Pairwise Metrics Research Articles

Uniform Tensor Clustering by Jointly Exploring Sample Affinities of Various Orders.

Traditional clustering methods rely on pairwise affinity to divide samples into different subgroups. However, high-dimensional small-sample (HDLSS) data are affected by the concentration effects, rendering traditional pairwise metrics unable to accurately describe relationships between samples, leading to suboptimal clustering results. This article advances the proposition of employing high-order affinities to characterize multiple sample relationships as a strategic means to circumnavigate the concentration effects. We establish a nexus between different order affinities by constructing specialized decomposable high-order affinities, thereby formulating a uniform mathematical framework. Building upon this insight, a novel clustering method named uniform tensor clustering (UTC) is proposed, which learns a consensus low-dimensional embedding for clustering by the synergistic exploitation of multiple-order affinities. Extensive experiments on synthetic and real-world datasets demonstrate two findings: 1) high-order affinities are better suited for characterizing sample relationships in complex data and 2) reasonable use of different order affinities can enhance clustering effectiveness, especially in handling high-dimensional data.

Dynamical irreversibility, high‐order interactions, and whole‐brain modeling of gamma oscillations in neurodegeneration

AbstractBackgroundAbnormal gamma oscillations (γ) have been systematically reported in preclinical animal models of Alzheimer’s disease (AD) and human AD patients. However, little is known about the underlying mechanisms, non‐linear dynamics, and high‐order interactions of γ in dementia.MethodTo bridge this gap, we combined EEG and fMRI with three novel approaches to offer a comprehensive pathophysiological characterization of γ oscillations: We assessed the irreversibility of large‐scale brain dynamics to provide a precise signature capable of distinguishing AD at the global, local and network levels and at different oscillatory regimes (study 1). A high‐order functional connectivity (HOFC) multivariate information theory approach capturing interactions between 3 or more regions across spatiotemporal scales was compared with standard connectivity metrics for dementia characterization (study 2). A whole‐brain semi‐empirical model with a perturbational approach combined EEG, structural connectivity, and atrophy to explain the disrupted high‐order interactions observed in AD (study 3).ResultsStudy 1: AD was associated with a breakdown of temporal irreversibility at the global, local, and network levels at γ and other oscillations. The irreversibility of brain dynamics provided higher accuracy and more distinctive information than classical neurocognitive measures when differentiating AD from controls. Study 2: EEG‐HOFC revealed hypoconnectivity in γ between frontal, limbic, and sensory regions in AD, and large effect sizes, compared to standard pairwise metrics, proving a more accurate and parsimonious characterization of AD and frontotemporal dementia (FTD) patients. Study 3: whole‐brain modeling showed that a combined mechanism involving reduced structural integration (DTI) and hypo‐excitability (altered excitation/inhibition balance) triggered the altered high‐order interaction γ dynamics observed in AD and FTD. A perturbational approach identified physiopathological regions (SFG, MTG, precuneus) deviating the trajectories from disease to a healthy brain state.ConclusionThe present results offer new evidence regarding the pathophysiological link, mechanistic processes, and non‐linear whole‐brain dynamics between altered resting‐state γ rhythms and the characterization of AD and FTD, opening new avenues for dementia research at different levels.

Reference standard for the evaluation of automatic segmentation algorithms: Quantification of inter observer variability of manual delineation of prostate contour on MRI

PurposeThe purpose of this study was to investigate the relationship between inter-reader variability in manual prostate contour segmentation on magnetic resonance imaging (MRI) examinations and determine the optimal number of readers required to establish a reliable reference standard. Materials and methodsSeven radiologists with various experiences independently performed manual segmentation of the prostate contour (whole-gland [WG] and transition zone [TZ]) on 40 prostate MRI examinations obtained in 40 patients. Inter-reader variability in prostate contour delineations was estimated using standard metrics (Dice similarity coefficient [DSC], Hausdorff distance and volume-based metrics). The impact of the number of readers (from two to seven) on segmentation variability was assessed using pairwise metrics (consistency) and metrics with respect to a reference segmentation (conformity), obtained either with majority voting or simultaneous truth and performance level estimation (STAPLE) algorithm. ResultsThe average segmentation DSC for two readers in pairwise comparison was 0.919 for WG and 0.876 for TZ. Variability decreased with the number of readers: the interquartile ranges of the DSC were 0.076 (WG) / 0.021 (TZ) for configurations with two readers, 0.005 (WG) / 0.012 (TZ) for configurations with three readers, and 0.002 (WG) / 0.0037 (TZ) for configurations with six readers. The interquartile range decreased slightly faster between two and three readers than between three and six readers. When using consensus methods, variability often reached its minimum with three readers (with STAPLE, DSC = 0.96 [range: 0.945–0.971] for WG and DSC = 0.94 [range: 0.912–0.957] for TZ, and interquartile range was minimal for configurations with three readers. ConclusionThe number of readers affects the inter-reader variability, in terms of inter-reader consistency and conformity to a reference. Variability is minimal for three readers, or three readers represent a tipping point in the variability evolution, with both pairwise-based metrics or metrics with respect to a reference. Accordingly, three readers may represent an optimal number to determine references for artificial intelligence applications.

Robust classification using average correlations as features (ACF)

MotivationIn single-cell transcriptomics and other omics technologies, large fractions of missing values commonly occur. Researchers often either consider only those features that were measured for each instance of their dataset, thereby accepting severe loss of information, or use imputation which can lead to erroneous results. Pairwise metrics allow for imputation-free classification with minimal loss of data.ResultsUsing pairwise correlations as metric, state-of-the-art approaches to classification would include the K-nearest-neighbor- (KNN) and distribution-based-classification-classifier. Our novel method, termed average correlations as features (ACF), significantly outperforms those approaches by training tunable machine learning models on inter-class and intra-class correlations. Our approach is characterized in simulation studies and its classification performance is demonstrated on real-world datasets from single-cell RNA sequencing and bottom-up proteomics. Furthermore, we demonstrate that variants of our method offer superior flexibility and performance over KNN classifiers and can be used in conjunction with other machine learning methods. In summary, ACF is a flexible method that enables missing value tolerant classification with minimal loss of data.

Revealing the spatiotemporal requirements for accurate subject identification with resting-state functional connectivity: a simultaneous fNIRS-fMRI study.

Brain fingerprinting refers to identifying participants based on their functional patterns. Despite its success with functional magnetic resonance imaging (fMRI), brain fingerprinting with functional near-infrared spectroscopy (fNIRS) still lacks adequate validation. We investigated how fNIRS-specific acquisition features (limited spatial information and nonneural contributions) influence resting-state functional connectivity (rsFC) patterns at the intra-subject level and, therefore, brain fingerprinting. We performed multiple simultaneous fNIRS and fMRI measurements in 29 healthy participants at rest. Data were preprocessed following the best practices, including the removal of motion artifacts and global physiology. The rsFC maps were extracted with the Pearson correlation coefficient. Brain fingerprinting was tested with pairwise metrics and a simple linear classifier. Our results show that average classification accuracy with fNIRS ranges from 75% to 98%, depending on the number of runs and brain regions used for classification. Under the right conditions, brain fingerprinting with fNIRS is close to the 99.9% accuracy found with fMRI. Overall, the classification accuracy is more impacted by the number of runs and the spatial coverage than the choice of the classification algorithm. This work provides evidence that brain fingerprinting with fNIRS is robust and reliable for extracting unique individual features at the intra-subject level once relevant spatiotemporal constraints are correctly employed.

Genuine high-order interactions in brain networks and neurodegeneration

Brain functional networks have been traditionally studied considering only interactions between pairs of regions, neglecting the richer information encoded in higher orders of interactions. In consequence, most of the connectivity studies in neurodegeneration and dementia use standard pairwise metrics. Here, we developed a genuine high-order functional connectivity (HOFC) approach that captures interactions between 3 or more regions across spatiotemporal scales, delivering a more biologically plausible characterization of the pathophysiology of neurodegeneration. We applied HOFC to multimodal (electroencephalography [EEG], and functional magnetic resonance imaging [fMRI]) data from patients diagnosed with behavioral variant of frontotemporal dementia (bvFTD), Alzheimer's disease (AD), and healthy controls. HOFC revealed large effect sizes, which, in comparison to standard pairwise metrics, provided a more accurate and parsimonious characterization of neurodegeneration. The multimodal characterization of neurodegeneration revealed hypo and hyperconnectivity on medium to large-scale brain networks, with a larger contribution of the former. Regions as the amygdala, the insula, and frontal gyrus were associated with both effects, suggesting potential compensatory processes in hub regions. fMRI revealed hypoconnectivity in AD between regions of the default mode, salience, visual, and auditory networks, while in bvFTD between regions of the default mode, salience, and somatomotor networks. EEG revealed hypoconnectivity in the γ band between frontal, limbic, and sensory regions in AD, and in the δ band between frontal, temporal, parietal and posterior areas in bvFTD, suggesting additional pathophysiological processes that fMRI alone can not capture. Classification accuracy was comparable with standard biomarkers and robust against confounders such as sample size, age, education, and motor artifacts (from fMRI and EEG). We conclude that high-order interactions provide a detailed, EEG- and fMRI compatible, biologically plausible, and psychopathological-specific characterization of different neurodegenerative conditions.

A spatiotemporal assessment of the high-resolution CHIRPS rainfall dataset in southwestern Colombia using combined principal component analysis

Long and temporal time-consistency rainfall time series are essential for studying climate; nevertheless, raingauge stations are unevenly distributed across southwestern Colombia. This research paper assesses the consistency of the satellite rainfall estimate from Climate Hazards Group Infrared Precipitation (CHIRPS) via pixel-to-point comparison with 46 observed monthly rainfall time series using four pairwise metrics and Combined Principal Component Analysis (CPCA). Two Combined Principal Components (CPC) were also used to determine the relationship with the Sea Surface Temperature (SST) through simultaneous linear correlation maps. The results showed that CHIRPS has a better performance in the Andean region than in the Pacific region. The correlation between CPC1 (CPC2) and SST showed a typical El Niño Southern Oscillation pattern with an inverse (direct) relationship between the rainfall in the Andean (Pacific) Region. Finally, our results validate that CHIRPS can be included in further studies of the spatiotemporal dynamics of rainfall in Southwestern Colombia.

Assessing the nonlinear decay of community similarity: Permutation and site‐block resampling significance tests

AbstractAimModelling how community similarity decays with spatial distance is a key tool for the study of the processes behind community variation (beta diversity). Distance‐decay models are computed from pairwise metrics (i.e. community similarity and spatial distance between localities) and hence suffer from pairwise dependence in the data, precluding the use of standard significance tests. Besides, distance‐decay patterns are inherently nonlinear because similarity is bounded between 1 and 0. However, the only standard method to assess model significance under pairwise dependency is the Mantel test, which considers a linear model. To allow the use of nonlinear models in the assessment of distance‐decay patterns, we introduce here a nonlinear significance test combining a pseudo‐R2 statistic with either permutations or block‐site resampling with replacement.LocationGlobal.TaxonApplicable to any taxon.MethodsTo assess the performance of the pseudo‐R2 significance test (i.e. type I error and statistical power), we have applied this method to exponential and power‐law generalized linear models of simulated distance‐decay data. We compared its performance with the one of a linear model (i.e. Mantel test) and illustrated its use with real data.ResultsThe pseudo‐R2 significance test has generally good type I error and statistical power, even for highly nonlinear relationships. The Mantel test shows an adequate empirical approximation of type I error, but presents lower statistical power when the distance‐decay relationship highly departs from linearity.Main conclusionsThe proposed pseudo‐R2 significance test for nonlinear models is adequate for exponential and power‐law models of distance‐decay of similarity and should be preferred over the linear Mantel test. Furthermore, this significance test could also be used in other nonlinear distance‐based regressions. The novel site‐block resampling method could also be applied to other statistical questions related to distance‐decay models that cannot be tackled using permutations, such as estimating the variance of model parameters.

A multi-site method to capture turnover in rare to common interactions in bipartite species networks.

Ecological network structure is maintained by a generalist core of common species. However, rare species contribute substantially to both the species and functional diversity of networks. Capturing changes in species composition and interactions, measured as turnover, is central to understanding the contribution of rare and common species and their interactions. Due to a large contribution of rare interactions, the pairwise metrics used to quantify interaction turnover are, however, sensitive to compositional change in the interactions of, often rare, peripheral specialists rather than common generalists in the network. Here we expand on pairwise interaction turnover using a multi-site metric that enables quantifying turnover in rare to common interactions (in terms of occurrence of interactions). The metric further separates this turnover into interaction turnover due to species turnover and interaction rewiring. We demonstrate the application and value of this method using a host-parasitoid system sampled along gradients of environmental modification. In the study system, both the type and amount of habitat needed to maintain interaction composition depended on the properties of the interactions considered, that is, from rare to common. The analyses further revealed the potential of host switching to prevent or delay species loss, and thereby buffer the system from perturbation. Multi-site interaction turnover provides a comprehensive measure of network change that can, for example, detect ecological thresholds to habitat loss for rare to common interactions. Accurate description of turnover in common, in addition to rare, species and their interactions is particularly relevant for understanding how network structure and function can be maintained.

A principal component analysis approach to assess CHIRPS precipitation dataset for the study of climate variability of the La Plata Basin, Southern South America

This article assesses the consistency of the satellite precipitation estimate CHIRPS v.2 to describe the spatiotemporal rainfall variability in the La Plata Basin (LPB), the second largest hydrographic basin in South America, by (a) pixel-to-point comparison of CHIRPS data with 167 observed monthly precipitation time series using three pairwise metrics (coefficient of correlation, bias and root mean square error) and (b) principal component analysis (PCA) to evaluate the large-scale coherence between CHIRPS and rain gauge data. The pairwise metrics indicate that CHIRPS better represents the rainfall in the coastal, northeastern and southeastern parts of the basin than in the Andean region to the west. The PCA shows that CHIRPS describes most of the observed rainfall variability in the LPB, but contains more variability, especially during December–February and March–May seasons. The two major modes observed are highly correlated spatially (empirical orthogonal functions—EOFs) and temporally (principal components—PCs) with the corresponding CHIRPS modes. The PCA allows the determination of the main rainfall variability modes and their possible relations with climate variability modes. Besides, the analyses of the precipitation anomaly modes show that the El Nino Southern Oscillation explains the first EOF modes of datasets. The PCA provides an alternative and effective means of assessing the consistency of CHIRPS data in representing spatial and temporal rainfall variability in the LPB.

Imputation-Aware Tag SNP Selection To Improve Power for Large-Scale, Multi-ethnic Association Studies.

The emergence of very large cohorts in genomic research has facilitated a focus on genotype-imputation strategies to power rare variant association. These strategies have benefited from improvements in imputation methods and association tests, however little attention has been paid to ways in which array design can increase rare variant association power. Therefore, we developed a novel framework to select tag SNPs using the reference panel of 26 populations from Phase 3 of the 1000 Genomes Project. We evaluate tag SNP performance via mean imputed r2 at untyped sites using leave-one-out internal validation and standard imputation methods, rather than pairwise linkage disequilibrium. Moving beyond pairwise metrics allows us to account for haplotype diversity across the genome for improve imputation accuracy and demonstrates population-specific biases from pairwise estimates. We also examine array design strategies that contrast multi-ethnic cohorts vs. single populations, and show a boost in performance for the former can be obtained by prioritizing tag SNPs that contribute information across multiple populations simultaneously. Using our framework, we demonstrate increased imputation accuracy for rare variants (frequency < 1%) by 0.5–3.1% for an array of one million sites and 0.7–7.1% for an array of 500,000 sites, depending on the population. Finally, we show how recent explosive growth in non-African populations means tag SNPs capture on average 30% fewer other variants than in African populations. The unified framework presented here will enable investigators to make informed decisions for the design of new arrays, and help empower the next phase of rare variant association for global health.

A meta‐analysis of nestedness and turnover components of beta diversity across organisms and ecosystems

AbstractAimThe number of studies investigating the nestedness and turnover components of beta diversity has increased substantially, but our general understanding of the drivers of turnover and nestedness remains elusive. Here, we examined the effects of species traits, spatial extent, latitude and ecosystem type on the nestedness and turnover components of beta diversity.LocationGlobal.Time period1968–2017.Major taxa studiedFrom bacteria to mammals.MethodsFrom the 99 studies that partition total beta diversity into its turnover and nestedness components, we assembled 269 and 259 data points for the pairwise and multiple site beta‐diversity metrics, respectively. Our data covered a broad variation in species dispersal type, body size and trophic position. The data were from freshwater, marine and terrestrial realms, and encompassed geographical areas from the tropics to near polar regions. We used linear modelling as a meta‐regression tool to analyse the data.ResultsPairwise turnover, multiple site turnover and total beta diversity all decreased significantly with latitude. In contrast, multiple site nestedness showed a positive relationship with latitude. Beta‐diversity components did not generally differ among the realms. The turnover component and total beta diversity increased with spatial extent, whereas nestedness was scale invariant for pairwise metrics. Multiple site beta‐diversity components did not vary with spatial extent. Surprisingly, passively dispersed organisms had lower turnover and total beta diversity than flying organisms. Body size showed a relatively weak relationship with beta diversity but had important interactions with trophic position, thus also affecting beta diversity via interactive effects. Producers had significantly higher average pairwise turnover and total beta diversity than carnivores.Main conclusionsThe present results provide evidence that species turnover, being consistently the larger component of total beta diversity, and nestedness are related to the latitude of the study area and intrinsic organismal features. We showed that two beta‐diversity components had generally opposing patterns with regard to latitude. We highlight that beta‐diversity partition may give additional insights into the underlying causes of spatial variability in biotic communities compared with total beta diversity alone.

A dynamic selection ensemble method for target recognition based on clustering and randomized reference classifier

In order to improve the generalization ability and recognition efficiency of the maritime surveillance radar, a novel selection ensemble technique, termed KMRRC, based on k-medoids clustering and random reference classifier (RRC) is proposed. By disturbing the training set base classifiers are generated, which are then divided into several clusters based on pairwise diversity metrics, finally the RRC model is used to select several most competent classifiers from each cluster to classify each query object. The performance of KMRRC is compared against nine ensemble learning methods using a self-built high range resolution profile (HRRP) data set and twenty UCI databases. The experimental results clearly show the KMMRRC’s feasibility and effectiveness. In addition, the influence of the selection of diversity measures is studied concurrently.

Optimum solution and evaluation of rectangular jigsaw puzzles based on branch and bound method and combinatorial accuracy

Automatic solution of rectangular jigsaw puzzles can be broken down into two separate steps of calculating pairwise compatibility metrics and jigsaw assembling algorithm. This work discriminates between two different sources of errors, each corresponding to one of these two steps. In this regard, type I error is defined as the imperfection of the used compatibility metric, and type II error is reserved to measure the imperfection of jigsaw assembling algorithm. Differentiating between these two types of error allows us to tweak and optimize different parts of the algorithm to achieve the best performance. Based upon these defined terms, this study argues that current jigsaw assembling algorithms mainly rely on either greedy methods or metaheuristic algorithms, which may impose a considerable amount of type II error to the final solution. This paper demonstrates that a powerful and perfect (i.e., type II error-free) jigsaw assembling algorithm is achievable by combining branch and bound technique with graph theory. This perfect jigsaw assembling algorithm is then utilized to measure the performances of various compatibility metrics and color models. The superiority of red-green-blue (RGB) color model and Mahalanobis gradient compatibility (MGC) metric in solving rectangular jigsaw puzzles is shown by providing conclusive evidence. Additionally, a mean opinion score (MOS) test is conducted to examine the accuracy of the existing metrics. According to the results from MOS test, we argue that the existing performance criteria are not concise and accurate; thus, a new accuracy metric is proposed on the basis of comparing different sub-blocks of solutions. Finally, the efficiency of jigsaw assembling algorithm is measured by proposing a new performance criterion.

Pairwise beta diversity resolves an underappreciated source of confusion in calculating species turnover

Beta diversity is an important metric in ecology quantifying differentiation or disparity in composition among communities, ecosystems, or phenotypes. To compare systems with different sizes (N, number of units within a system), beta diversity is often converted to related indices such as turnover or local/regional differentiation. Here we use simulations to demonstrate that these naive measures of dissimilarity depend on sample size and design. We show that when N is the number of sampled units (e.g., quadrats) rather than the "true" number of communities in the system (if such exists), these differentiation measures are biased estimators. We propose using average pairwise dissimilarity as an intuitive solution. That is, instead of attempting to estimate an N-community measure, we advocate estimating the expected dissimilarity between any random pair of communities (or sampling units)-especially when the "true" N is unknown or undefined. Fortunately, measures of pairwise dissimilarity or overlap have been used in ecology for decades, and their properties are well known. Using the same simulations, we show that average pairwise metrics give consistent and unbiased estimates regardless of the number of survey units sampled. We advocate pairwise dissimilarity as a general standardization to ensure commensurability of different study systems.

Beyond connectedness: why pairwise metrics cannot capture community stability.

The connectedness of species in a trophic web has long been a key structural characteristic for both theoreticians and empiricists in their understanding of community stability. In the past decades, there has been a shift from focussing on determining the number of interactions to taking into account their relative strengths. The question is: How do the strengths of the interactions determine the stability of a community? Recently, a metric has been proposed which compares the stability of observed communities in terms of the strength of three‐ and two‐link feedback loops (cycles of interaction strengths). However, it has also been suggested that we do not need to go beyond the pairwise structure of interactions to capture stability. Here, we directly compare the performance of the feedback and pairwise metrics. Using observed food‐web structures, we show that the pairwise metric does not work as a comparator of stability and is many orders of magnitude away from the actual stability values. We argue that metrics based on pairwise‐strength information cannot capture the complex organization of strong and weak links in a community, which is essential for system stability.

Mapping Averaged Pairwise Information (MAPI): a new exploratory tool to uncover spatial structure

Summary Visualisation of spatial networks based on pairwise metrics such as (dis)similarity coefficients provides direct information on spatial organisation of biological systems. However, for large networks, graphical representations are often unreadable as nodes (samples), and edges (links between samples) strongly overlap. We present a new method, MAPI, allowing translation from spatial networks to variation surfaces. MAPI relies on (i) a spatial network in which samples are linked by ellipses and (ii) a grid of hexagonal cells encompassing the study area. Pairwise metric values are attributed to ellipses and averaged within the cells they intersect. The resulting surface of variation can be displayed as a colour map in Geographical Information System (GIS), along with other relevant layers, such as land cover. The method also allows the identification of significant discontinuities in grid cell values through a nonparametric randomisation procedure. The interest of MAPI is here demonstrated in the field of spatial and landscape genetics. Using simulated test data sets, as well as observed data from three biological models, we show that MAPI is (i) relatively insensitive to confounding effects resulting from isolation by distance (i.e. over‐structuring), (ii) efficient in detecting barriers when they are not too permeable to gene flow and, (iii) useful to explore relationships between spatial genetic patterns and landscape features. MAPI is freely provided as a PostgreSQL/PostGIS data base extension allowing easy interaction with GIS or the r software and other programming languages. Although developed for spatial and landscape genetics, the method can also be useful to visualise spatial organisation from other kinds of data from which pairwise metrics can be computed.

Genic insights from integrated human proteomics in GeneCards.

GeneCards is a one-stop shop for searchable human gene annotations (http://www.genecards.org/). Data are automatically mined from ∼120 sources and presented in an integrated web card for every human gene. We report the application of recent advances in proteomics to enhance gene annotation and classification in GeneCards. First, we constructed the Human Integrated Protein Expression Database (HIPED), a unified database of protein abundance in human tissues, based on the publically available mass spectrometry (MS)-based proteomics sources ProteomicsDB, Multi-Omics Profiling Expression Database, Protein Abundance Across Organisms and The MaxQuant DataBase. The integrated database, residing within GeneCards, compares favourably with its individual sources, covering nearly 90% of human protein-coding genes. For gene annotation and comparisons, we first defined a protein expression vector for each gene, based on normalized abundances in 69 normal human tissues. This vector is portrayed in the GeneCards expression section as a bar graph, allowing visual inspection and comparison. These data are juxtaposed with transcriptome bar graphs. Using the protein expression vectors, we further defined a pairwise metric that helps assess expression-based pairwise proximity. This new metric for finding functional partners complements eight others, including sharing of pathways, gene ontology (GO) terms and domains, implemented in the GeneCards Suite. In parallel, we calculated proteome-based differential expression, highlighting a subset of tissues that overexpress a gene and subserving gene classification. This textual annotation allows users of VarElect, the suite’s next-generation phenotyper, to more effectively discover causative disease variants. Finally, we define the protein–RNA expression ratio and correlation as yet another attribute of every gene in each tissue, adding further annotative information. The results constitute a significant enhancement of several GeneCards sections and help promote and organize the genome-wide structural and functional knowledge of the human proteome.Database URL: http://www.genecards.org/

Functional Mesh Model with Temporal Measurements for brain decoding.

We propose a method called Functional Mesh Model with Temporal Measurements (FMM-TM) to estimate a functional relationship among voxels using temporal data, and employ these relationships for brain decoding. For each sample, we measure Blood Oxygenation Level Dependent (BOLD) responses from each voxel, and construct a functional mesh around each voxel (called seed voxel) with its nearest neighbors selected using distance metrics namely Pearson correlation, cosine similarity and Euclidean distance. Then, we represent the BOLD response of a seed voxel in terms of linear combination of BOLD responses of its p-nearest neighbors. The relationship between the seed voxel and its neighbors is represented using a set of weights which are estimated by employing linear regression. We train Support Vector Machine and k-Nearest Neighbor classifiers using the estimated weights as features. We test our model in an event-related design experiment, namely object recognition, and observe that our features perform better than raw voxel intensity values, features obtained using various pairwise distance metrics, and local mesh model features extracted using stationary and temporal measurements.

LDlink: a web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants.

Assessing linkage disequilibrium (LD) across ancestral populations is a powerful approach for investigating population-specific genetic structure as well as functionally mapping regions of disease susceptibility. Here, we present LDlink, a web-based collection of bioinformatic modules that query single nucleotide polymorphisms (SNPs) in population groups of interest to generate haplotype tables and interactive plots. Modules are designed with an emphasis on ease of use, query flexibility, and interactive visualization of results. Phase 3 haplotype data from the 1000 Genomes Project are referenced for calculating pairwise metrics of LD, searching for proxies in high LD, and enumerating all observed haplotypes. LDlink is tailored for investigators interested in mapping common and uncommon disease susceptibility loci by focusing on output linking correlated alleles and highlighting putative functional variants. LDlink is a free and publically available web tool which can be accessed at http://analysistools.nci.nih.gov/LDlink/. mitchell.machiela@nih.gov.