Characterization and mitigation of fragmentation enzyme-induced dual stranded artifacts

Characterization and mitigation of artifacts derived from NGS library preparation due to structure-specific sequences in the human genome

BACKGROUND

Hybridization capture-based targeted next generation sequencing (NGS) is gaining importance in routine cancer clinical practice. DNA library preparation is a fundamental step to produce high-quality sequencing data. Numerous unexpected, low variant allele frequency calls were observed in libraries using sonication fragmentation and enzymatic fragmentation. In this study, we investigated the characteristics of the artifact reads induced by sonication and enzymatic fragmentation. We also developed a bioinformatic algorithm to filter these sequencing errors.

RESULTS

We used pairwise comparisons of somatic single nucleotide variants (SNVs) and insertions and deletions (indels) of the same tumor DNA samples prepared using both ultrasonic and enzymatic fragmentation protocols. Our analysis revealed that the number of artifact variants was significantly greater in the samples generated using enzymatic fragmentation than using sonication. Most of the artifacts derived from the sonication-treated libraries were chimeric artifact reads containing both cis- and trans-inverted repeat sequences of the genomic DNA. In contrast, chimeric artifact reads of endonuclease-treated libraries contained palindromic sequences with mismatched bases. Based on these distinctive features, we proposed a mechanistic hypothesis model, PDSM (pairing of partial single strands derived from a similar molecule), by which these sequencing errors derive from ultrasonication and enzymatic fragmentation library preparation. We developed a bioinformatic algorithm to generate a custom mutation "blacklist" in the BED region to reduce errors in downstream analyses.

CONCLUSIONS

We first proposed a mechanistic hypothesis model (PDSM) of sequencing errors caused by specific structures of inverted repeat sequences and palindromic sequences in the natural genome. This new hypothesis predicts the existence of chimeric reads that could not be explained by previous models, and provides a new direction for further improving NGS analysis accuracy. A bioinformatic algorithm, ArtifactsFinder, was developed and used to reduce the sequencing errors in libraries produced using sonication and enzymatic fragmentation.

A T cell receptor targeting a recurrent driver mutation in FLT3 mediates elimination of primary human acute myeloid leukemia in vivo

Acute myeloid leukemia (AML), the most frequent leukemia in adults, is driven by recurrent somatically acquired genetic lesions in a restricted number of genes. Treatment with tyrosine kinase inhibitors has demonstrated that targeting of prevalent FMS-related receptor tyrosine kinase 3 (FLT3) gain-of-function mutations can provide significant survival benefits for patients, although the efficacy of FLT3 inhibitors in eliminating FLT3-mutated clones is variable. We identified a T cell receptor (TCR) reactive to the recurrent D835Y driver mutation in the FLT3 tyrosine kinase domain (TCRFLT3D/Y). TCRFLT3D/Y-redirected T cells selectively eliminated primary human AML cells harboring the FLT3D835Y mutation in vitro and in vivo. TCRFLT3D/Y cells rejected both CD34+ and CD34- AML in mice engrafted with primary leukemia from patients, reaching minimal residual disease-negative levels, and eliminated primary CD34+ AML leukemia-propagating cells in vivo. Thus, T cells targeting a single shared mutation can provide efficient immunotherapy toward selective elimination of clonally involved primary AML cells in vivo.

svCapture: efficient and specific detection of very low frequency structural variant junctions by error-minimized capture sequencing

Error-corrected sequencing of genomic targets enriched by probe-based capture has become a standard approach for detecting single-nucleotide variants (SNVs) and small insertion/deletions (indels) present at very low variant allele frequencies. Less attention has been given to comparable strategies for rare structural variant (SV) junctions, where different error mechanisms must be addressed. Working from samples with known SV properties, we demonstrate that duplex sequencing (DuplexSeq), which demands confirmation of variants on both strands of a source DNA molecule, eliminates false SV junctions arising from chimeric PCR. DuplexSeq could not address frequent intermolecular ligation artifacts that arise during Y-adapter addition prior to strand denaturation without requiring multiple source molecules. In contrast, tagmentation libraries coupled with data filtering based on strand family size greatly reduced both artifact classes and enabled efficient and specific detection of single-molecule SV junctions. The throughput of SV capture sequencing (svCapture) and base-level accuracy of DuplexSeq provided detailed views of the microhomology profile and limited occurrence of de novo SNVs near the junctions of hundreds of newly created SVs, suggesting end joining as a possible formation mechanism. The open source svCapture pipeline enables rare SV detection as a routine addition to SNVs/indels in properly prepared capture sequencing libraries.

CRISPR-KRISPR: a method to identify on-target and random insertion of donor DNAs and their characterization in knock-in mice

CRISPR tools can generate knockout and knock-in animal models easily, but the models can contain off-target genomic lesions or random insertions of donor DNAs. Simpler methods to identify off-target lesions and random insertions, using tail or earpiece DNA, are unavailable. We develop CRISPR-KRISPR (CRISPR-Knock-ins and Random Inserts Searching PRotocol), a method to identify both off-target lesions and random insertions. CRISPR-KRISPR uses as little as 3.4 μg of genomic DNA; thus, it can be easily incorporated as an additional step to genotype founder animals for further breeding.

OPUSeq simplifies detection of low-frequency DNA variants and uncovers fragmentase-associated artifacts

Detection of low-frequency DNA variants (below 1%) is becoming increasingly important in biomedical research and clinical practice, but is challenging to do with standard sequencing approaches due to high error rates. The use of double-stranded unique molecular identifiers (dsUMIs) allows correction of errors by comparing reads arising from the same original DNA duplex. However, the implementation of such approaches is still challenging. Here, we present a novel method, one-pot dsUMI sequencing (OPUSeq), which allows incorporation of dsUMIs in the same reaction as the library PCR. This obviates the need for adapter pre-synthesis or additional enzymatic steps. OPUSeq can be incorporated into standard DNA library preparation approaches and coupled with hybridization target capture. We demonstrate successful error correction and detection of variants down to allele frequency of 0.01%. Using OPUSeq, we also show that the use of enzymatic fragmentation can lead to the appearance of spurious double-stranded variants, interfering with detection of variant fractions below 0.1%.