Haystack, a web-based tool for metabolomics research.

Stephen C Grace,Heng Luo,Stephen Embry

doi:10.1186/1471-2105-15-s11-s12

Abstract

BackgroundLiquid chromatography coupled to mass spectrometry (LCMS) has become a widely used technique in metabolomics research for differential profiling, the broad screening of biomolecular constituents across multiple samples to diagnose phenotypic differences and elucidate relevant features. However, a significant limitation in LCMS-based metabolomics is the high-throughput data processing required for robust statistical analysis and data modeling for large numbers of samples with hundreds of unique chemical species.ResultsTo address this problem, we developed Haystack, a web-based tool designed to visualize, parse, filter, and extract significant features from LCMS datasets rapidly and efficiently. Haystack runs in a browser environment with an intuitive graphical user interface that provides both display and data processing options. Total ion chromatograms (TICs) and base peak chromatograms (BPCs) are automatically displayed, along with time-resolved mass spectra and extracted ion chromatograms (EICs) over any mass range. Output files in the common .csv format can be saved for further statistical analysis or customized graphing. Haystack's core function is a flexible binning procedure that converts the mass dimension of the chromatogram into a set of interval variables that can uniquely identify a sample. Binned mass data can be analyzed by exploratory methods such as principal component analysis (PCA) to model class assignment and identify discriminatory features. The validity of this approach is demonstrated by comparison of a dataset from plants grown at two light conditions with manual and automated peak detection methods. Haystack successfully predicted class assignment based on PCA and cluster analysis, and identified discriminatory features based on analysis of EICs of significant bins.ConclusionHaystack, a new online tool for rapid processing and analysis of LCMS-based metabolomics data is described. It offers users a range of data visualization options and supports non-biased differential profiling studies through a unique and flexible binning function that provides an alternative to conventional peak deconvolution analysis methods.

Highlights

Untargeted metabolomics has become an increasingly powerful tool to investigate biological systems [1,2,3]. This approach typically employs gas or liquid chromatography combined with mass spectrometry or nuclear magnetic resonance to survey the metabolome and identify features associated with the genotype and/or biological state of the organism [4,5]
We have found the online tool MetaboAnalyst 2.0 to be useful for analysis of binned mass data since it provides a range of data normalization options as well as a full suite of statistical processing functions [16,22]
We show that bin analysis can be used as a fast and practical alternative to peak detection methods to streamline and simplify the analysis of Liquid chromatography coupled to mass spectrometry (LCMS)-based metabolomic data

Summary

Introduction

Untargeted metabolomics has become an increasingly powerful tool to investigate biological systems [1,2,3]. Given the enormous range and variability associated with metabolomic data, identifying a distinctive but weak signal from the data pool presents a formidable challenge [11]. Manual identification of these metabolites can be a cumbersome and error-prone task when dealing with large metabolomic studies that involve multiple files and groups. Liquid chromatography coupled to mass spectrometry (LCMS) has become a widely used technique in metabolomics research for differential profiling, the broad screening of biomolecular constituents across multiple samples to diagnose phenotypic differences and elucidate relevant features. A significant limitation in LCMS-based metabolomics is the high-throughput data processing required for robust statistical analysis and data modeling for large numbers of samples with hundreds of unique chemical species

Objectives

Methods

Results

Discussion

Conclusion