Highly sensitive detection of mutations in CHO cell recombinant DNA using multi-parallel single molecule real-time DNA sequencing

Next-generation sequencing methodologies to detect low-frequency mutations: "Catch me if you can"

Mutations, the irreversible changes in an organism's DNA sequence, are present in tissues at a variant allele frequency (VAF) ranging from ∼10-8 per bp for a founder mutation to ∼10-3 for a histologically normal tissue sample containing several independent clones - compared to 1%- 50% for a heterozygous tumor mutation or a polymorphism. The rarity of these events poses a challenge for accurate clinical diagnosis and prognosis, toxicology, and discovering new disease etiologies. Standard Next-Generation Sequencing (NGS) technologies report VAFs as low as 0.5% per nt, but reliably observing rarer precursor events requires additional sophistication to measure ultralow-frequency mutations. We detail the challenge; define terms used to characterize the results, which vary between laboratories and sometimes conflict between biologists and bioinformaticists; and describe recent innovations to improve standard NGS methodologies including: single-strand consensus sequence methods such as Safe-SeqS and SiMSen-Seq; tandem-strand consensus sequence methods such as o2n-Seq and SMM-Seq; and ultrasensitive parent-strand consensus sequence methods such as DuplexSeq, PacBio HiFi, SinoDuplex, OPUSeq, EcoSeq, BotSeqS, Hawk-Seq, NanoSeq, SaferSeq, and CODEC. Practical applications are also noted. Several methods quantify VAF down to 10-5 at a nt and mutation frequency (MF) in a target region down to 10-7 per nt. By expanding to > 1 Mb of sites never observed twice, thus forgoing VAF, other methods quantify MF < 10-9 per nt or < 15 errors per haploid genome. Clonal expansion cannot be directly distinguished from independent mutations by sequencing, so it is essential for a paper to report whether its MF counted only different mutations - the minimum independent-mutation frequency MFminI - or all mutations observed including recurrences - the larger maximum independent-mutation frequency MFmaxI which may reflect clonal expansion. Ultrasensitive methods reveal that, without their use, even mutations with VAF 0.5-1% are usually spurious.

PacBio sequencing detects genome-wide ultra-low-frequency substitution mutations resulting from exposure to chemical mutagens

Genetic toxicology uses several assays to identity mutagens and protects the public. Most of these assays, however, rely on reporter genes, can only measure mutation indirectly based on phenotype, and often require specific cell lines or animal models-features that impede their integration with existing and emerging toxicological models, such as organoids. In this study, we show that PacBio Single-Molecule, Real-Time (PB SMRT) sequencing identified substitution mutations caused by chemical mutagens in Escherichia coli by generating nearly error-free consensus reads after repeatedly inspecting both strands of circular DNA molecules. Using DNA from E. coli exposed to ethyl methanosulfonate (EMS) or N-ethyl-N-nitrosourea (ENU), PB SMRT sequencing detected mutation frequencies (MFs) and spectra comparable to those obtained by clone-sequencing from the same exposures. The optimized background MF of PB SMRT sequencing was ≤ 1 × 10^-7 mutations per base pair (mut/bp).

Benchmarking and optimization of a high-throughput sequencing based method for transgene sequence variant analysis in biotherapeutic cell line development

In recent years, High-Throughput Sequencing (HTS) based methods to detect mutations in biotherapeutic transgene products have become a key quality step deployed during the development of manufacturing cell line clones. Previously we reported on a higher throughput, rapid mutation detection method based on amplicon sequencing (targeting transgene RNA) and detailed its implementation to facilitate cell line clone selection. By gaining experience with our assay in a diverse set of cell line development programs, we improved the computational analysis as well as experimental protocols. Here we report on these improvements as well as on a comprehensive benchmarking of our assay. We evaluated assay performance by mixing amplicon samples of a verified mutated antibody clone with a non-mutated antibody clone to generate spike-in mutations from ∼60% down to ∼0.3% frequencies. We subsequently tested the effect of 16 different sample and HTS library preparation protocols on the assay's ability to quantify mutations and on the occurrence of false-positive background error mutations (artifacts). Our evaluation confirmed assay robustness, established a high confidence limit of detection of ∼0.6%, and identified protocols that reduce error levels thereby significantly reducing a source of false positives that bottlenecked the identification of low-level true mutations.

A hybrid correcting method considering heterozygous variations by a comprehensive probabilistic model

Background

The emergence of the third generation sequencing technology, featuring longer read lengths, has demonstrated great advancement compared to the next generation sequencing technology and greatly promoted the biological research. However, the third generation sequencing data has a high level of the sequencing error rates, which inevitably affects the downstream analysis. Although the issue of sequencing error has been improving these years, large amounts of data were produced at high sequencing errors, and huge waste will be caused if they are discarded. Thus, the error correction for the third generation sequencing data is especially important. The existing error correction methods have poor performances at heterozygous sites, which are ubiquitous in diploid and polyploidy organisms. Therefore, it is a lack of error correction algorithms for the heterozygous loci, especially at low coverages.

Results

In this article, we propose a error correction method, named QIHC. QIHC is a hybrid correction method, which needs both the next generation and third generation sequencing data. QIHC greatly enhances the sensitivity of identifying the heterozygous sites from sequencing errors, which leads to a high accuracy on error correction. To achieve this, QIHC established a set of probabilistic models based on Bayesian classifier, to estimate the heterozygosity of a site and makes a judgment by calculating the posterior probabilities. The proposed method is consisted of three modules, which respectively generates a pseudo reference sequence, obtains the read alignments, estimates the heterozygosity the sites and corrects the read harboring them. The last module is the core module of QIHC, which is designed to fit for the calculations of multiple cases at a heterozygous site. The other two modules enable the reads mapping to the pseudo reference sequence which somehow overcomes the inefficiency of multiple mappings that adopt by the existing error correction methods.

Conclusions

To verify the performance of our method, we selected Canu and Jabba to compare with QIHC in several aspects. As a hybrid correction method, we first conducted a groups of experiments under different coverages of the next-generation sequencing data. QIHC is far ahead of Jabba on accuracy. Meanwhile, we varied the coverages of the third generation sequencing data and compared performances again among Canu, Jabba and QIHC. QIHC outperforms the other two methods on accuracy of both correcting the sequencing errors and identifying the heterozygous sites, especially at low coverage. We carried out a comparison analysis between Canu and QIHC on the different error rates of the third generation sequencing data. QIHC still performs better. Therefore, QIHC is superior to the existing error correction methods when heterozygous sites exist.

A combined computational and experimental strategy identifies mutations conferring resistance to drugs targeting the BCR-ABL fusion protein

Drug resistance is of increasing concern, especially during the treatments of infectious diseases and cancer. To accelerate the drug discovery process in combating issues of drug resistance, here we developed a computational and experimental strategy to predict drug resistance mutations. Using BCR-ABL as a case study, we successfully recaptured the clinically observed mutations that confer resistance imatinib, nilotinib, dasatinib, bosutinib, and ponatinib. We then experimentally tested the predicted mutants in vitro. We found that although all mutants showed weakened binding strength as expected, the binding constants alone were not a good indicator of drug resistance. Instead, the half-maximal inhibitory concentration (IC50) was shown to be a good indicator of the incidence of the predicted mutations, together with change in catalytic efficacy. Our suggested strategy for predicting drug-resistance mutations includes the computational prediction and in vitro selection of mutants with increased IC50 values beyond the drug safety window.

A comparative evaluation of hybrid error correction methods for error-prone long reads

BACKGROUND

Third-generation sequencing technologies have advanced the progress of the biological research by generating reads that are substantially longer than second-generation sequencing technologies. However, their notorious high error rate impedes straightforward data analysis and limits their application. A handful of error correction methods for these error-prone long reads have been developed to date. The output data quality is very important for downstream analysis, whereas computing resources could limit the utility of some computing-intense tools. There is a lack of standardized assessments for these long-read error-correction methods.

RESULTS

Here, we present a comparative performance assessment of ten state-of-the-art error-correction methods for long reads. We established a common set of benchmarks for performance assessment, including sensitivity, accuracy, output rate, alignment rate, output read length, run time, and memory usage, as well as the effects of error correction on two downstream applications of long reads: de novo assembly and resolving haplotype sequences.

CONCLUSIONS

Taking into account all of these metrics, we provide a suggestive guideline for method choice based on available data size, computing resources, and individual research goals.

Long-read sequencing across the C9orf72 'GGGGCC' repeat expansion: implications for clinical use and genetic discovery efforts in human disease

BACKGROUND

Many neurodegenerative diseases are caused by nucleotide repeat expansions, but most expansions, like the C9orf72 'GGGGCC' (G4C2) repeat that causes approximately 5-7% of all amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) cases, are too long to sequence using short-read sequencing technologies. It is unclear whether long-read sequencing technologies can traverse these long, challenging repeat expansions. Here, we demonstrate that two long-read sequencing technologies, Pacific Biosciences' (PacBio) and Oxford Nanopore Technologies' (ONT), can sequence through disease-causing repeats cloned into plasmids, including the FTD/ALS-causing G4C2 repeat expansion. We also report the first long-read sequencing data characterizing the C9orf72 G4C2 repeat expansion at the nucleotide level in two symptomatic expansion carriers using PacBio whole-genome sequencing and a no-amplification (No-Amp) targeted approach based on CRISPR/Cas9.

RESULTS

Both the PacBio and ONT platforms successfully sequenced through the repeat expansions in plasmids. Throughput on the MinION was a challenge for whole-genome sequencing; we were unable to attain reads covering the human C9orf72 repeat expansion using 15 flow cells. We obtained 8× coverage across the C9orf72 locus using the PacBio Sequel, accurately reporting the unexpanded allele at eight repeats, and reading through the entire expansion with 1324 repeats (7941 nucleotides). Using the No-Amp targeted approach, we attained > 800× coverage and were able to identify the unexpanded allele, closely estimate expansion size, and assess nucleotide content in a single experiment. We estimate the individual's repeat region was > 99% G4C2 content, though we cannot rule out small interruptions.

CONCLUSIONS

Our findings indicate that long-read sequencing is well suited to characterizing known repeat expansions, and for discovering new disease-causing, disease-modifying, or risk-modifying repeat expansions that have gone undetected with conventional short-read sequencing. The PacBio No-Amp targeted approach may have future potential in clinical and genetic counseling environments. Larger and deeper long-read sequencing studies in C9orf72 expansion carriers will be important to determine heterogeneity and whether the repeats are interrupted by non-G4C2 content, potentially mitigating or modifying disease course or age of onset, as interruptions are known to do in other repeat-expansion disorders. These results have broad implications across all diseases where the genetic etiology remains unclear.