Researchers from The University of Queensland assessed the usefulness of four different dimensionality reduction methods (PCA, multidimensional scaling [MDS], t-SNE, and UMAP) to analyze 71 large bulk transcriptomic datasets.
“When you consider that we’re looking at large datasets of patients, who each have more than 10,000 genes, we need a really good method to reduce the complexity of this big data for better interpretation,” explains Di Yu, senior author of the study published recently in Cell Reports.
“Of the four tools we compared, UMAP stood out as incredibly powerful. It performed significantly better than PCA, which is the tool many clinicians currently use to try and stratify patients,” he adds.
Search Antibodies Search Now Use our Antibody Search Tool to find the right antibody for your research. Filter
by Type, Application, Reactivity, Host, Clonality, Conjugate/Tag, and Isotype.
UMAP was the most efficient at reporting patient clustering. Using the tool, the researchers were able to separate healthy samples from those with lupus, as well as segregate the lupus patients into disease subgroups. They could also show which patients were getting better and who were getting worse.
The UMAP tool is relatively new, and is currently only used in biomedical research, however, Yu hopes the publication of his team’s results will lead to its future adoption in the clinic. “UMAP’s algorithm is based more on machine learning, and this makes it much more powerful than the popular PCA tool, which has a linear approach,” says Yu.