In modern laboratory research, the generation and analysis of large amounts of data are becoming increasingly common. This is particularly true in fields such as genomics, proteomics, and metabolomics, where experiments can produce thousands of features per sample. While this wealth of data can provide valuable insights, it also presents a major challenge: how to effectively visualize and analyze these complex and high-dimensional datasets.

One approach that has gained popularity in recent years is the use of multivariate techniques, such as principal component analysis (PCA) or t-SNE, to reduce the dimensionality of data and visualize patterns or clustering in a 2D or 3D scatterplot. Another promising technique in this area is UMAP (Uniform Manifold Approximation and Projection), a dimension reduction method that is particularly well-suited for visualizing complex datasets and preserving local structure.

So how does UMAP work? In simple terms, UMAP takes each sample's unique "fingerprint" and spreads them out in a 2D or 3D scatterplot. Samples that are close to each other in the scatterplot have similar fingerprints, while samples that are further away have different profiles. This allows researchers to easily identify abnormal samples, which can be useful in troubleshooting laboratory issues.

In the context of laboratory research, UMAP can be a powerful tool for identifying and solving issues with experiments, particularly when it comes to detecting abnormal samples. This is because UMAP can effectively capture patterns and relationships in the data, allowing for the easy identification of samples that are distinct from the rest of the dataset.

For example, in a metabolomics experiment, abnormal samples could be identified by their unique metabolic fingerprint, which can be visualized using UMAP. By analyzing the fingerprints of these abnormal samples, researchers can gain valuable insights into potential issues with the experiment, such as contamination, technical problems with the instrumentation, or issues with sample preparation.

Additionally, UMAP can also be used to compare the fingerprints of different groups of samples, such as control and treatment groups, or different conditions or timepoints. This can help researchers identify differences in metabolic pathways or identify metabolic biomarkers that are associated with a particular condition.

Let's take one example that is close to my heart since it relates to the work of my graduate studies. When running my samples in triplicate on the two-dimensinoal gas chromatography coupled to time of flight mass spectrometry (say that 10x fast), I'd often have an issue by the third sample. What would happen is that the concentrations of that sample would be way off by the third triplicate. What I didn't know was how the fingerprint was changing on account of this change in response. So what could be happening between these runs? Well experienced analytical chemists would tell you that there is a problem with the injection, but often there isn't a quick way to check this and who wants to spend another few hours re-running a sample. What I realised is the volatile poly-aromatic hydrocarbons within my sample were starting to evaporate between runs 💡. What I couldn't say is how does this affect my profiles. Using UMAP and the visual video attached we can see runs from different tripliates are changing. The thing that is changing is the relative amount of the volatile compounds which shifts the entire profile. However, this isn't always the case, so you don't have to spend hours of your life re-running samples if the profiles are alright. 

If you're struggling with laboratory issues and need help saving time, I'm here to help! With my expertise in UMAP and other multivariate techniques, I can assist you in analyzing your data and uncovering patterns and relationships. Learn more about UMAP and how it can help you solve laboratory issues by reading this blog. And if you need any assistance, feel free to reach out!