Introduction to metabolomics

Jessica Cooperstone

Introductions 👋

• Name
• Research focus area
• How you see yourself using metabolomics
• What you hope to learn

What will we cover in this course?

Lectures

• Introduction (this lecture)
• Study design and sample collection
• LC-MS data acquisition and pre-processing
• Data analysis
• Compound ID

Hands-on activities

• Design your own experiment
• LC-MS data processing with MZmine
• Data analysis with MetaboAnalyst

Course logistics

• Let me know if you want to use the Ohio Supercomputer Center for deconvolution with MZmine
• Other logistics

What is metabolomics?

• The study of the totality of small molecules (< 1500 Da) within a given system
• Metabolomics is a tool that allows us to study global metabolism

Metabolites are the downstream products of the system biology cascade

The food metabolome can be influenced by:

• Variety/genetics
• Environment
• Post-harvest/processing
• Storage

The metabolome is really BIG!

Adapted from David Wishart’s Canadian Bioinformatics Workshop

The metabolome is very chemically diverse

From Roche, Biochemical Pathways

The metabolome is constantly changing

How is metabolomics different from targeted analyses?

Metabolomics:

• 100s-1,000s of analytes
• Work on the back end
• Comparative (i.e. relative concentration)

Targeted analyses:

• 1-20 analytes
• Work on the front end
• Quantitative (i.e. absolute concentration)

Metabolomics workflow

Metabolomics is a comparative analysis

• What can food scientists use metabolomics for?
• If you have specific compounds of interest, develop a targeted method!

What do we want to compare?

• It’s critical to select comparable samples as our approach is comparative.
• Foods: plants, animal products, raw ingredients, finished product
• Biological sample: plasma, urine, tissue, other fluids, cells

Preparation dictates compounds detected

• You can only detect what you present to an instrument for analysis
• Sample prep depends on intended method of analysis (e.g., water extraction, polar compounds; non-polar extraction, non-polar compounds)
• Dilute, centrifuge/filter, inject (e.g. urine, juice, olive oil)

Collect comprehensive metabolite data

3 most popular methods for analysis:

• Liquid-chromatography, mass spectrometry (LC-MS)
• Gas chromatpgrahy, MS (GC-MS)
• Nuclear magnetic resonance spectroscopy (NMR)

All methods have benefits and drawbacks

Convert spectral data into feature table

• From raw spectra, ions are selected, chromatograms drawn, peaks detected, masses and retention times aligned, features dereplicated
• Result is a data file that includes m/z, retention time, compound identifier (usually mz_rt), and relative abundance of each feature in each sample
• With MZmine, samples are columns, features are rows

Use statistics and chemometrics to understand group differences

• Significance testing (e.g., t-test, Wilcoxon rank sum test, ANOVA)
• Unsupervised analyses (e.g., PCA, hierarchical clustering)
• Supervised analyses (e.g., PLS-DA or PLS-R, random forest)

What metabolites do we have?

• Searching publicly available databases (e.g., HMDB, Mass Bank of North America (MoNA), GNPS) at the MS1 and MS2 level
• Conduct MS/MS experiments
• Comparison with authentic standards

Putting findings into a broader context

• Understanding which metabolic pathways are most deregulated
• Typically for enzymatic pathways
• Requires compound IDs (a big limitation)

Ensure findings are real and reproducible

• Mass spectrometry is not inherently quantitative (i.e., if the intensity of analyte A is higher than analyte B, it doesn’t necessarily mean there is more of A than B)
• Knowing the absolute concentration allows comparison with literature/other data
• Validation in a separate sample set ensures robustness

Metabolomics workflow