In 2012, NIST convened the Genome in a Bottle Consortium to develop the metrology infrastructure needed to enable confidence in human whole genome variant calls.
2015 SMRT Informatics Developers Conference Presentation Slides: Kevin Corcoran of PacBio provided a brief review of community involvement in the development of analysis tools and showed a preview of upcoming sample preparation, chemistry and informatics improvements.
The Genome in a Bottle Consortium is developing the reference materials, reference methods , and reference data n
Purpose: Clinical laboratories, research laboratories and technology developers all need DNA samples with reliably known genotypes in order to help validate and improve their methods. The Genome in a Bottle Consortium (genomeinabottle.org) has been developing Reference Materials with high-accuracy whole genome sequences to support these efforts.Methodology: Our pilot reference material is based on Coriell sample NA12878 and was released in May 2015 as NIST RM 8398 (tinyurl.com/giabpilot). To minimize bias and improve accuracy, 11 whole-genome and 3 exome data sets produced using 5 different technologies were integrated using a systematic arbitration method [1]. The Genome in a Bottle Analysis Group is adapting these methods and developing new methods to characterize 2 families, one Asian and one Ashkenazi Jewish from the Personal Genome Project, which are consented for public release of sequencing and phenotype data. We have generated a larger and even more diverse data set on these samples, including high-depth Illumina paired-end and mate-pair, Complete Genomics, and Ion Torrent short-read data, as well as Moleculo, 10X, Oxford Nanopore, PacBio, and BioNano Genomics long-read data. We are analyzing these data to provide an accurate assessment of not just small variants but also large structural variants (SVs) in both “easy” regions of the genome and in some “hard” repetitive regions. We have also made all of the input data sources publicly available for download, analysis, and publication.Results: Our arbitration method produced a reference data set of 2,787,291 single nucleotide variants (SNVs), 365,135 indels, 2744 SVs, and 2.2 billion homozygous reference calls for our pilot genome. We found that our call set is highly sensitive and specific in comparison to independent reference data sets. We have also generated preliminary assemblies and structural variant calls for the next 2 trios from long read data and are currently integrating and validating these.Discussion: We combined the strengths of each of our input datasets to develop a comprehensive and accurate benchmark call set. In the short time it has been available, over 20 published or submitted papers have used our data. Many challenges exist in comparing to our benchmark calls, and thus we have worked with the Global Alliance for Genomics and Health to develop standardized methods, performance metrics, and software to assist in its use.[1] Zook et al, Nat Biotech. 2014.
The human immunoglobulin heavy chain locus (IGH) remains among the most understudied regions of the human genome. Recent efforts have shown that haplotype diversity within IGH is elevated and exhibits population specific patterns; for example, our re-sequencing of the locus from only a single chromosome uncovered >100 Kb of novel sequence, including descriptions of six novel alleles, and four previously unmapped genes. Historically, this complex locus architecture has hindered the characterization of IGH germline single nucleotide, copy number, and structural variants (SNVs; CNVs; SVs), and as a result, there remains little known about the role of IGH polymorphisms in inter-individual antibody repertoire variability and disease. To remedy this, we are taking a multi-faceted approach to improving existing genomic resources in the human IGH region. First, from whole-genome and fosmid-based datasets, we are building the largest and most ethnically diverse set of IGH reference assemblies to date, by employing PacBio long-read sequencing combined with novel algorithms for phased haplotype assembly. In total, our effort will result in the characterization of >15 phased haplotypes from individuals of Asian, African, and European descent, to be used as a representative reference set by the genomics and immunogenetics community. Second, we are utilizing this more comprehensive sequence catalogue to inform the design and analysis of novel targeted IGH genotyping assays. Standard targeted DNA enrichment methods (e.g., exome capture) are currently optimized for the capture of only very short (100’s of bp) DNA segments. Our platform uses a modified bench protocol to pair existing capture-array technologies with the enrichment of longer fragments of DNA, enabling the use of PacBio sequencing of DNA segments up to 7 Kb. This substantial increase in contiguity disambiguates many of the complex repeated structures inherent to the locus, while yielding the base pair fidelity required to call SNVs. Together these resources will establish a stronger framework for further characterizing IGH genetic diversity and facilitate IGH genomic profiling in the clinical and research settings, which will be key to fully understanding the role of IGH germline variation in antibody repertoire development and disease.
Structural variant calling combining Illumina and low-coverage Pacbio Detection of large genomic variation (structural variants) has proven challenging using short-read methods. Long-read approaches which can span these large events have promise to dramatically expand the ability to accurately call structural variants. Although sequencing with Pacific Biosciences (Pacbio) long-read technology has become increasingly high throughput, generating high coverage with the technology can still be limiting and investigators often would like to know what pacbio coverages are adequate to call structural variants. Here, we present a method to identify a substantially higher fraction of structural variants in the human genome using low-coverage pacbio data by multiple strategies for ensembling data types and algorithms. Algorithmically, we combine three structural variant callers: PBHoney by Adam English, Sniffles by Fritz Sedlazeck, and Parliament by Adam English (which we have modified to improve for speed). Parliament itself uses a combination of Pacbio and Illumina data with a number of short-read callers (Breakdancer, Pindel, Crest, CNVnator, Delly, and Lumpy). We show that the outputs of these three programs are largely complementary to each other, with each able to uniquely access different sets of structural variants at different coverages. Combining them together can more than double the recall of true structural variants from a truth set relative to sequencing with Illumina alone, with substantial improvements even at low pacbio coverages (3x – 7x). This allows us to present for the first time cost-benefit tradeoffs to investigators about how much pacbio sequencing will yield what improvements in SV-calling. This work also builds upon the foundational work of Genome in a Bottle led by Justin Zook in establishing a truth set for structural variants in the Ashkenazim-Jewish trio data recently released. This work demonstrates the power of this benchmark set – one of the first of its kind for structural variation data – to help understand and refine the accuracies of calling structural variants with a number of approaches.
Each human genome has thousands of structural variants compared to the reference assembly, up to 85% of which are difficult or impossible to detect with Illumina short reads and are only visible with long, multi-kilobase reads. The PacBio RS II and Sequel single molecule, real-time (SMRT) sequencing platforms have made it practical to generate long reads at high throughput. These platforms enable the discovery of structural variants just as short-read platforms did for single nucleotide variants. Numerous software algorithms call structural variants effectively from PacBio long reads, but algorithm sensitivity is lower for insertion variants and all heterozygous variants. Furthermore, the impact of coverage depth and read lengths on sensitivity is not fully characterized. To quantify how zygosity, coverage depth, and read lengths impact the sensitivity of structural variant detection, we obtained high coverage PacBio sequences for three human samples: haploid CHM1, diploid NA12878, and diploid SK-BR-3. For each dataset, reads were randomly subsampled to titrate coverage from 0.5- to 50-fold. The structural variants detected at each coverage were compared to the set at “full” 50-fold coverage. For the diploid samples, additional titrations were performed with reads first partitioned by phase using single nucleotide variants for essentially haploid structural variant discovery. Even at low coverages (1- to 5-fold), PacBio long reads reveal hundreds of structural variants that are not seen in deep 50-fold Illumina whole genome sequences. At moderate 10-fold PacBio coverage, a majority of structural variants are detected. Sensitivity begins to level off at around 40-fold coverage, though it does not fully saturate before 50-fold. Phasing improves sensitivity for all variant types, especially at moderate 10- to 20-fold coverage. Long reads are an effective tool to identify and phase structural variants in the human genome. The majority of variants are detected at moderate 10-fold coverage, and even extremely low long-read coverage (1- to 5-fold) reveals variants that are invisible to short-read sequencing. Performance will continue to improve with better software and longer reads, which will empower studies to connect structural variants to healthy and disease traits in the human population.
Human genomic variations range in size from single nucleotide substitutions to large chromosomal rearrangements. Sequencing technologies tend to be optimized for detecting particular variant types and sizes. Short reads excel at detecting SNVs and small indels, while long or linked reads are typically used to detect larger structural variants or phase distant loci. Long reads are more easily mapped to repetitive regions, but tend to have lower per-base accuracy, making it difficult to call short variants. The PacBio Sequel System produces two main data types: long continuous reads (up to 100 kbp), generated by single passes over a long template, and Circular Consensus Sequence (CCS) reads, generated by calculating the consensus of many sequencing passes over a single shorter template (500 bp to 20 kbp). The long-range information in continuous reads is useful for genome assembly and structural variant detection. The higher base accuracy of CCS effectively detects and phases short variants in single molecules. Recent improvements in library preparation protocols and sequencing chemistry have increased the length, accuracy, and throughput of CCS reads. For the human sample HG002, we collected 28-fold coverage 15 kbp high-fidelity CCS reads with an average read quality above Q20 (99% accuracy). The length and accuracy of these reads allow us to detect SNVs, indels, and structural variants not only in the Genome in a Bottle (GIAB) high confidence regions, but also in segmental duplications, HLA loci, and clinically relevant “difficult-to-map” genes. As with continuous long reads, we call structural variants at 90.0% recall compared to the GIAB structural variant benchmark “truth” set, with the added advantages of base pair resolution for variant calls and improved recall at compound heterozygous loci. With minimap2 alignments, GATK4 HaplotypeCaller variant calls, and simple variant filtration, we have achieved a SNP F-Score of 99.51% and an INDEL F-Score of 80.10% against the GIAB short variant benchmark “truth” set, in addition to calling variants outside of the high confidence region established by GIAB using previous technologies. With the long-range information available in 15 kbp reads, we applied the read-backed phasing tool WhatsHap to generate phase blocks with a mean length of 65 kbp across the entire genome. Using an alignment-based approach, we typed all major MHC class I and class II genes to at least 3-field precision. This new data type has the potential to expand the GIAB high confidence regions and “truth” benchmark sets to many previously difficult-to-map genes and allow a single sequencing protocol to address both short variants and large structural variants.
Recent improvements in sequencing chemistry and instrument performance combine to create a new PacBio data type, Single Molecule High-Fidelity reads (HiFi reads). Increased read length and improvement in library construction enables average read lengths of 10-20 kb with average sequence identity greater than 99% from raw single molecule reads. The resulting reads have the accuracy comparable to short read NGS but with 50-100 times longer read length. Here we benchmark the performance of this data type by sequencing and genotyping the Genome in a Bottle (GIAB) HG0002 human reference sample from the National Institute of Standards and Technology (NIST). We further demonstrate the general utility of HiFi reads by analyzing multiple clones of Cabernet Sauvignon. Three different clones were sequenced and de novo assembled with the CANU assembly algorithm, generating draft assemblies of very high contiguity equal to or better than earlier assembly efforts using PacBio long reads. Using the Cabernet Sauvignon Clone 8 assembly as a reference, we mapped the HiFi reads generated from Clone 6 and Clone 47 to identify single nucleotide polymorphisms (SNPs) and structural variants (SVs) that are specific to each of the three samples.
Past large scale cancer genome sequencing efforts, including The Cancer Genome Atlas and the International Cancer Genome Consortium, have utilized short-read sequencing, which is well-suited for detecting single nucleotide variants (SNVs) but far less reliable for detecting variants larger than 20 base pairs, including insertions, deletions, duplications, inversions and translocations. Recent same-sample comparisons of short- and long-read human reference genome data have revealed that short-read resequencing typically uncovers only ~4,000 structural variants (SVs, =50 bp) per genome and is biased towards deletions, whereas sequencing with PacBio long-reads consistently finds ~20,000 SVs, evenly balanced between insertions and deletions. This discovery has important implications for cancer research, as it is clear that SVs are both common and biologically important in many cancer subtypes, including colorectal, breast and ovarian cancer. Without confident and comprehensive detection of structural variants, it is unlikely we have a sufficiently complete picture of all the genomic changes that impact cancer development, disease progression, treatment response, drug resistance, and relapse. To begin to address this unmet need, we have sequenced the COLO829 tumor and matched normal lymphoblastoid cell lines to 49- and 51-fold coverage, respectively, with PacBio SMRT Sequencing, with the goal of developing a high-confidence structural variant call set that can be used to empirically evaluate cost-effective experimental designs for larger scale studies and develop structural variation calling software suitable for cancer genomics. Structural variant calling revealed over 21,000 deletions and 19,500 insertions larger than 20 bp, nearly four times the number of events detected with short-read sequencing. The vast majority of events are shared between the tumor and normal, with about 100 putative somatic deletions and 400 insertions, primarily in microsatellites. A further 40 rearrangements were detected, nearly exclusively in the tumor. One rearrangement is shared between the tumor and normal, t(5;X) which disrupts the mismatch repeat gene MSH3, and is likely a driver mutation. Generating high-confidence call sets that cover the entire size-spectrum of somatic variants from a range of cancer model systems is the first step in determining what will be the best approach for addressing an ongoing blind spot in our current understanding of cancer genomes. Here the application of PacBio sequencing to a melanoma cancer cell line revealed thousands of previously overlooked variants, including a mutation likely involved in tumorogenesis.
Genomics studies have shown that the insertions, deletions, duplications, translocations, inversions, and tandem repeat expansions in the structural variant (SV) size range (>50 bp) contribute to the evolution of traits and often have significant associations with agronomically important phenotypes. However, most SVs are too small to detect with array comparative genomic hybridization and too large to reliably discover with short-read DNA sequencing. While de novo assembly is the most comprehensive way to identify variants in a genome, recent studies in human genomes show that PacBio SMRT Sequencing sensitively detects structural variants at low coverage. Here we present SV characterization in the major crop species Oryza sativa subsp. indica (rice) with low-fold coverage of long reads. In addition, we provide recommendations for sequencing and analysis for the application of this workflow to other important agricultural species.
Introduction: Long-read sequencing has been applied successfully to assemble genomes and detect structural variants. However, due to high raw-read error rates (10-15%), it has remained difficult to call small variants from long reads. Recent improvements in library preparation and sequencing chemistry have increased length, accuracy, and throughput of PacBio circular consensus sequencing (CCS) reads, resulting in 10-20kb reads with average read quality above 99%. Materials and Methods: We sequenced a 12kb library from human reference sample HG002 to 18-fold coverage on the PacBio Sequel II System with three SMRT Cells 8M. The CCS algorithm was used to generate highly-accurate (average 99.8%) 11.4kb reads, which were mapped to the hg19 reference with pbmm2. We detected small variants using Google DeepVariant with a model trained for CCS and phased the variants using WhatsHap. Structural variants were detected with pbsv. Variant calls were evaluated against Genome in a Bottle (GIAB) benchmarks. Results: With these reads, DeepVariant achieves SNP and Indel F1 scores of 99.82% and 96.70% against the GIAB truth set, and pbsv achieves 95.94% recall on structural variants longer than 50bp. Using WhatsHap, small variants were phased into haplotype blocks with 105kb N50. The improved mappability of long reads allows us to align to and detect variants in medically relevant genes such as CYP2D6 and PMS2 that have proven “difficult-to-map” with short reads. Conclusions: These highly-accurate long reads combine the mappability and ability to detect structural variants of long reads with the accuracy and ability to detect small variants of short reads.
To comprehensively detect large variants in human genomes, we have extended pbsv – a structural variant caller for long reads – to call copy-number variants (CNVs) from read-clipping and read-depth signatures. In human germline benchmark samples, we detect more than 300 CNVs spanning around 10 Mb, and we call hundreds of additional events in re-arranged cancer samples. Long-read sequencing of diverse humans has revealed more than 20,000 insertion, deletion, and inversion structural variants spanning more than 12 Mb in a typical human genome. Most of these variants are too large to detect with short reads and too small for array comparative genome hybridization (aCGH). While the standard approaches to calling structural variants with long reads thrive in the 50 bp to 10 kb size range, they tend to miss exactly the large (>50 kb) copy-number variants that are called more readily with aCGH and short reads. Standard algorithms rely on reference-based mapping of reads that fully span a variant or on de novo assembly; and copy-number variants are often too large to be spanned by a single read and frequently involve segmentally duplicated sequence that is not yet included in most de novo assemblies.
Introduction: Long-read PacBio SMRT Sequencing has been applied successfully to assemble genomes and detect structural variants. However, due to high raw read error rates of 10-15%, it has remained difficult to call small variants from long reads. Recent improvements in library preparation, sequencing chemistry, and instrument yield have increased length, accuracy, and throughput of PacBio Circular Consensus (CCS) reads, resulting in 10-20 kb “HiFi” reads with mean read quality above 99%. Materials and Methods: We sequenced 11 kb size-selected libraries from the Genome in a Bottle (GIAB) human reference samples HG001, HG002, and HG005 to approximately 30-fold coverage on the Sequel II System with six SMRT Cells 8M each. The CCS algorithm was used to generate highly accurate (average 99.8%) reads of mean length 10-11 kb, which were then mapped to the hs37d5 reference with pbmm2. We detected small variants using Google DeepVariant and compared these variant calls to GIAB benchmarks. Small variants were then phased with WhatsHap. Results: With these long, highly accurate CCS reads, DeepVariant achieves high SNP and Indel accuracy against the GIAB benchmark truth set for all three reference samples. Using WhatsHap, small variants were phased into haplotype blocks with N50 from 82 to 146 kb. The improved mappability of long reads allows detection of variants in many medically relevant genes such as CYP2D6and PMS2that have proven ‘difficult-to-map’ with short reads. We show that small variant precision and recall remain high down to 15-fold coverage. Conclusions: These highly accurate long reads combine the mappability of noisy long reads with the accuracy and small variant detection utility of short reads, which will allow the detection and phasing of variants in regions that have proven recalcitrant to short read sequencing and variant detection.
HiFi reads (>99% accurate, 15-20 kb) from the PacBio Sequel II System consistently provide complete and contiguous genome assemblies. In addition to completeness and contiguity, accuracy is of critical importance, as assembly errors complicate downstream analysis, particularly by disrupting gene frames. Metrics used to assess assembly accuracy include: 1) in-frame gene count, 2) kmer consistency, and 3) concordance to a benchmark, where discordances are interpreted as assembly errors. Genome in a Bottle (GIAB) provides a benchmark for the human genome with estimated accuracy of 99.9999% (Q60). Concordance for human HiFi assemblies exceeds Q50, which provides excellent genomes for downstream analysis, but presents a challenge that any new benchmark must significantly exceed Q50 or the discordance will represent the error rate of the benchmark. To establish benchmarks for Oryza sativa and Drosophila melanogaster, we collected draft references, Illumina short reads, and PacBio HiFi reads. By species, the benchmark was defined as regions of normal coverage that are not within 5 bp of a small variant or 50 bp of a structural variant. For both species, the benchmark regions span around 60% of the genome and HiFi assemblies achieve Q50 accuracy, which is notably more accurate than assemblies with other technologies and meets typical standards for a finished, reference-grade assembly. Here we present a protocol to generate benchmarks for any sample that rival the GIAB benchmark in accuracy. These benchmarks allow the comparison and improvement of genome assemblies and highlight the superior accuracy of assemblies generated with PacBio HiFi reads.
If you have a question, need to check the status of an order, or are interested in purchasing an instrument, we're here to help.