following these findings, a study was undertaken into the value of using human evaluators to judge the outputs. The Synonym method was compared to the original author keyphrases of the Reuters’ News Corpus. The findings show that authors of Reuters’ news articles provide good keyphrases but that more often than not they do not provide any keyphrases.

The correlation of ULF waves and auroral intensity before, during and after substorm expansion phase onset

We present case studies of the evolution of magnetic wave amplitudes and auroral intensity through the late growth phase and the expansion phase of the substorm cycle. We present strong evidence that substorm-related auroral enhancements are clearly and demonstrably linked to ULF wave amplitudes observed at the same location. In most cases, we find that the highest correlations are observed when the magnetometer time series is advanced in time, indicating that the ULF wave amplitudes start to grow before measured auroral intensities, though interestingly this is not always the case. Further we discuss the four possible reasons that may be able to explain both the timing and the high correlations between these two phenomena, including: a simple coincidence, an artifact of instrumental effects, the response of the ionosphere to magnetic waves and auroral particle precipitation, and finally that ULF waves and auroral particle precipitation are physically linked. We discount coincidence and instrumental effects since in the studies presented here they are unlikely or in general will contribute negligible effects, and we find that the ionospheric response to waves and precipitation can explain some, but not all of the results contained within this paper. Specifically, ionospheric response to substorm waves and auroral precipitation cannot explain that the result that previous studies have shown, that onset of ULF wave activity and the onset of auroral particle precipitation occur at the same time and in the same location. This leaves the possibility that ULF waves and auroral particles are physically linked.

Consistent and contrasting decadal Arctic sea ice thickness predictions from a highly optimized sea ice model

[1] Decadal hindcast simulations of Arctic Ocean sea ice thickness made by a modern dynamic-thermodynamic sea ice model and forced independently by both the ERA-40 and NCEP/NCAR reanalysis data sets are compared for the first time. Using comprehensive data sets of observations made between 1979 and 2001 of sea ice thickness, draft, extent, and speeds, we find that it is possible to tune model parameters to give satisfactory agreement with observed data, thereby highlighting the skill of modern sea ice models, though the parameter values chosen differ according to the model forcing used. We find a consistent decreasing trend in Arctic Ocean sea ice thickness since 1979, and a steady decline in the Eastern Arctic Ocean over the full 40-year period of comparison that accelerated after 1980, but the predictions of Western Arctic Ocean sea ice thickness between 1962 and 1980 differ substantially. The origins of differing thickness trends and variability were isolated not to parameter differences but to differences in the forcing fields applied, and in how they are applied. It is argued that uncertainty, differences and errors in sea ice model forcing sets complicate the use of models to determine the exact causes of the recently reported decline in Arctic sea ice thickness, but help in the determination of robust features if the models are tuned appropriately against observations.

Automated determination of auroral breakup during the substorm expansion phase using all sky imager data

This technique paper describes a novel method for quantitatively and routinely identifying auroral breakup following substorm onset using the Time History of Events and Macroscale Interactions During Substorms (THEMIS) all-sky imagers (ASIs). Substorm onset is characterised by a brightening of the aurora that is followed by auroral poleward expansion and auroral breakup. This breakup can be identified by a sharp increase in the auroral intensity i(t) and the time derivative of auroral intensity i'(t). Utilising both i(t) and i'(t) we have developed an algorithm for identifying the time interval and spatial location of auroral breakup during the substorm expansion phase within the field of view of ASI data based solely on quantifiable characteristics of the optical auroral emissions. We compare the time interval determined by the algorithm to independently identified auroral onset times from three previously published studies. In each case the time interval determined by the algorithm is within error of the onset independently identified by the prior studies. We further show the utility of the algorithm by comparing the breakup intervals determined using the automated algorithm to an independent list of substorm onset times. We demonstrate that up to 50% of the breakup intervals characterised by the algorithm are within the uncertainty of the times identified in the independent list. The quantitative description and routine identification of an interval of auroral brightening during the substorm expansion phase provides a foundation for unbiased statistical analysis of the aurora to probe the physics of the auroral substorm as a new scientific tool for aiding the identification of the processes leading to auroral substorm onset.

Automated identification and tracking of polar-cap plasma patches at solar minimum

A method of automatically identifying and tracking polar-cap plasma patches, utilising data inversion and feature-tracking methods, is presented. A well-established and widely used 4-D ionospheric imaging algorithm, the Multi-Instrument Data Assimilation System (MIDAS), inverts slant total electron content (TEC) data from ground-based Global Navigation Satellite System (GNSS) receivers to produce images of the free electron distribution in the polar-cap ionosphere. These are integrated to form vertical TEC maps. A flexible feature-tracking algorithm, TRACK, previously used extensively in meteorological storm-tracking studies is used to identify and track maxima in the resulting 2-D data fields. Various criteria are used to discriminate between genuine patches and "false-positive" maxima such as the continuously moving day-side maximum, which results from the Earth's rotation rather than plasma motion. Results for a 12-month period at solar minimum, when extensive validation data are available, are presented. The method identifies 71 separate structures consistent with patch motion during this time. The limitations of solar minimum and the consequent small number of patches make climatological inferences difficult, but the feasibility of the method for patches larger than approximately 500 km in scale is demonstrated and a larger study incorporating other parts of the solar cycle is warranted. Possible further optimisation of discrimination criteria, particularly regarding the definition of a patch in terms of its plasma concentration enhancement over the surrounding background, may improve results.

Prediction of Staphylococcus aureus antimicrobial resistance by whole-genome sequencing

Whole-genome sequencing (WGS) could potentially provide a single platform for extracting all the information required to predict an organism’s phenotype. However, its ability to provide accurate predictions has not yet been demonstrated in large independent studies of specific organisms. In this study, we aimed to develop a genotypic prediction method for antimicrobial susceptibilities. The whole genomes of 501 unrelated Staphylococcus aureus isolates were sequenced, and the assembled genomes were interrogated using BLASTn for a panel of known resistance determinants (chromosomal mutations and genes carried on plasmids). Results were compared with phenotypic susceptibility testing for 12 commonly used antimicrobial agents (penicillin, methicillin, erythromycin, clindamycin, tetracycline, ciprofloxacin, vancomycin, trimethoprim, gentamicin, fusidic acid, rifampin, and mupirocin) performed by the routine clinical laboratory. We investigated discrepancies by repeat susceptibility testing and manual inspection of the sequences and used this information to optimize the resistance determinant panel and BLASTn algorithm. We then tested performance of the optimized tool in an independent validation set of 491 unrelated isolates, with phenotypic results obtained in duplicate by automated broth dilution (BD Phoenix) and disc diffusion. In the validation set, the overall sensitivity and specificity of the genomic prediction method were 0.97 (95% confidence interval [95% CI], 0.95 to 0.98) and 0.99 (95% CI, 0.99 to 1), respectively, compared to standard susceptibility testing methods. The very major error rate was 0.5%, and the major error rate was 0.7%. WGS was as sensitive and specific as routine antimicrobial susceptibility testing methods. WGS is a promising alternative to culture methods for resistance prediction in S. aureus and ultimately other major bacterial pathogens.