hide
Free keywords:
-
Abstract:
We define breathomics as the metabolomics study of exhaled air. It is a
strongly emerging metabolomics research field that mainly focuses on
health-related Volatile Organic Compounds (VOCs). Since the composition
of these compounds varies depending on health status, breathomics holds
great promise as non-invasive diagnostic tool. Thusthe main aim of
breathomics is to find the patterns of VOCs relatedto deviant (for instance
inflammatory) metabolic processes occurring e.g. inthe human body.
Consequently, methods for recording VOCs in exhaledair for diagnosis and
monitoring health status gained increased attentionover the last years. As
a result, measuring breath air high-throughput and in high resolution has
enormously developed. Yet machine learning solutions for fingerprinting
VOCs profiles in the breathomics research field arestill in their infancy.
Therefore in this review/tutorial we describe the current state of the art in
data pre-processing and analysis. We start with detailed pre-processing
pipelines for breathomics data obtained from Gas-Chromatography Mass
Spectrometry and Ion Mobility Spectrometer coupled to Multi-Capillary
Columns. The final result of such pipelines is a matrix containing the
relative abundances of a set of VOCs for a group ofpatients under different
conditions (e.g. disease stage, treatment). Independently of the utilized
analytical technique the most important question: �Which VOCs are
discriminatory�, remains the same. Hence, in the main part of our
review/tutorial we focus on several modern machine learning methods
(multivariate statistics). We demonstrate the advantages as well the
drawbacks of such techniques. We aim to help the breath analysis
community to understand when and how one can profitfrom a certain
method. In parallel, we hope to make the community aware of the
existing, yet in breathomics unmet research data fusion methods.
