Omics–Bioinformatics in the Context of Clinical Data

Abstract

The Omics revolution has provided the researcher with tools and methodologies for qualitative and quantitative assessment of a wide spectrum of molecular players spanning from the genome to the meta-bolome level. As a consequence, explorative analysis (in contrast to purely hypothesis driven research procedures) has become applicable. However, numerous issues have to be considered for deriving meaningful results from Omics, and bioinformatics has to respect these in data analysis and interpretation. Aspects include sample type and quality, concise definition of the (clinical) question, and selection of samples ideally coming from thoroughly defined sample and data repositories. Omics suffers from a principal shortcoming, namely unbalanced sample-to-feature matrix denoted as "curse of dimensionality", where a feature refers to a specific gene or protein among the many thousands assayed in parallel in an Omics experiment. This setting makes the identification of relevant features with respect to a phenotype under analysis error prone from a statistical perspective. From this sample size calculation for screening studies and for verification of results from Omics, bioinformatics is essential. Here we present key elements to be considered for embedding Omics bioinformatics in a quality controlled workflow for Omics screening, feature identification, and validation. Relevant items include sample and clinical data management, minimum sample quality requirements, sample size estimates, and statistical procedures for computing the significance of findings from Omics bioinformatics in validation studies.

Genome-wide studies to identify risk factors for kidney disease with a focus on patients with diabetes

“…Alternatively to the use of classical regression models and an adjusted P-value, there are more complex approaches such as lasso, ridge regression or elastic net to account for the huge abundance of predicting variables compared with the number of observed subjects [19,20].…”

Multi-subgroup Gene Screening Using Semi-parametric Hierarchical Mixture Models and the Optimal Discovery Procedure: Application to a Randomized Clinical Trial in Multiple Myeloma

“…Screening of genes associated with a phenotype of interest is one of fundamental elements in many genomic studies to understand biological mechanisms. Standard statistical analysis for gene screening is to apply multiple tests regarding the association of individual genes with the phenotypic variable (Speed, ; McLachlan et al, 2004; Simon et al, ; Mayer, ). In such an analysis, it is often the case that there are some important subgroups of samples or conditions and one is interested in whether the profile of association with the phenotype is different across the subgroups at the gene level.…”

“…Distinguishing genes whose association profiles are similar from those different across subgroups may facilitate deeper understanding of the biological mechanisms across subgroups. Typically, the difference in association profiles across subgroups is investigated via a regression model that associates the phenotypic variable with covariates representing genes, subgroups, and their interactions, where some parametric form for covariates’ effects (e.g., linear form) is specified to capture differential association profiles (Speed, ; McLachlan et al, 2004; Simon et al, ; Mayer, ).…”

BcCluster: A Bladder Cancer Database at the Molecular Level

“…Though omics profiling has provided an abundance of data, technical boundaries involving incompleteness of the individual molecular datasets together with the static representation of cellular activity limit the insights on molecular processes and their interaction dynamics [ 32–34 ]. A large number of biological pathway analysis tools are available, including KEGG [ 35 ], PANTHER [ 36 ], REACTOME [ 37 ] and AmiGO [ 38 ] described in PathGuide ( http://www.pathguide.org/ ), and allow detection of significant metabolic and signaling pathways.…”