They also demonstrated the possibility that genomic diversity might have an impact on polygenic risk score (PRS), a method to estimate the risk of a disease or disease-related trait for an individual.

Anatomically modern humans migrated out of Africa to Europe, the Middle East, and Asia. The history of migration, admixture, and adaptation shapes the diversity recorded in the human genome. Thus, it is possible to roughly identify races of people in the world through principal component analysis (PCA), a classical linear dimensionality reduction method, of genetic data. However, to what degree our genomes differ due to subtle environmental and regional diversity was not well known.

This research group proposed methods to visualize regional diversity on a two-dimensional space by applying recently developed machine learning methods to genome data of 170,000 people collected from various regions throughout Japan.

Using these methods, they discovered that the Japanese population was roughly separated into two groups: Hondo (primarily in the mainland) and Ryukyu (Okinawa and part of Kyushu). The application of PCA–UMAP to the Ryukyu cluster dissolved it into different sub-clusters.

The researchers also applied these five dimensionality reduction methods to genotype data from the UK, Malaysia, and some Arab States, identifying subpopulations within each population.

In addition, they demonstrated that the fine-scale population structure in the Japanese population also affected the PRS, suggesting that a careful assessment of the genetic and phenotypic architecture of complex traits was necessary when planning the risk stratification of individuals based on the relative rank of PRSs.

This study showed that the PRS differences might reflect biases from the uncorrected structure and that it was also necessary to understand the adaptive evolution in regions and their history. This will give a critical suggestion to tailor-made medical treatment in the future.

Figure 1

Figure 2

Figure 3