WYMM Tour: Bay Area

Reference genome assemblies have historically excluded repetitive DNA sequences found in and near centromeres, telomeres, and ribosomal DNA regions, limiting the ability to study their organization, evolution, and function using modern genomic and epigenomic tools. Thanks to recent advances in long-read sequencing and assembly technologies, as part of the Telomere-to-Telomere Consortium, we helped to produce the first truly complete assembly of a human genome, including formerly missing repetitive regions that make up ~8% of the genome. We also co-developed DiMeLo-seq, a method that uses nanopore sequencing to map protein-DNA interactions and endogenous DNA methylation marks at high resolution on long, single molecules of DNA, whose sequences can be mapped reliably to repetitive regions. Ongoing work continues to extend the DiMeLo-seq method and to leverage its many advantages for studying chromatin organization at the single-molecule level, including within the most challenging regions of the human genome.

Periodontal disease (PD) is a prolific and chronic inflammatory condition driven by aberrant immune function and perturbation of the oral microbiome. Though specific bacterial species are implicated in PD, strain level identity and functional contributions at the microbial community level remain underexplored. Oral samples have low microbial burden, leading to microbiome study limitations. Newer collection techniques and sequencing technologies have vastly improved microbiome resolution, though these tools have been underutilized in the oral microbiome field. Our study aims to take advantage of Oxford Nanopore’s adaptive sampling technology, which permits selective rejection of non-target reads during sequencing. This technique will enhance coverage of oral metagenomes and identify, at strain-level resolution, persistent microbes that associate severe PD. Longitudinal subgingival plaque, saliva, and oral wash samples were collected from patients with severe PD and from healthy controls. A clinical assessment of health and demographic information was also collected. Samples were collected and preserved in nucleic acid preservative at an initial evaluation, following dental cleaning, and, for the periodontal patients, at a 4-6 week follow up appointment. DNA and RNA were extracted from plaque samples and 16S rRNA sequencing and RNA transcriptomics performed. PD plaque samples will also be sequenced using Oxford Nanopore metagenomic adaptive sampling using the PromethION. This longitudinal study will permit identification of pathogenic strains that persist despite dental intervention in PD patients, paving the way for targeted PD treatment.

Our goal is to develop a Lab Developed Test for clinical confirmation of TREs and CNVs using multiplexed long read adaptive sampling, in order to reduce cost and test turnaround time via rapid Oxford Nanopore sequencing. We optimized Adaptive Sampling (AS) conditions (e.g. loading DNA amount and wash-reload timing) on GridION and P2 solo for output maximization using reference cancel panel and GIAB 12878. We found optimal library loading conditions on GridION and P2 and achieved average 20X on-target depth for gene targets (8-9 fold enrichment) with 5 samples multiplexing. We designed our own AS panel that contains over 130 gene regions (1.4% of genome) targeting TREs, CNVs and difficult genes. Our customized AS panel had similar target enrichment performance as cancer panel after tested with GIAB 24385. We tested our AS panel on P2 using positive samples of known CNVs and TREs and achieved high accuracy (>95% concordance). In addition, we evaluated the performance of our version 5 AS panel for enrichment of Mitochondria genome using positive samples of Mt CNV.

Esophageal cancer is one of the deadliest cancers in the world, with a 5-year U.S. survival rate of only 6% once it has metastasized. There is thus an urgent need for better technologies that can detect esophageal cancer at the earliest stages. We have developed an RNA liquid biopsy platform technology that takes advantage of the unique capabilities and strengths of single-molecule, long-read Oxford Nanopore sequencing technology to characterize the entire cell-free RNA transcriptome in the blood. We discovered over 250,000 novel, unannotated cell-free RNA transcripts in the blood of esophageal cancer patients, which we used to create a custom annotation of the transcriptome to leverage these cancer-specific RNA biomarkers. By training our machine learning models using these novel cell-free RNA features, we were able to classify whether someone has esophageal cancer with near perfect sensitivity and specificity using a minimally invasive blood sample. Notably, our RNA liquid biopsy technology also enabled the early detection of precancerous states, highlighting the power and potential of nanopore sequencing for precision health and cancer prevention.

Metagenomics (mNGS) enables detection of rare and novel pathogens and is a critical tool in public health surveillance. Key barriers to its adoption include cost and complexity of sequencing and bioinformatics analysis. Long-read sequencers from Oxford Nanopore Technologies (ONT) have the advantages of being cost-effective and having fewer infrastructure requirements. Here we develop an optimized pipeline for ONT metagenomics and use it to assess the suitability of ONT for unbiased pathogen detection in host-derived samples. We also compare it to the recently published ONT protocol and find that it performs comparably, across a range of sample types such as stool, respiratory swabs, and serum/plasma. We evaluate the sensitivity of ONT mNGS, and find that ONT is able to detect targets even with very low input RNA of a few picograms.

The onset of age-related diseases is shaped by disease-specific risk factors (genetic and environmental), and a gradual decline in cellular resilience. Genetic risk factors for complex traits have been widely investigated using genome-wide association studies (GWAS), and colocalization analyses of GWAS results with quantitative trait loci (QTL) in disease-relevant tissues have proven useful to identify and prioritize causal genes. However, variants underlying differential stress responses or mediating resilience may become apparent only under distinct physiological or pathological conditions (e.g., young vs. old cells, healthy vs. frail states). Cell villages offer a unique and economic opportunity for large-scale state-dependent QTL mapping. We have developed villages of primary skin fibroblasts from genetically diverse donors. Our aim is to map genes that (a) drive differential responses to various stressors, (b) modulate the capacity to recover from stress (resilience), and (c) mediate variable responses to therapeutic agents. It is also known that epigenetics, especially DNA methylation, plays a major role in age-related diseases. Biological age can be inferred from DNA methylation patterns, and epigenetic changes correlate strongly with diverse disease processes. To enable both genome- and epigenome-wide association studies (GWAS and EWAS), we subjected fibroblasts from all donors to Oxford Nanopore sequencing to obtain both reference genomes and epigenomes. The fibroblast villages can now be used in experiments involving cellular stress responses and therapeutic approaches, allowing us to identify genetic and epigenetic correlates of cellular resilience.