Revealing multi-scale population structure in large cohorts

By A. Diaz-Papkovich, Luke Anderson-Trocmé, Simon Gravel. 2018

24 citations recorded by the source corpus.

Population structure in genetic data depends on complex demographic processes including geographic isolation, genetic drift, migration, and admixture. Together with technical artifacts, population structure is a prominent confounder of genomic studies. Identifying such patterns is therefore central to the genomic enterprise. Whereas many methods can identify specific types of population structure, few are able to provide simple representations of genomic diversity across a range of scales. Here we investigate an approach to dimension reduction and visualization of genomic data that combines principal components analysis (PCA) with uniform manifold approximation and projection (UMAP) to succinctly illustrate population structure in large cohorts and capture their relationships on local and global scales. We demonstrate using genotype data from the 1000 Genome Project, the Health and Retirement Study, and the UK Biobank that projections using PCA-UMAP effectively cluster individuals who are genetically closely related while placing them in a global continuum of genetic variation. These projections reveal non-trivial population groupings, reflect ethnicity and geography on fine-scale levels, and uncover patterns in the distributions of a variety of phenotypes, establishing PCA-UMAP as a general-purpose approach toexploratory analysis in genomics.

Subfields: Computational Biology & Bioinformatics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Biology & Bioinformatics

1. Dmitry Kobak
2. George C. Linderman
3. Stefan Steinerberger
4. Yuval Kluger
5. Philipp Berens
6. Jukka Corander

Citations in the corpus

• Heavy-tailed kernels reveal a finer cluster structure in t-SNE visualisations (2019, background): “They range from single-cell cytometry [1] and transcriptomics [16, 19], where samples are cells and features are proteins or genes, to population genetics [4], where samples are people and features are single-nucleotide polymorphisms, to humanities [14], where samples are books and features are word”
• Fast hierarchical Bayesian analysis of population structure (2018, methodology): “UMAP is a non-linear dimensionality reduction technique that has been shown to perform well on large biological datasets including visualizing large genetic population datasets as well as in the analysis of single cell RNAseq data (48,49).”