Identification of metabolomic biomarkers

Question

[Na Sed]

Currently I am working in the field of metabolic network analysis and would like to extend my research to identification of metabolomics biomarkers. I am not familiar with different types of metablomics data and try to find text as well as research papers in this area. I have searched a lot but I have not yet found suitable reference. Additionally, I do not know any well-known database (something similar to TCGA) in this field that I can obtain metabolomics data. I would really appreciated if you could introduce me databases, papers, or text books to me.

Answer 1

[Kevin Blighe]

Metabolomics has been around for a very long time, primarily amongst biochemists, but is only recently gaining interest amongst non-biochemists as a means of research. Obviously, it has long been used in hospital clinical laboratory testing. However, there is still skepticism amongst some about the utility of metabolomics in terms of disease biomarkers. Some appear to dismiss it because they know nothing about it, whilst others dismiss it due to the huge challenges that the field faces, challenges which ironically inspire others to continue research in this particular area.

Whilst I am not an expert in this area, I have worked in it at postdoctoral level in Boston, USA. I can therefore go over the basics that anyone entering the field should know. I have also published 2 papers (so far) in this area:

• Vitamin D prenatal programming of childhood metabolomics profiles at age 3 y.
• Applications of metabolomics in the study and management of preeclampsia: a review of the literature

Apart from individual tests for specific metabolites, the high throughput methods include nuclear magnetic resonance (NMR) and liquid chromatography-tandem mass spectrometry (LC-MS). I am not overly familiar with NMR, but LC-MS is run four times with different configurations to identify different classes of metabolites:

1. HILIC-positive platform: Amines and polar metabolites that ionise in the positive ion mode using hydrophilic interaction liquid chromatography (HILIC) and MS analyses
2. HILIC-negative platform: Central metabolites and polar metabolites that ionise in the negative ion mode using HILIC chromatography with an amine column and targeted MS
3. C8 platform: Polar and non-polar lipids using reverse phase chromatography and full scan MS
4. C18 platform: Free fatty acids, bile acids, and metabolites of intermediate polarity using reversed chromatography with a T3 UPLC column (C18 chromatography) and MS analyses in the negative ion mode

Metabolites are then identified by examining their mass-to-charge (m/z) ratio and retention time (rt) on the chromatographic output. The experiment will be spiked with known standards.

----------------------------------

People

The people of whose research you should be aware in this area include:

• Hector Kuhn
• Karsten Suhr
• Clary Clish / Jessica Lasky-Su

--------------------------------------

Publications

Some major publications in this area about which you should be aware:

• Whole-genome sequencing identifies common-to-rare variants associated with human blood metabolites.
• Metabolomics: an emerging but powerful tool for precision medicine
• Genetic variation in metabolic phenotypes: study designs and applications.
• A genome-wide association study of metabolic traits in human urine.

--------------------------------------

Databases / analysis tools

In terms of databases, the main one that is becoming more standardised (for naming) is HMDB: The Human Metabolome Database. However, the majority of metabolites that can be detected still don't have a HMDB ID because we simply don't know what they are. Another database that you'll come across is CAS, although this is more geared toward chemistry, generally, and not quite metabolomics.

A very popular online analysis suite that has grown in popularity is MetaboAnalyst. I encourage you to explore its features.

--------------------------------------

Challenges and network analysis below in a comment

Kevin

[Kevin Blighe]

Challenges:

• Metabolite samples require much care and degrade rapidly. They also degrade at different rates depending on the type of metabolite: for example, fatty acids will degrade more rapidly than amino acids or other misc compounds.
• Identification of metabolites, as the vast majority (90+%) that can be detected (in blood) remain unknown
• Correct separation of individual metabolites from spectrometer output (sometimes what looks like 2 metabolites on the output may actually be one)
• Mapping of metabolites back to their genetic / DNA origins
• Mapping of metabolites to pathways
• Analysis methods are not perfected. There is no standard way of normalising data and it suffers from large technical and sampling variability. For example: if you take a sample in the morning and then one in the evening, they may have entirely different profiles due to diet and just general human metabolic activity. Thus, experiments require much better design than, say, GWAS studies, in order to reduce confounding factors. The data also suffers from high missingness, which typically has to be imputed or excluded.

Network analysis

As to which I alluded in the challenges, we simply don't know much about the majority of metabolites, nor do we know to which pathways they belong, etc (even for the known ones). In a typical experiment, one would identify dozens of, for example, fatty acids with different length carbon chains. These may be part of separate biochemical processes or part of the same biochemical cascade... impossible to know. One can easily build a correlation network of the data but it may be entirely meaningless. My feeling is that, in the future, we will have to build metabolite networks using known biochemical pathways as a 'scaffold'. For example, we know (very well) pathways such as citric acid synthesis, vitamin D metabolism, etc. One could take that information and use it as a scaffold for the purposes of building metabolite networks. Without these types of guides, I fear that network analysis in metabolomics will be just meaningless.

[Na Sed]

Thank you very much. Your response is a great help. :)

[Kevin Blighe]

No problem. Could have written a fair bit more!

[Na Sed]

Hi Kevin, I am looking for a data set containing concentration of metabolites in two groups of samples (normal vs. patient) to test my own biomarker analysis algorithm. Is there any data set available to download? Thanks.

[Kevin Blighe]

Hello again. In relation to that, there is a previous answer here that was unfortunately not up-voted: Publicly-available Metabolomics Data

I have now just given it a up-vote.

[Na Sed]

Thank you very much.

Answer 2

[Buffo]

You can use MetaboAnalyst to analyze your data, and also to look for biomarkers.