Multidimensionality of microarrays: statistical challenges and (im)possible solutions

What did we learn from multiple omics studies in asthma?

More than a decade has passed since the finalization of the Human Genome Project. Omics technologies made a huge leap from trendy and very expensive to routinely executed and relatively cheap assays. Simultaneously, we understood that omics is not a panacea for every problem in the area of human health and personalized medicine. Whilst in some areas of research omics showed immediate results, in other fields, including asthma, it only allowed us to identify the incredibly complicated molecular processes. Along with their possibilities, omics technologies also bring many issues connected to sample collection, analyses and interpretation. It is often impossible to separate the intrinsic imperfection of omics from asthma heterogeneity. Still, many insights and directions from applied omics were acquired-presumable phenotypic clusters of patients, plausible biomarkers and potential pathways involved. Omics technologies develop rapidly, bringing improvements also to asthma research. These improvements, together with our growing understanding of asthma subphenotypes and underlying cellular processes, will likely play a role in asthma management strategies.

The use of pharmacogenomics, epigenomics, and transcriptomics to improve childhood asthma management: Where do we stand?

Asthma is a complex multifactorial disease and it is the most common chronic disease in children. There is a high variability in response to asthma treatment, even in patients with good adherence to maintenance treatment, and a correct inhalation technique. Distinct underlying disease mechanisms in childhood asthma might be the reason of this heterogeneity. A deeper knowledge of the underlying molecular mechanisms of asthma has led to the recent development of advanced and mechanism-based treatments such as biologicals. However, biologicals are recommended only for patients with specific asthma phenotypes who remain uncontrolled despite high dosages of conventional asthma treatment. One of the main unmet needs in their application is lack of clinically available biomarkers to individualize pediatric asthma management and guide treatment. Pharmacogenomics, epigenomics, and transcriptomics are three omics fields that are rapidly advancing and can provide tools to identify novel asthma mechanisms and biomarkers to guide treatment. Pharmacogenomics focuses on variants in the DNA, epigenomics studies heritable changes that do not involve changes in the DNA sequence but lead to alteration of gene expression, and transcriptomics investigates gene expression by studying the complete set of mRNA transcripts in a cell or a population of cells. Advances in high-throughput technologies and statistical tools together with well-phenotyped patient inclusion and collaborations between different centers will expand our knowledge of underlying molecular mechanisms involved in disease onset and progress. Furthermore, it could help to select and stratify appropriate therapeutic strategies for subgroups of patients and hopefully bring precision medicine to daily practice.

A computational framework for complex disease stratification from multiple large-scale datasets

BACKGROUND

Multilevel data integration is becoming a major area of research in systems biology. Within this area, multi-'omics datasets on complex diseases are becoming more readily available and there is a need to set standards and good practices for integrated analysis of biological, clinical and environmental data. We present a framework to plan and generate single and multi-'omics signatures of disease states.

METHODS

The framework is divided into four major steps: dataset subsetting, feature filtering, 'omics-based clustering and biomarker identification.

RESULTS

We illustrate the usefulness of this framework by identifying potential patient clusters based on integrated multi-'omics signatures in a publicly available ovarian cystadenocarcinoma dataset. The analysis generated a higher number of stable and clinically relevant clusters than previously reported, and enabled the generation of predictive models of patient outcomes.

CONCLUSIONS

This framework will help health researchers plan and perform multi-'omics big data analyses to generate hypotheses and make sense of their rich, diverse and ever growing datasets, to enable implementation of translational P4 medicine.

Statistical controversies in clinical research: prognostic gene signatures are not (yet) useful in clinical practice

With the genomic revolution and the era of targeted therapy, prognostic and predictive gene signatures are becoming increasingly important in clinical research. They are expected to assist prognosis assessment and therapeutic decision making. Notwithstanding, an evidence-based approach is needed to bring gene signatures from the laboratory to clinical practice. In early breast cancer, multiple prognostic gene signatures are commercially available without having formally reached the highest levels of evidence-based criteria. We discuss specific concepts for developing and validating a prognostic signature and illustrate them with contemporary examples in breast cancer. When a prognostic signature has not been developed for predicting the magnitude of relative treatment benefit through an interaction effect, it may be wishful thinking to test its predictive value. We propose that new gene signatures be built specifically for predicting treatment effects for future patients and outline an approach for this using a cross-validation scheme in a standard phase III trial. Replication in an independent trial remains essential.

Bayesian clinical classification from high-dimensional data: Signatures versus variability

When data exhibit imbalance between a large number d of covariates and a small number n of samples, clinical outcome prediction is impaired by overfitting and prohibitive computation demands. Here we study two simple Bayesian prediction protocols that can be applied to data of any dimension and any number of outcome classes. Calculating Bayesian integrals and optimal hyperparameters analytically leaves only a small number of numerical integrations, and CPU demands scale as O(nd). We compare their performance on synthetic and genomic data to the mclustDA method of Fraley and Raftery. For small d they perform as well as mclustDA or better. For d = 10,000 or more mclustDA breaks down computationally, while the Bayesian methods remain efficient. This allows us to explore phenomena typical of classification in high-dimensional spaces, such as overfitting and the reduced discriminative effectiveness of signatures compared to intra-class variability.

Distance-based classifiers as potential diagnostic and prediction tools for human diseases

Typically, gene expression biomarkers are being discovered in course of high-throughput experiments, for example, RNAseq or microarray profiling. Analytic pipelines that extract so-called signatures suffer from the "Dimensionality curse": the number of genes expressed exceeds the number of patients we can enroll in the study and use to train the discriminator algorithm. Hence, problems with the reproducibility of gene signatures are more common than not; when the algorithm is executed using a different training set, the resulting diagnostic signature may turn out to be completely different.

In this paper we propose an alternative novel approach which takes into account quantifiable expression levels of all genes assayed. In our analysis, the cumulative gene expression pattern of an individual patient is represented as a point in the multidimensional space formed by all gene expression profiles assayed in given system, where the clusters of "normal samples" and "affected samples" and defined. The degree of separation of the given sample from the space occupied by "normal samples" reflects the drift of the sample away from homeostasis in the course of development of the pathophysiological process that underly the disease. The outlined approach was validated using the publicly available glioma dataset deposited in Rembrandt and associated with survival data. Additionally, the applicability of the distance analysis to the classification of non-malignant sampled was tested using psoriatic lesions and non-lesional matched controls as a model.

Keywords: biomarkers; clustering; human diseases; RNA

P, Anderson KC, Li C, Avet-Loiseau H, Munshi NC. Gene expression profile alone is inadequate in predicting complete response in multiple myeloma

With advent of several treatment options in multiple myeloma (MM), a selection of effective regimen has become an important issue. Use of gene expression profile (GEP) is considered an important tool in predicting outcome; however, it is unclear whether such genomic analysis alone can adequately predict therapeutic response. We evaluated the ability of GEP to predict complete response (CR) in MM. GEP from pretreatment MM cells from 136 uniformly treated MM patients with response data on an IFM, France led study were analyzed. To evaluate variability in predictive power due to microarray platform or treatment types, additional data sets from three different studies (n=511) were analyzed using same methods. We used several machine learning methods to derive a prediction model using training and test subsets of the original four data sets. Among all methods employed for GEP-based CR predictive capability, we got accuracy range of 56-78% in test data sets and no significant difference with regard to GEP platforms, treatment regimens or in newly diagnosed or relapsed patients. Importantly, permuted P-value showed no statistically significant CR predictive information in GEP data. This analysis suggests that GEP-based signature has limited power to predict CR in MM, highlighting the need to develop comprehensive predictive model using integrated genomics approach.

Gene expression profiling for targeted cancer treatment

INTRODUCTION

There is certain degree of frustration and discontent in the area of microarray gene expression data analysis of cancer datasets. It arises from the mathematical problem called 'curse of dimensionality,' which is due to the small number of samples available in training sets, used for calculating transcriptional signatures from the large number of differentially expressed (DE) genes, measured by microarrays. The new generation of causal reasoning algorithms can provide solutions to the curse of dimensionality by transforming microarray data into activity of a small number of cancer hallmark pathways. This new approach can make feature space dimensionality optimal for mathematical signature calculations.

AREAS COVERED

The author reviews the reasons behind the current frustration with transcriptional signatures derived from DE genes in cancer. He also provides an overview of the novel methods for signature calculations based on differentially variable genes and expression regulators. Furthermore, the authors provide perspectives on causal reasoning algorithms that use prior knowledge about regulatory events described in scientific literature to identify expression regulators responsible for the differential expression observed in cancer samples.

EXPERT OPINION

The author advocates causal reasoning methods to calculate cancer pathway activity signatures. The current challenge for these algorithms is in ensuring quality of the knowledgebase. Indeed, the development of cancer hallmark pathway collections, together with statistical algorithms to transform activity of expression regulators into pathway activity, are necessary for causal reasoning to be used in cancer research.

Challenges and prospects for biomarker research: a current perspective from the developing world

Majority of deaths due to communicable and non-communicable diseases occur in the low and middle-income nations (LMNs), mainly due to the lack of early diagnoses and timely treatments. In such a scenario, biomarkers serve as an indispensible resource that can be used as indicators of biological processes, specific disease conditions or response to therapeutic interventions. Evaluation, diagnosis and management of diseases in developing world by following/extrapolating the findings obtained on the basis of the research work involving only the populations from the developed countries, could often be highly misleading due to existence of diverse patterns of diseases in developing countries compared to the developed world. Biomarker candidates identified from high-throughput integrated omics technologies have promising potential; however, their actual clinical applications are found to be limited, primarily due to the challenges of disease heterogeneity and pre-analytical variability associated with the biomarker discovery pipeline. Additionally, in the developing world, economic crunches, lack of awareness and education, paucity of biorepositories, enormous diversities in socio-epidemiological background, ethnicity, lifestyle, diet, exposure to various environmental risk factors and infectious agents, and ethical and social issues also cumulatively hinder biomarker discovery ventures. Establishment of standard operating procedures, comprehensive data repositories and exchange of scientific findings are crucial for reducing the variability and fragmentation of data. This review highlights the challenges associated with the discovery, validation and translational phases of biomarker research in LMNs with some of their amenable solutions and future prospects. This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge.

Genome-wide assessment in Escherichia coli reveals time-dependent nanotoxicity paradigms

The use of engineered nanomaterials (eNM) in consumer and industrial products is increasing exponentially. Our ability to rapidly assess their potential effects on human and environmental health is limited by our understanding of nanomediated toxicity. High-throughput screening (HTS) enables the investigation of nanomediated toxicity on a genome-wide level, thus uncovering their novel mechanisms and paradigms. Herein, we investigate the toxicity of zinc-containing nanomaterials (Zn-eNMs) using a time-resolved HTS methodology in an arrayed Escherichia coli genome-wide knockout (KO) library. The library was screened against nanoscale zerovalent zinc (nZn), nanoscale zinc oxide (nZnO), and zinc chloride (ZnCl(2)) salt as reference. Through sequential screening over 24 h, our method identified 173 sensitive clones from diverse biological pathways, which fell into two general groups: early and late responders. The overlap between these groups was small. Our results suggest that bacterial toxicity mechanisms change from pathways related to general metabolic function, transport, signaling, and metal ion homeostasis to membrane synthesis pathways over time. While all zinc sources shared pathways relating to membrane damage and metal ion homeostasis, Zn-eNMs and ZnCl(2) displayed differences in their sensitivity profiles. For example, ZnCl(2) and nZnO elicited unique responses in pathways related to two-component signaling and monosaccharide biosynthesis, respectively. Single isolated measurements, such as MIC or IC(50), are inadequate, and time-resolved approaches utilizing genome-wide assays are therefore needed to capture this crucial dimension and illuminate the dynamic interplay at the nano-bio interface.

Multidimensionality of microarrays: statistical challenges and (im)possible solutions

A typical array experiment yields at least tens of thousands of measurements on often not more than a hundred patients, a situation often denoted as the curse of dimensionality. With a focus on prognostic multi-biomarker scores derived from microarrays, we highlight the multidimensionality of the problem and the issues in the multidimensionality of the data. We go over several statistical challenges raised by this curse occurring in each step of microarray analysis on patient data, from the hypothesis and the experimental design to the analysis methods, interpretation of results and clinical utility. Different analytical tools and solutions to answer these challenges are provided and discussed.