Omics-Based Strategies in Precision Medicine: Toward a Paradigm Shift in Inborn Errors of Metabolism Investigations

Abstract

The rise of technologies that simultaneously measure thousands of data points represents the heart of systems biology. These technologies have had a huge impact on the discovery of next-generation diagnostics, biomarkers, and drugs in the precision medicine era. Systems biology aims to achieve systemic exploration of complex interactions in biological systems. Driven by high-throughput omics technologies and the computational surge, it enables multi-scale and insightful overviews of cells, organisms, and populations. Precision medicine capitalizes on these conceptual and technological advancements and stands on two main pillars: data generation and data modeling. High-throughput omics technologies allow the retrieval of comprehensive and holistic biological information, whereas computational capabilities enable high-dimensional data modeling and, therefore, accessible and user-friendly visualization. Furthermore, bioinformatics has enabled comprehensive multi-omics and clinical data integration for insightful interpretation. Despite their promise, the translation of these technologies into clinically actionable tools has been slow. In this review, we present state-of-the-art multi-omics data analysis strategies in a clinical context. The challenges of omics-based biomarker translation are discussed. Perspectives regarding the use of multi-omics approaches for inborn errors of metabolism (IEM) are presented by introducing a new paradigm shift in addressing IEM investigations in the post-genomic era.

Keywords: bioinformatics; biomarkers; chemometrics; data integration; inborn errors of metabolism; machine learning; mass spectrometry; next-generation sequencing; omics; precision medicine.

Figure 1

Multi-scale biology overview of systems medicine. Three main drivers define phenotype: (i) the molecular phenome, which is defined by the underlying molecular supports of biological information. The different omics strategies enable to interrogate these supports for information retrieval; (ii) environmental effects spanning from exposures to toxic substances or drugs to diet define the exposome; and (iii) the different clinical metrics used to define the clinical phenome. These different biological and clinical metrics should be approached in a multi-dimensional fashion and should take into account the inherent spatial and temporal scales of both measurement technologies and disease dynamics from the molecular to the population level.

Figure 2

Schematic illustration of bioinformatics pipelines in next-generation sequencing (NGS) and mass spectrometry (MS)-based omics. Left: The NGS pipeline comprises library construction and capture, sequencing reaction, and signal processing. Then, a base-calling step is performed to define the unaligned nucleotide sequence. The data are stored in FASTAQ file format containing quality scores. Subsequently, read alignment to a reference sequence is performed, followed by variant calling and annotation. The final output is a list of variants in VCF format for visualization and interpretation; Right: MS pipeline starts with sample preparation, depending on the MS instruments and the combined separation method. Data acquisition is performed according to the chosen mode (full scan or tandem MS). Subsequently, a pre-processing step is needed for feature extraction and data cleaning. The result is a list of features that will undergo data analysis, molecular annotation, and identification before biological interpretation. Signal processing is platform-dependent in NGS; however, open source solutions are available for pre-processing MS data.

Figure 3

Biomarker development pipeline milestones.

Figure 4

Illustration of the two main machine learning techniques on which omics-based biomarker strategies rely. Left: All samples are unlabeled in unsupervised learning. A model separates samples into different clusters based on their biological similarity. A new sample (red circle) is classified according to its similarity to a particular cluster; Right: In supervised learning, a training dataset of samples with known class labels is used to build a model (blue circle for condition 1 and green circle for condition 2). The model maximizes the difference between samples from condition 1 and condition 2. Based on this learning, a label for a new sample (red circle) is determined.

Figure 5

A stepwise approach to using machine learning methods for the prediction of clinical phenotypes. A training dataset is first collected. Then, a subset of features associated with the phenotype of interest is selected. Based on these features, a multi-variate model is built by the training data. A validation set acquired using the same omics profiling methods is collected and treated as new input to the established multi-variate model. The predictions provided by the model are used to assess the classification performance of the test input by comparing the model output and the actual clinical phenotypes of the patients in the validation set.

Figure 6

Paradigm shift in Inborn Errors of Metabolism (IEM) diagnosis workflow. Laboratory workflow using high-throughput analytical technologies, integrative bioinformatics, and computational frameworks recovers molecular information for more effective medical decision-making.