Enhancing breakpoint resolution with deep segmentation model: A general refinement method for read-depth based structural variant callers

Read-depths (RDs) are frequently used in identifying structural variants (SVs) from sequencing data. For existing RD-based SV callers, it is difficult for them to determine breakpoints in single-nucleotide resolution due to the noisiness of RD data and the bin-based calculation. In this paper, we propose to use the deep segmentation model UNet to learn base-wise RD patterns surrounding breakpoints of known SVs. We integrate model predictions with an RD-based SV caller to enhance breakpoints in single-nucleotide resolution. We show that UNet can be trained with a small amount of data and can be applied both in-sample and cross-sample. An enhancement pipeline named RDBKE significantly increases the number of SVs with more precise breakpoints on simulated and real data. The source code of RDBKE is freely available at https://github.com/yaozhong/deepIntraSV.

Characterization of FMR1 Repeat Expansion and Intragenic Variants by Indirect Sequence Capture

Traditional methods for the analysis of repeat expansions, which underlie genetic disorders, such as fragile X syndrome (FXS), lack single-nucleotide resolution in repeat analysis and the ability to characterize causative variants outside the repeat array. These drawbacks can be overcome by long-read and short-read sequencing, respectively. However, the routine application of next-generation sequencing in the clinic requires target enrichment, and none of the available methods allows parallel analysis of long-DNA fragments using both sequencing technologies. In this study, we investigated the use of indirect sequence capture (Xdrop technology) coupled to Nanopore and Illumina sequencing to characterize FMR1, the gene responsible of FXS. We achieved the efficient enrichment (> 200×) of large target DNA fragments (~60–80 kbp) encompassing the entire FMR1 gene. The analysis of Xdrop-enriched samples by Nanopore long-read sequencing allowed the complete characterization of repeat lengths in samples with normal, pre-mutation, and full mutation status (> 1 kbp), and correctly identified repeat interruptions relevant for disease prognosis and transmission. Single-nucleotide variants (SNVs) and small insertions/deletions (indels) could be detected in the same samples by Illumina short-read sequencing, completing the mutational testing through the identification of pathogenic variants within the FMR1 gene, when no typical CGG repeat expansion is detected. The study successfully demonstrated the parallel analysis of repeat expansions and SNVs/indels in the FMR1 gene at single-nucleotide resolution by combining Xdrop enrichment with two next-generation sequencing approaches. With the appropriate optimization necessary for the clinical settings, the system could facilitate both the study of genotype–phenotype correlation in FXS and enable a more efficient diagnosis and genetic counseling for patients and their relatives.

Amplification Refractory Mutation System (ARMS)-PCR for Waxy Sorghum Authentication with Single-Nucleotide Resolution

Waxy sorghum has greater economic value than wild sorghum in relation to their use in food processing and the brewing industry. Thus, the authentication of the waxy sorghum species is an important issue. Herein, a rapid and sensitive Authentication Amplification Refractory Mutation System-PCR (aARMS-PCR) method was employed to identify sorghum species via its ability to resolve single-nucleotide in genes. As a proof of concept, we chose a species of waxy sorghum containing the wxc mutation which is abundantly used in liquor brewing. The aARMS-PCR can distinguish non-wxc sorghum from wxc sorghum to guarantee identification of specific waxy sorghum species. It allowed to detect as low as 1% non-wxc sorghum in sorghum mixtures, which ar one of the most sensitive tools for food authentication. Due to its ability for resolving genes with single-nucleotide resolution and high sensitivity, aARMS-PCR may have wider applicability in monitoring food adulteration, offering a rapid food authenticity verification in the control of adulteration.

Mapping Human Transient Transcriptomes Using Single Nucleotide Resolution 4sU Sequencing (SNU-Seq)

Genomes are pervasively transcribed leading to stable and unstable transcripts that define functional regions of genomes and contribute to cellular phenotypes. Defining comprehensive nascent transcriptomes is pivotal to understand gene regulation, disease processes, and the impact of extracellular signals on cells. However, currently employed methods are laborious, technically challenging and costly. We developed single-nucleotide resolution 4sU-sequencing (SNU-Seq), involving pulse labelling, biotinylation and direct isolation of nascent transcripts. Artificial poly-(A)-tailing of the 3' most nucleotide of nascent transcripts ensures oligo-d(T) primer-based library preparation and sequencing using commercial 3' RNA-Seq kits. We show that SNU-Seq is a cost-effective new method generating even read profiles across transcription units. We used SNU-Seq to identify transcription elongation parameters, to map usage of polyadenylation (PAS) sites and novel enhancers. Remarkably, 4sU labelled nascent RNA accumulates short ~100nt transcripts that map to the 5' end of genes. We show that isolation of these short nascent RNA and sequencing the 5' and 3' ends using size-selected SNU-Seq (ssSNU-Seq) provides highly sensitive annotations of mapped and novel TSSs, promoter-proximal pause/termination sites. Thus, SNU-seq and ssSNU-seq combined yield comprehensive transcriptomics data at low cost with high spatial and temporal resolution.

Global analysis of the specificities and targets of endoribonucleases from E. coli toxin-antitoxin systems

Toxin-antitoxin systems are widely distributed genetic modules typically featuring toxins that can inhibit bacterial growth and antitoxins that can reverse inhibition. Although Escherichia coli encodes 11 toxins with known or putative endoribonuclease activity, the target of most of these toxins remain poorly characterized. Using a new RNA-seq pipeline that enables the mapping and quantification of RNA cleavage with single-nucleotide resolution, we characterize the targets and specificities of 9 endoribonuclease toxins from E. coli. We find that these toxins use low-information cleavage motifs to cut a significant proportion of mRNAs in E. coli, but not tRNAs or the rRNAs from mature ribosomes. However, all the toxins, including those that are ribosome-dependent and cleave only translated RNA, inhibit ribosome biogenesis. This inhibition likely results from the cleavage of ribosomal protein transcripts, which disrupts the stoichiometry and biogenesis of new ribosomes and causes the accumulation of aberrant ribosome precursors. Collectively, our results provide a comprehensive, global analysis of endoribonuclease-based toxin-antitoxin systems in E. coli and support the conclusion that, despite their diversity, each disrupts translation and ribosome biogenesis.

Multiplexed single-molecule analysis of human telomerase synthesizing DNA

Genomic stability in proliferating cells critically depends on telomere maintenance by telomerase reverse transcriptase. Here we developed a real-time single-molecule RNA sequencing approach that visualizes telomerase catalysis and structural dynamics at single-nucleotide resolution using FRET and zero-mode waveguides. The method permits direct detection of dynamic steps and structural states throughout the telomerase catalytic cycle and can be generalized to other nucleic acid polymerase systems.

Sequencing uracil in DNA at single-nucleotide resolution

As an aberrant base in DNA, uracil is generated by dUMP misincorporation or cytosine deamination, and involved in multiple physiological and pathological processes. Current methods for whole-genome mapping of uracil all rely on uracil-DNA N-glycosylase (UNG) and are limited in resolution or specificity. Here, we present a UNG-independent Single-Nucleotide resolution Uracil Sequencing (SNU-seq) method utilizing the UdgX protein which specifically excises the uracil and forms a covalent bond with the resulting deoxyribose. SNU-seq was validated on synthetic DNA and applied to mammalian genomes. We found that the uracil content of pemetrexed-treated cells fluctuated along with DNA replication timing. We also accurately detected uracil introduced through cytosine deamination by the cytosine base editor (nCas9-APOBEC) and verified uracil occurrence in "WRC" motif within Activation-Induced Cytidine Deaminase (AID) hotspot regions in CSR-activated UNG-/- B cells.

Profiling the binding sites of RNA-binding protein by LACE-seq

RNA-binding proteins (RBPs) directly interact with various RNAs in living cells to regulate their processing, translation, and stability. Identifying the precise binding sites of RBPs is critical for appreciating their physiological or pathological roles in germline and early embryo development. Current methods typically need millions of cells to map RBP binding positions, which prevents us from appreciating the crucial role of RBPs in early development. Here, we present the LACE-seq method for unbiased mapping of RBP-binding sites at single-nucleotide resolution in fewer cells or even single oocytes. LACE-seq depends on RBP-mediated reverse transcription termination, and linear amplification of the cDNA ends for deep sequencing. To further promote its application, we describe a step-by-step protocol about how to construct a successful LACE-seq library.