Retrospective Change-Points Detection for Multidimensional Time Series of Arbitrary Nature: Model-Free Technology Based on the ϵ-Complexity Theory.

We consider a retrospective change-point detection problem for multidimensional time series of arbitrary nature (in particular, panel data). Change-points are the moments at which the changes in generating mechanism occur. Our method is based on the new theory of ϵ-complexity of individual continuous vector functions and is model-free. We present simulation results confirming the effectiveness of the method.

A simple approximation to the bias of gene-environment interactions in case-control studies with silent disease.

One of the most important research areas in case-control Genome-Wide Association Studies is to determine how the effect of a genotype varies across the environment or to measure the gene-environment interaction (G × E). We consider the scenario when some of the "healthy" controls actually have the disease and when the frequency of these latent cases varies by the environmental variable of interest. In this scenario, performing logistic regression with the clinically diagnosed disease status as an outcome variable and will result in biased estimates of G × E interaction. Here, we derive a general theoretical approximation to the bias in the estimates of the G × E interaction and show, through extensive simulation, that this approximation is accurate in finite samples. Moreover, we apply this approximation to evaluate the bias in the effect estimates of the genetic variants related to mitochondrial proteins a large-scale prostate cancer study.

Binary classification of multichannel-EEG records based on the ϵ-complexity of continuous vector functions.

BACKGROUND AND OBJECTIVE: A crucial step in a classification of electroencephalogram (EEG) records is the feature selection. The feature selection problem is difficult because of the complex structure of EEG signals. To classify the EEG signals with good accuracy, most of the recently published studies have used high-dimensional feature spaces. Our objective is to create a low-dimensional feature space that enables binary classification of EEG records. METHODS: The proposed approach is based on our theory of the ϵ-complexity of continuous functions, which is extended here (see Appendix) to the case of vector functions. This extension permits us to handle multichannel-EEG records. The method consists of two steps. Firstly, we estimate the ϵ-complexity coefficients of the original signal and its finite differences. Secondly, we utilize the random forest (RF) or support vector machine (SVM) classifier. RESULTS: We demonstrated the performance of our method on simulated data. We also applied it to the problem of classification of multichannel-EEG records related to a group of healthy adolescents (39 subjects) and a group of adolescents with schizophrenia (45 subjects). We found that the random forest classifier provides a superior result. In particular, out-of-bag accuracy in the case of RF was 85.3%. Using 10-fold cross-validation (CV), RF gave an average accuracy of 84.5% on a test set, whereas SVM gave an accuracy of 81.07%. We note that the highest accuracy on CV was 89.3%. To compare our method with the classical approach, we performed classification using the spectral features. In this case, the best performance was achieved using seven-dimensional feature space, with an average accuracy of 83.6%. CONCLUSIONS: We developed a model-free method for binary classification of EEG records. The feature space was reduced to four dimensions. The results obtained indicate the effectiveness of the proposed method.

Glycoproteins in Claudin-Low Breast Cancer Cell Lines Have a Unique Expression Profile.

Claudin proteins are components of epithelial tight junctions; a subtype of breast cancer has been defined by the reduced expression of mRNA for claudins and other genes. Here, we characterize the expression of glycoproteins in breast cell lines for the claudin-low subtype using liquid chromatography/tandem mass spectrometry. Unsupervised clustering techniques reveal a group of claudin-low cell lines that is distinct from nonmalignant, basal, and luminal lines. The claudin-low cell lines express F11R, EPCAM, and other proteins at very low levels, whereas CD44 is expressed at a high level. Comparison of mRNA expression to glycoprotein expression shows modest correlation; the best agreement occurs when the mRNA expression level is lowest and little or no protein is detected. These findings from cell lines are compared to those for tumor samples by the Clinical Proteomic Tumor Analysis Consortium (CPTAC). The CPTAC samples contain a group low in CLDN3. The samples low in CLDN3 proteins share many differentially expressed glycoproteins with the claudin-low cell lines. In contrast to the situation for cell lines or patient samples classified as claudin-low by RNA expression, however, most of the tumor samples low in CLDN3 protein express the estrogen receptor or HER2. These tumor samples express CD44 protein at low rather than high levels. There is no correlation between CLDN3 gene expression and protein expression in these CPTAC samples; hence, the claudin-low subtype defined by gene expression is not the same group of tumors as that defined by low expression of CLDN3 protein.

Mining the Breast Cancer Proteome for Predictors of Drug Sensitivity.

Approximately 20 drugs have been approved by the FDA for breast cancer treatment, yet predictive biomarkers are known for only a few of these. The identification of additional biomarkers would be useful both for drugs currently approved for breast cancer treatment and for new drug development. Using glycoprotein expression data collected via mass spectrometry, in conjunction with statistical models constructed by elastic net or lasso regression, we modeled quantitatively the responses of breast cancer cell lines to ~90 drugs. Lasso and elastic net regression identified HER2 as a predictor protein for lapatinib, afatinib, gefitinib and erlotinib, which target HER2 or the EGF receptor, thus providing an internal control for the approach. Two additional protein datasets and two RNA datasets were also tested as sources of predictor proteins for modeling drug sensitivity. Protein expression measured by mass spectrometry gave models with higher coefficients of determination than did reverse phase protein array (RPPA) predictor data. Further, cross validation of the elastic net models shows that, for many drugs, the prediction error is lower when the predictor data is from proteins, rather than mRNA expression measured on microarrays. Drugs that could be modeled effectively include PI3K inhibitors, Akt inhibitors, paclitaxel and docetaxel, rapamycin, everolimus and temsirolimus, gemcitabine and vinorelbine. Strikingly, this modeling approach with protein predictors often succeeds for drugs that are targeted agents, even when the nominal target is not in the dataset.

Mathematical modeling of tumor cell proliferation kinetics and label retention in a mouse model of lung cancer.

Slowly cycling tumor cells that may be present in human tumors may evade cytotoxic therapies, which tend to be more efficient at destroying cells with faster growth rates. However, the proportion and growth rate of slowly cycling tumor cells is often unknown in preclinical model systems used for drug discovery. Here, we report a quantitative approach to quantitate slowly cycling malignant cells in solid tumors, using a well-established mouse model of Kras-induced lung cancer (Kras(G12D/+)). 5-Bromo-2-deoxyuridine (BrdUrd) was administered to tumor-bearing mice, and samples were collected at defined times during pulse and chase phases. Mathematical and statistical modeling of the label-retention data during the chase phase supported the existence of a slowly cycling label-retaining population in this tumor model and permitted the estimation of its proportion and proliferation rate within a tumor. The doubling time of the slowly cycling population was estimated at approximately 5.7 weeks, and this population represented approximately 31% of the total tumor cells in this model system. The mathematical modeling techniques implemented here may be useful in other tumor models where direct observation of cell-cycle kinetics is difficult and may help evaluate tumor cell subpopulations with distinct cell-cycling rates.

Optimal channel selection for analysis of EEG-sleep patterns of neonates.

This paper extends our previous work on automated detection and classification of neonate EEG sleep stages. In [19] we adapted and integrated a range of computational, mathematical and statistical tools for the analysis of neonatal electroencephalogram (EEG) sleep recordings with the aim of facilitating the assessment of neonatal brain maturation and dismaturity by studying the structure and temporal patterns of their sleep. That work relied on algorithms using a single channel of EEG. The present paper builds on our previous work by incorporating a larger selection of EEG channels that capture both the spatial distribution and temporal patterns of EEG during sleep. Using a multivariate analysis approach, we obtain the "optimal" selection of the EEG channels and characteristics that are most suitable for EEG sleep state separation.

Automated detection of neonate EEG sleep stages.

The paper integrates and adapts a range of advanced computational, mathematical and statistical tools for the purpose of analysis of neonate sleep stages based on extensive electroencephalogram (EEG) recordings. The level of brain dysmaturity of a neonate is difficult to assess by direct physical or cognitive examination, but dysmaturity is known to be directly related to the structure of neonatal sleep as reflected in the nonstationary time series produced by EEG signals which, importantly, can be collected trough a noninvasive procedure. In the past, the assessment of sleep EEG structure has often been done manually by experienced clinicians. The goal of this paper is to develop rigorous algorithmic tools for the same purpose by providing a formal scheme to separate different sleep stages corresponding to different stationary segments of the EEG signal based on statistical analysis of the spectral and nonlinear characteristics of the sleep EEG recordings. The methods developed in this paper can, potentially, be translated to other areas of biomedical research.