Multi-platform discovery of haplotype-resolved structural variation in human genomes

Mark Chaisson, Ashley D. Sanders, Xuefang Zhao, Ankit Malhotra, David Porubský, Tobias Rausch, Eugene J. Gardner, Oscar L. Rodriguez, Li Guo, Ryan L. Collins, Xian Fan, Jia Wen, Robert E. Handsaker, Susan Fairley, Zev Kronenberg, Xiangmeng Kong, Fereydoun Hormozdiari, Dillon Lee, Aaron M. Wenger, Alex Hastie, Danny Antaki, Peter A. Audano, Harrison Brand, Stuart Cantsilieris, Han Cao, Eliza Cerveira, Chong Chen, Xintong Chen, Chen-Shan Chin, Zechen Chong, Nelson T. Chuang, Christine Lambert, Deanna M. Church, Laura Clarke, Andrew Farrell, Joey Flores, Timur R. Galeev, David U. Gorkin, Madhusudan Gujral, Victor Guryev, Haynes Heaton, Jonas Korlach, Sushant Kumar, Jee Young Kwon, Jong Eun Lee, Joyce Lee, Wan‐Ping Lee, Sau Peng Lee, Shantao Li, Patrick Marks, Karine A. Viaud‐Martinez, Sascha Meiers, Katherine M. Munson, Fábio C. P. Navarro, Bradley J. Nelson, Conor Nodzak, Amina Noor, Sofia Kyriazopoulou-Panagiotopoulou, Andy Wing Chun Pang, Yunjiang Qiu, Gabriel Rosanio, Xian Mallory, Adrian M. Stütz, Diana C.j. Spierings, Alistair Ward, AnneMarie E. Welch, Ming Xiao, Wei Xu, Chengsheng Zhang, Qihui Zhu, Xiangqun Zheng-Bradley, Ernesto Lowy, Sergei Yakneen, Steven A. McCarroll, Goo Jun, Li Ding, Chong‐Lek Koh, Bing Ren, Paul Flicek, Ken Chen, Mark Gerstein, Pui–Yan Kwok, Peter M. Lansdorp, Gábor Marth, Jonathan Sebat, Xinghua Shi, Ali Bashir, Kai Ye, Scott E. Devine, Michael E. Talkowski, Ryan E. Mills, Tobias Marschall, Jan O. Korbel, Evan E. Eichler, Charles LeebioRxiv 2017 · DOI

TL;DR

This paper presents a comprehensive, multi-platform approach to identify structural variations (SVs) in human genomes with unprecedented resolution, achieving a sevenfold increase in SV detection compared to previous studies. By integrating diverse sequencing technologies and advanced algorithms, the researchers provide a gold standard dataset and methodology for understanding human genetic diversity and its implications for disease.

What Problem Does This Paper Address?

Identifying structural variants (SVs) – large-scale changes in DNA like insertions, deletions, duplications, and inversions – is crucial for understanding human genetic diversity and its role in diseases. However, current methods, primarily relying on short-read sequencing, struggle to accurately and comprehensively detect these variations, especially in complex or repetitive genomic regions. This limitation hinders our ability to fully grasp the extent of genetic variation and its impact on human health. Specifically, many SVs, particularly inversions and smaller structural changes, have historically escaped detection, leaving a significant gap in our knowledge of the human genome's variability.

Methods

The researchers employed a multi-platform strategy, integrating several cutting-edge genomic technologies:

  • Short-read sequencing (Illumina WGS): Standard for high-throughput sequencing, providing broad coverage but limited resolution for complex regions.
  • Long-read sequencing (PacBio): Crucial for spanning repetitive regions and resolving larger structural variations.
  • Strand-specific sequencing (Strand-seq): Particularly effective for identifying inversions by detecting strand switches within the DNA.
  • Optical mapping (Bionano Genomics): Provides a physical map of the genome, aiding in the assembly and validation of SVs.
  • 10X Chromium, Hi-C, and Illumina synthetic long reads: Used for haplotype phasing, allowing researchers to distinguish between the two copies of each chromosome and assign SVs to specific haplotypes.

They developed novel computational algorithms and integrated existing ones to process data from these diverse platforms. Key algorithmic advancements included methods for haplotype-aware SV discovery and the merging of callsets from different technologies to maximize sensitivity and specificity.

Datasets

The study analyzed three human parent-child trios:

  • A Han Chinese (CHS) trio (HG00513, HG00512, HG00514)
  • A Puerto Rican (PUR) trio (HG00732, HG00731, HG00733)
  • A Yoruban (YRI) Nigerian trio (NA19238, NA19239, NA19240)

These trios were chosen to represent different population groups and genetic diversity levels. The children's genomes were sequenced to high coverage, and parental data was used for phasing and validation. The resulting dataset serves as a "gold standard" for SV assessment.

Key Insights and Ideas

  • Synergy of Technologies: No single technology or algorithm is sufficient for comprehensive SV detection. Combining diverse approaches, especially long-read and strand-specific sequencing with short-read data, is essential for capturing the full spectrum of variation.
  • Haplotype Resolution is Key: Resolving SVs to specific haplotypes significantly improves detection accuracy and provides a more complete picture of genetic variation within an individual.
  • Inversions are Underestimated: A substantial number of inversions, particularly larger ones, were previously undetected. These inversions are often found in critical regions associated with genetic disorders.
  • Repetitive Regions are a Major Source of Variation: Tandem repeats and mobile elements are hotspots for SVs, and long-read technologies are crucial for resolving variations in these complex regions.
  • Implications for Disease: The identified SVs, especially those in critical regions and those affecting gene expression, highlight their potential role in human diseases and the "missing heritability" in complex disorders.

Key Results

  • Massive SV Discovery: Identified an average of 818,054 indel variants (<50 bp) and 27,622 SVs (≥50 bp) per human genome.
  • Sevenfold Increase in SV Detection: Achieved a sevenfold increase in SV detection compared to standard high-throughput sequencing studies, including the 1000 Genomes Project.
  • Discovery of Novel Inversions: Discovered an average of 156 inversions per genome, many of which were previously undetected.
  • Association with Disease Regions: 58 of the identified inversions intersect with critical regions of recurrent microdeletion and microduplication syndromes.
  • Improved Sensitivity: Demonstrated that combining multiple technologies and algorithms significantly increases sensitivity, particularly for insertions and variations in repetitive regions. For example, PB-SV callsets showed a threefold increase in sensitivity over IL-SV callsets.
  • Haplotype Phasing Success: Achieved high-quality chromosomal-level phasing, with Strand-seq and CHRO data showing the lowest mismatch error rate.

Limitations and Future Work

  • Computational Burden: The comprehensive approach requires significant computational resources, making it challenging for routine large-scale studies. The paper suggests a "triaged application" of technologies for future studies.
  • SV Resolution in Segmental Duplications: While progress was made, resolving SVs within highly duplicated regions remains a challenge, necessitating the continued use of read-depth methods.
  • Forward Validation: The authors note the pressing need to reduce the false discovery rate (FDR) of SV calling, as forward validation of all potentially disease-relevant events is impractical.
  • Population Diversity: While three trios were analyzed, the authors suggest that a deeper understanding of germline SVs from more individuals across diverse global populations is needed for proper application in patient care.
  • Future Directions: The paper advocates for the development of new methods to sequence-resolve copy number variants (CNVs) and emphasizes the need for integrated algorithms to improve diagnostic yields in genetic testing.

Summary updated

Abstract

ABSTRACT The incomplete identification of structural variants (SVs) from whole-genome sequencing data limits studies of human genetic diversity and disease association. Here, we apply a suite of long-read, short-read, and strand-specific sequencing technologies, optical mapping, and variant discovery algorithms to comprehensively analyze three human parent–child trios to define the full spectrum of human genetic variation in a haplotype-resolved manner. We identify 818,054 indel variants (&lt;50 bp) and 27,622 SVs (≥50 bp) per human genome. We also discover 156 inversions per genome—most of which previously escaped detection. Fifty-eight of the inversions we discovered intersect with the critical regions of recurrent microdeletion and microduplication syndromes. Taken together, our SV callsets represent a sevenfold increase in SV detection compared to most standard high-throughput sequencing studies, including those from the 1000 Genomes Project. The method and the dataset serve as a gold standard for the scientific community and we make specific recommendations for maximizing structural variation sensitivity for future large-scale genome sequencing studies.
BibTeX
@misc{chaisson2017multi,
  author = {Mark Chaisson and Ashley D. Sanders and Xuefang Zhao and Ankit Malhotra and David Porubský and Tobias Rausch and Eugene J. Gardner and Oscar L. Rodriguez and Li Guo and Ryan L. Collins and Xian Fan and Jia Wen and Robert E. Handsaker and Susan Fairley and Zev Kronenberg and Xiangmeng Kong and Fereydoun Hormozdiari and Dillon Lee and Aaron M. Wenger and Alex Hastie and Danny Antaki and Peter A. Audano and Harrison Brand and Stuart Cantsilieris and Han Cao and Eliza Cerveira and Chong Chen and Xintong Chen and Chen-Shan Chin and Zechen Chong and Nelson T. Chuang and Christine Lambert and Deanna M. Church and Laura Clarke and Andrew Farrell and Joey Flores and Timur R. Galeev and David U. Gorkin and Madhusudan Gujral and Victor Guryev and Haynes Heaton and Jonas Korlach and Sushant Kumar and Jee Young Kwon and Jong Eun Lee and Joyce Lee and Wan‐Ping Lee and Sau Peng Lee and Shantao Li and Patrick Marks and Karine A. Viaud‐Martinez and Sascha Meiers and Katherine M. Munson and Fábio C. P. Navarro and Bradley J. Nelson and Conor Nodzak and Amina Noor and Sofia Kyriazopoulou-Panagiotopoulou and Andy Wing Chun Pang and Yunjiang Qiu and Gabriel Rosanio and Xian Mallory and Adrian M. Stütz and Diana C.J. Spierings and Alistair Ward and AnneMarie E. Welch and Ming Xiao and Wei Xu and Chengsheng Zhang and Qihui Zhu and Xiangqun Zheng-Bradley and Ernesto Lowy and Sergei Yakneen and Steven A. McCarroll and Goo Jun and Li Ding and Chong‐Lek Koh and Bing Ren and Paul Flicek and Ken Chen and Mark Gerstein and Pui–Yan Kwok and Peter M. Lansdorp and Gábor Marth and Jonathan Sebat and Xinghua Shi and Ali Bashir and Kai Ye and Scott E. Devine and Michael E. Talkowski and Ryan E. Mills and Tobias Marschall and Jan O. Korbel and Evan E. Eichler and Charles Lee},
  title = {{Multi-platform discovery of haplotype-resolved structural variation in human genomes}},
  year = {2017},
  doi = {10.1101/193144},
  howpublished = {bioRxiv},
}
Figure