This study aimed to evaluate whether current electroencephalographic spectral measures can predict participant's performance during future sessions of a motor imagery task. By investigating this point, we hope to understand which spectral components are related to MI "literacy". Twelve healthy subjects performed a neurofeedback task whereby a cursor was moved to one of two targets (left and right) using only motor imagery of the corresponding hands. To evaluate the effect of aptitude, we measured the Mahalanobis' distances between whole-scalp spectral patterns in four frequency bands (theta, alpha, beta, and gamma) during the first session of left and right motor imagery. Later, we used these features as inputs in a Support Vector Regressor to predict performance during the following two sessions. The performance was measured as the percentage of trials where the cursor correctly reached the target. Since our sample was balanced, this approach predicted performance on sessions two and three with mean absolute errors of 15.07±12.94% and 11.98±11.40%, respectively. The most relevant feature in both cases was the Mahalanobis' distance in alpha. These results suggest that participants who can not evoke different patterns of alpha power during left- and right-hand motor imagery during the first session, also are less likely to improve during the following training sessions. The investigation of whole-scalp differences allows a holistic comprehension of the neural basis of motor imagery. This method also characterizes a potential predictor of performance for future applications of MI-based neurofeedback and brain-computer interfaces.

This study aimed to evaluate on-task electroencephalographic spectral measures and its correlation to performance during a motor imagery (MI) task. By investigating this aspect, we hope to understand what makes some individuals MI "illliterates". Eighteen healthy subjects performed an experimental task whereby a cursor was moved to one of two targets (left and right) using only MI of the left and right hands. To evaluate the effect of aptitude, performance was measured as percentage of correct movement to target, and Mahalanobis distances were calculated between whole-scalp spectral patterns during left and right motor imagery. Then the correlation between performance and Mahalanobis distance was investigated for central, and whole-head topographies using Spearman's correlations. In central topographies, distances on alpha band were positively correlated with performance (ρ=0.562, p=0.032), while distances on theta band were negatively correlated to performance (ρ--0.648, p=0.018) in whole-head maps. The investigation of on-task whole-scalp differences allows a holistic comprehension of the neural basis of motor imagery, as well as how this leads to performance variations.

Hippocampal damage results in profound retrograde, but no anterograde amnesia in contextual fear conditioning (CFC). Although the content learned in the latter have been discussed, alternative regions supporting CFC learning were seldom proposed and never empirically addressed. Here, we employed network analysis of pCREB expression quantified from brain slices of rats with dorsal hippocampal lesion (dHPC) after undergoing CFC session. Using inter-regional correlations of pCREB-positive nuclei between brain regions, we modelled functional networks using different thresholds. The dHPC network showed small-world topology, equivalent to SHAM (control) network. However, diverging hubs were identified in each network. In a direct comparison, hubs in both networks showed consistently higher centrality values compared to the other network. Further, the distribution of correlation coefficients was different between the groups, with most significantly stronger correlation coefficients belonging to the SHAM network. These results suggest that dHPC network engaged in CFC learning is partially different, and engage alternative hubs. We next tested if pre-training lesions of dHPC and one of the new dHPC network hubs (perirhinal, Per; or disgranular retrosplenial, RSC, cortices) would impair CFC. Only dHPC-RSC, but not dHPC-Per, impaired CFC. Interestingly, only RSC showed a consistently higher centrality in the dHPC network, suggesting that the increased centrality reflects an increased functional dependence on RSC. Our results provide evidence that, without hippocampus, the RSC, an anatomically central region in the medial temporal lobe memory system might support CFC learning and memory.

Recent studies have suggested abnormal brain network organization in subjects with Autism Spectrum Disorders (ASD). Here we applied spectral clustering algorithm, diverse centrality measures (betweenness (BC), clustering (CC), eigenvector (EC), and degree (DC)), and also the network entropy (NE) to identify brain sub-systems associated with ASD. We have found that BC increases in the following ASD clusters: in the somatomotor, default-mode, cerebellar, and fronto-parietal. On the other hand, CC, EC, and DC decrease in the somatomotor, default-mode, and cerebellar clusters. Additionally, NE decreases in ASD in the cerebellar cluster. These findings reinforce the hypothesis of under-connectivity in ASD and suggest that the difference in the network organization is more prominent in the cerebellar system. The cerebellar cluster presents reduced NE in ASD, which relates to a more regular organization of the networks. These results might be important to improve current understanding about the etiological processes and the development of potential tools supporting diagnosis and therapeutic interventions.

Several gene regulatory network models containing concepts of directionality at the edges have been proposed. However, only a few reports have an interpretable definition of directionality. Here, differently from the standard causality concept defined by Pearl, we introduce the concept of contagion in order to infer directionality at the edges, i.e., asymmetries in gene expression dependences of regulatory networks. Moreover, we present a bootstrap algorithm in order to test the contagion concept. This technique was applied in simulated data and, also, in an actual large sample of biological data. Literature review has confirmed some genes identified by contagion as actually belonging to the TP53 pathway.

Abstract Summary: We propose a likelihood ratio test (LRT) with Bartlett correction in order to identify Granger causality between sets of time series gene expression data. The performance of the proposed test is compared to a previously published bootstrap-based approach. LRT is shown to be significantly faster and statistically powerful even within non-Normal distributions. An R package named gGranger containing an implementation for both Granger causality identification tests is also provided. Availability: http://dnagarden.ims.u-tokyo.ac.jp/afujita/en/doku.php?id=ggranger. Contact: [email protected] Supplementary information: Supplementary data are available at Bioinformatics online.

BACKGROUND: There are several studies in the literature depicting measurement error in gene expression data and also, several others about regulatory network models. However, only a little fraction describes a combination of measurement error in mathematical regulatory networks and shows how to identify these networks under different rates of noise. RESULTS: This article investigates the effects of measurement error on the estimation of the parameters in regulatory networks. Simulation studies indicate that, in both time series (dependent) and non-time series (independent) data, the measurement error strongly affects the estimated parameters of the regulatory network models, biasing them as predicted by the theory. Moreover, when testing the parameters of the regulatory network models, p-values computed by ignoring the measurement error are not reliable, since the rate of false positives are not controlled under the null hypothesis. In order to overcome these problems, we present an improved version of the Ordinary Least Square estimator in independent (regression models) and dependent (autoregressive models) data when the variables are subject to noises. Moreover, measurement error estimation procedures for microarrays are also described. Simulation results also show that both corrected methods perform better than the standard ones (i.e., ignoring measurement error). The proposed methodologies are illustrated using microarray data from lung cancer patients and mouse liver time series data. CONCLUSIONS: Measurement error dangerously affects the identification of regulatory network models, thus, they must be reduced or taken into account in order to avoid erroneous conclusions. This could be one of the reasons for high biological false positive rates identified in actual regulatory network models.

MOTIVATION: A variety of biological cellular processes are achieved through a variety of extracellular regulators, signal transduction, protein-protein interactions and differential gene expression. Understanding of the mechanisms underlying these processes requires detailed molecular description of the protein and gene networks involved. To better understand these molecular networks, we propose a statistical method to estimate time-varying gene regulatory networks from time series microarray data. One well known problem when inferring connectivity in gene regulatory networks is the fact that the relationships found constitute correlations that do not allow inferring causation, for which, a priori biological knowledge is required. Moreover, it is also necessary to know the time period at which this causation occurs. Here, we present the Dynamic Vector Autoregressive model as a solution to these problems. RESULTS: We have applied the Dynamic Vector Autoregressive model to estimate time-varying gene regulatory networks based on gene expression profiles obtained from microarray experiments. The network is determined entirely based on gene expression profiles data, without any prior biological knowledge. Through construction of three gene regulatory networks (of p53, NF-kappaB and c-myc) for HeLa cells, we were able to predict the connectivity, Granger-causality and dynamics of the information flow in these networks. SUPPLEMENTARY INFORMATION: Additional figures may be found at http://mariwork.iq.usp.br/dvar/.

BACKGROUND: Several mathematical and statistical methods have been proposed in the last few years to analyze microarray data. Most of those methods involve complicated formulas, and software implementations that require advanced computer programming skills. Researchers from other areas may experience difficulties when they attempting to use those methods in their research. Here we present an user-friendly toolbox which allows large-scale gene expression analysis to be carried out by biomedical researchers with limited programming skills. RESULTS: Here, we introduce an user-friendly toolbox called GEDI (Gene Expression Data Interpreter), an extensible, open-source, and freely-available tool that we believe will be useful to a wide range of laboratories, and to researchers with no background in Mathematics and Computer Science, allowing them to analyze their own data by applying both classical and advanced approaches developed and recently published by Fujita et al. CONCLUSION: GEDI is an integrated user-friendly viewer that combines the state of the art SVR, DVAR and SVAR algorithms, previously developed by us. It facilitates the application of SVR, DVAR and SVAR, further than the mathematical formulas present in the corresponding publications, and allows one to better understand the results by means of available visualizations. Both running the statistical methods and visualizing the results are carried out within the graphical user interface, rendering these algorithms accessible to the broad community of researchers in Molecular Biology.

BACKGROUND: With the development of DNA hybridization microarray technologies, nowadays it is possible to simultaneously assess the expression levels of thousands to tens of thousands of genes. Quantitative comparison of microarrays uncovers distinct patterns of gene expression, which define different cellular phenotypes or cellular responses to drugs. Due to technical biases, normalization of the intensity levels is a pre-requisite to performing further statistical analyses. Therefore, choosing a suitable approach for normalization can be critical, deserving judicious consideration. RESULTS: Here, we considered three commonly used normalization approaches, namely: Loess, Splines and Wavelets, and two non-parametric regression methods, which have yet to be used for normalization, namely, the Kernel smoothing and Support Vector Regression. The results obtained were compared using artificial microarray data and benchmark studies. The results indicate that the Support Vector Regression is the most robust to outliers and that Kernel is the worst normalization technique, while no practical differences were observed between Loess, Splines and Wavelets. CONCLUSION: In face of our results, the Support Vector Regression is favored for microarray normalization due to its superiority when compared to the other methods for its robustness in estimating the normalization curve.