Bioinformatics

Biography

Biography: Mari van Reenen

Abstract

Statement of the Problem: Variable selection and classification can become complex due to the large number and sources of missing values often present in data, specifically data generated in GC-MS based metabolomics experiments. Missing values are set to zero under certain conditions resulting in a mixture distribution which are difficult for most statistical approaches to account for.

Methodology: ERp is a variable selection and classification method based on minimized classification error rates, from a control and experimental group. ERp tests the null hypothesis that there is no difference between the distributions of the two groups. Significant variables are can discriminate between the two groups and provide insight into the biological mechanisms driving group differences. XERp is an extension of ERp that takes zero-inflated data into account. XERp addresses two sources of zero-valued observations: (i) zeros reflecting the complete absence of a metabolite from a sample (true zeros); and (ii) zeros reflecting a measurement below the detection limit.

Findings: XERp performs well with regard to bias and power. XERp was also applied to a GC-MS dataset from a metabolomics study on tuberculosis meningitis in infants and children and generated a list of discriminatory variables which informed the biological interpretation of the data. XERp also attained satisfactory classification accuracy for previously unseen cases in a leave-one-out cross-validation context.

Conclusion & Significance: XERp is able to identify variables that discriminate between two groups by simultaneously extracting information from the difference in the proportion of zeros and shifts in the distributions of the non-zero observations. XERp uses simple rules to classify new subjects and a weight pair to adjust for unequal sample sizes or sensitivity and specificity requirements.

Recent Publications

1. Van Reenen, M Westerhuis J A, Reinecke C J and Venter J H (2017) Metabolomics variable selection and classification in the presence of observations below the detection limit using an extension of ERp. BMC Bioinformatics, 18(1):83.

2. Irwin C, van Reenen M, Mason S, Mienie L J, Westerhuis J A, and Reinecke C J (2016) Contribution towards a metabolite profile of the detoxification of benzoic acid through glycine conjugation: an intervention study. PloS one. 11(12): e0167309. 

3. Van Reenen M Reinecke C J, Westerhuis J A and Venter J H (2016) Variable selection for binary classification using error rate p-values applied to metabolomics data. BMC Bioinformatics. 17(1):33.

4. Moutloatse G P, Bunders M J, van Reenen M, Mason S Kuijpers, T W Engelke U F and Reinecke C J (2016) Metabolic risks at birth of neonates exposed in utero to HIV-antiretroviral therapy relative to unexposed neonates: an NMR metabolomics study of cord blood. Metabolomics. 12(11):175.

5. Mason S, van Furth A M T, Solomons R, Wevers R A, van Reenen M and Reinecke C J (2016) A putative urinary biosignature for diagnosis and follow-up of tuberculous meningitis in children: outcome of a metabolomics study disclosing host–pathogen responses. Metabolomics, 12(7):1-16.