Background

Limitations of Existing Genomic Analyses and Data Imbalance

Human genome research is a crucial foundation for elucidating disease causes and realizing personalized precision medicine. However, over 80% of global genomic data is biased towards individuals of European ancestry, making the expansion to multi-ethnic research an urgent task. Existing genome microarray chips rely on pre-designed probes, which makes it difficult to capture genetic variations in non-European populations. In particular, there have been ongoing concerns that unique variations in African and Latin American populations are significantly missed in chip analysis, leading to racial inequality in genomic research.

Overcoming the Barrier of High Analysis Costs

Whole-genome sequencing (WGS), which decodes the entire DNA information, is an ideal alternative, but the cost burden in large-scale sample analysis makes it difficult to directly apply to multi-ethnic cohort studies. Whole-exome sequencing (WES), which targets only the protein-coding exon regions, has also shown limitations in capturing common variations in non-coding regions and population information. In an attempt to overcome these barriers, a hybrid sequencing technique that leverages the synergistic effects of the two methods has finally been developed.

Key Findings

Chemical Combination of Genomes and Exomes with Computational Correction

The research team at the Broad Institute introduced a blended genome-exome (BGE) technique that integrates low-pass WGS and deep WES into a single sequencing run. BGE first creates a genomic library, then takes a portion of it and amplifies only the exome region. The two libraries are then mixed in a 67:33 ratio and sequenced in a single tube, thereby acquiring low-resolution genome data (average 1-4x coverage depth) and high-resolution exome data (average 30-40x coverage depth). This data is corrected using the 'GLIMPSE2' genotype imputation algorithm to cleanly restore missing sequences. Sequencing costs have been reduced to as low as 28% of existing high-resolution WGS with 30x coverage. For example, the cost per sample has been reduced to one-quarter of the original level, creating a foundation for analyzing the genomes of four times more patients with the same budget.

Performance Demonstrated in a Large-Scale Multi-Ethnic Cohort of 53,000

The researchers applied the BGE technique to over 53,000 samples from the Populations Underrepresented in Mental Illness Association Studies (PUMAS), including African and Latin American populations. The analysis showed that in the common genetic variation region with a minor allele frequency (MAF) of 1% or more, all cohorts had a high correlation with an R2 value of 95% or more. Even for rare variations with a frequency of less than 1%, the R2 value was 90% or more, demonstrating reliable stability. In addition, in the analysis of copy number variations (CNVs) in protein-coding regions, a positive predictive value (PPV) of approximately 90% was confirmed when identifying variations spanning three or more exon regions. This data demonstrates that the combination of low-resolution sequencing data and high-performance algorithms can achieve accuracy comparable to high-cost WGS.

Significance and Prospects

A Stepping Stone to Resolving Global Genomic Imbalance

The BGE technique is considered a major tool for improving the imbalance in genomic data that has been biased towards specific populations. It has provided an efficient analysis alternative for researchers in Asia and Latin America, where large-scale research has been difficult due to high costs. As a multi-ethnic genomic map is accumulated, it is expected to reduce the racial bias in polygenic risk scores (PRS), which predict the risk of disease. This is expected to be a catalyst for extending the benefits of precision medicine, which has been advantageous only to specific races, to all of humanity. At the same time, it is expected to contribute to the democratization of genetics by enabling research institutions with limited resources to lead large-scale precision studies with diverse populations.

Practical Challenges of Hybrid Analysis

However, the BGE technique cannot perfectly replace all genomic research. Due to the imputation characteristic of filling in the gaps with algorithms, there are limitations in perfectly detecting extremely rare variations or mutations in non-coding regions for which there is no information in the reference panel. In fact, when analyzing populations with poor databases, accuracy can drop sharply. As the importance of non-coding variation analysis increases in the future, it will be necessary to support complementary coordination with high-resolution WGS.