We illustrate the power of the method using two cancer data sets. In both cases, we can quickly and accurately classify test samples from any number of specified a priori groups and identify the genes which characterize these groups. We obtained very high rates of correct classification, as determined by jack-knife or validation experiments with training and test sets. The results are comparable to those from other methods in terms of accuracy but the power and flexibility of BGA make it an especially attractive method for the analysis of microarray cancer data.

No abstract

Despite the proposal of minimum reporting guidelines for metabolomics over a decade ago, reporting on the data analysis step in metabolomics studies has been shown to be unclear and incomplete. Major omissions and a lack of logical flow render the data analysis’ sections in metabolomics studies impossible to follow, and therefore replicate or even imitate. Here, we propose possible reasons why the original reporting guidelines have had poor adherence and present an approach to improve their uptake. We present in this paper an R markdown reporting template file that guides the production of text and generates workflow diagrams based on user input. This R Markdown template contains, as an example in this instance, a set of minimum information requirements specifically for the data pre-treatment and data analysis section of biomarker discovery metabolomics studies, (gleaned directly from the original proposed guidelines by Goodacre at al). These minimum requirements are presented in the format of a questionnaire checklist in an R markdown template file. The R Markdown reporting template proposed here can be presented as a starting point to encourage the data analysis section of a metabolomics manuscript to have a more logical presentation and to contain enough information to be understandable and reusable. The idea is that these guidelines would be open to user feedback, modification and updating by the metabolomics community via GitHub.

Unmet clinical diagnostic needs exist for many complex diseases, which (it is hoped) will be solved by the discovery of metabolomics biomarkers. However, at present, no diagnostic tests based on metabolomics have yet been introduced to the clinic. This review is presented as a research perspective on how data analysis methods in metabolomics biomarker discovery may contribute to the failure of biomarker studies and suggests how such failures might be mitigated. The study design and data pretreatment steps are reviewed briefly in this context, and the actual data analysis step is examined more closely.

The aim of this preliminary study was to investigate the potential of maternal serum to provide metabolomic biomarker candidates for the prediction of spontaneous preterm birth (SPTB) in asymptomatic pregnant women at 15 and/or 20 weeks’ gestation. Metabolomics LC-MS datasets from serum samples at 15- and 20-weeks’ gestation from a cohort of approximately 50 cases (GA < 37 weeks) and 55 controls (GA > 41weeks) were analysed for candidate biomarkers predictive of SPTB. Lists of the top ranked candidate biomarkers from both multivariate and univariate analyses were produced. At the 20 weeks’ GA time-point these lists had high concordance with each other (85%). A subset of 4 of these features produce a biomarker panel that predicts SPTB with a partial Area Under the Curve (pAUC) of 12.2, a sensitivity of 87.8%, a specificity of 57.7% and a p-value of 0.0013 upon 10-fold cross validation using PanelomiX software. This biomarker panel contained mostly features from groups already associated in the literature with preterm birth and consisted of 4 features from the biological groups of “Bile Acids”, “Prostaglandins”, “Vitamin D and derivatives” and “Fatty Acids and Conjugates”.