Saori Sakaue, Yukinori Okada ≪Statistical Genetics≫ Celebrating our genomic diversity—Fine-scale differences in the Japanese population

Journal: Nature Communications

Researchers led by Osaka University use dimensionality reduction analyses to identify fine-scale differences in the Japanese population, revealing greater population structure than previously thought

Fig.1 The overview of this study

Click to enlarge

Osaka, Japan – Geographically isolated from mainland Eurasia, the island nation of Japan has seen little in the way of population mixing for thousands of years. In fact, after two waves of human migration—one 40,000 years ago from Southeast Asia and one 3,000 years ago from the Korean Peninsula—Japanese people are recognized as a relatively homogeneous and isolated population.

But in a study in Nature Communications, an international research team led by Osaka University revealed that there is more than meets the eye when it comes to the Japanese population.

“At first glance, the Japanese population appears to be demographically homogenous,” says lead author of the study Saori Sakaue. “However, there are obvious cultural and linguistic differences from region to region and from island to island in Japan, leading us to suspect that there may actually be greater population structure than has previously been shown.”

Using novel machine-learning methods, the researchers examined the genomic sequences of 170,000 individuals from Japan, looking for minute genetic differences. As it turns out, they were right.

“Dimensionality reduction analyses revealed fine-scale diversity within the Japanese population,” explains Sakaue. “Not only did we identify the two main population clusters observed in previous studies, we successfully distinguished eight sub-clusters within one of these populations, confirming that Japanese people are not as homogenous as we previously thought.”

Interestingly, the eight sub-clusters correlated with distinct islands located in the Ryukyu region of Japan, some of which are separated by as little as 50 km.

The researchers also successfully illuminated genetic diversity within populations from the United Kingdom, Malaysia, and an Arab cohort, suggesting that the machine-learning approach is also applicable to more complex populations.

They then turned their attention to the implications of their research, particularly the future disease risk prediction of individuals based on their unique genetic profile.

“Given the fine-scale differences within the Japanese population, we were interested in how subtle population differences may affect the genetic prediction of human diseases and complex traits,” explains senior author Yukinori Okada. “Using phenome-wide polygenic risk score analyses of 67 complex traits, we confirmed that the sub-structures within the Japanese population did indeed impact disease prediction.”

“The results of our study have significant implications for the wider field of individualized medicine, highlighting the importance of a thorough understanding of genetic diversity before embarking on disease prediction.”


Fig.2 Application of a machine-learning method, PCA-UMAP, to large-scale genetic data from Japanese, the United Kingdom, Malaysia, and Arab population.

Click to enlarge



Fig.3 (top) PRS of height and the measured height were smaller in non-mainland Japan. While PRS of BMI was smaller in non-mainland Japan, the measured BMI was larger in non-mainland Japan. (bottom) Longitudinal census data suggested that height was smaller in non-mainland over the past 50 years, whereas BMI in non-mainland exceeded that of mainland Japan only a decade ago.

Click to enlarge



The article, “Dimensionality reduction reveals fine-scale structure in the Japanese population with consequences for polygenic risk prediction,” was published in Nature Communications at DOI: 10.1038/s41467-020-15194-z.

Summary: The Japanese population has long been thought of as genetically homogeneous as a result of limited population mixing. However, an international research team led by Osaka University used novel machine-learning techniques to identify subtle genetic differences within the Japanese population. Regionality was identified among the sub-populations, reflecting the linguistic and cultural differences between regions in Japan. These minute genomic differences were shown to impact genetic prediction of human disease, with implications for personalized medicine.

Primary Keyword: Biology

Additional Keywords: Evolution, Genetics, Molecular Biology, Population Biology, Medicine/Health

Title: “Dimensionality reduction reveals fine-scale structure in the Japanese 5 population with consequences for polygenic risk prediction”
Journal: Nature Communications
Authors: Saori Sakaue, Jun Hirata, Masahiro Kanai, Ken Suzuki, Masato Akiyama, Chun Lai Too, Thurayya Arayssi, Mohammed Hammoudeh, Samar Al Emadi, Basel K. Masri, Hussein Halabi, Humeira Badsha, Imad W. Uthman, Richa Saxena, Leonid Padyukov, Makoto Hirata, Koichi Matsuda, Yoshinori Murakami, Yoichiro Kamatani, Yukinori Okada

DOI: 10.1038/s41467-020-15194-z

Funded by: Japan Society for the Promotion of Science, Japan Agency for Medical Research and Development, Takeda Science Foundation