WYMM Tour: Stockholm

We demonstrate that CpG methylation detection from 7,179 nanopore sequenced DNA samples is highly accurate and consistent with 132 oxidative bisulfite sequenced (oxBS) samples, isolated from the same blood draws. We introduce quality filters for CpGs, that further enhance accuracy of CpG methylation detection from nanopore sequenced DNA, while removing at most 30% of CpGs. We evaluate the per-site performance of CpG methylation detection across different genomic features and CpG methylation rates and demonstrate how the latest R10.4 flowcell chemistry and base-calling algorithms improve CpG methylation detection from nanopore sequencing. We show how we can use ONT CpG methylation data to identify variants associating with allele specific methylation (ASM-QTLs) and how the ASM-QTLs drive most of the correlation found between gene expression and CpG methylation.

Structural variants (SVs) affect more of the cancer genome than any other type of somatic alterations and include aneuploidies and rearrangements of the genome that can span up to megabases in size. SVs can manifest as simple events such as deletions, duplications and translocations, and more complex events such as chromothripsis or breakage fusion bridge cycles. Accumulation of SVs in cancer genomes is associated with aggressive disease, and the first targeted anti-cancer therapies were developed against the proteins affected by specific SVs. SVs have been challenging to resolve using standard short-read sequencing due to their often complex nature and when occurring in difficult-to-map regions of the genome. Here, we applied Oxford nanopore long-read sequencing (LRS) to resolve SVs in cancer genomes. We find that LRS can detect a highly challenging driver SV in cancer genomes, which involves duplication of a repetitive sequences. Moreover, we find that LRS can aid to resolve highly complex SVs in cancer genomes that involve dysregulation of several known cancer genes.

Individuals with intellectual disability (ID) and/or neurodevelopment disorders (NDDs) are currently investigated with several different approaches in clinical genetic diagnostics. Long-read genome sequencing (lrGS) enables analyses such as structural variant detection, phasing, and methylomics in one single experiment. Here we evaluate PromethION GS lrGS as a comprehensive genetic diagnostic test for pediatric neurological disorders. Our aim is to sequence 100 individuals with pediatric neurological disorders with lrGS in parallel with short read genome sequencing (srGS), allowing for a direct comparison. Here we present our initial results. On average, we obtain 35X median coverage, read-length N50 of 10 kbp, and error-rate of 3%. Using a custom pipeline, we perform SNV, SV, repeat expansion, and methylome analyses. Through these analyses, we detect the full range of pathogenic variation, including SV and SNV. Next, we perform a diversity of epigenetic analyses, including allele-specific methylation, aberrant promoter usage, and overlay with public RNA-seq datasets.

Correlating the promoter methylation level with GTEX expression levels, we find that genes not expressed in blood are prone to be heavily methylated, while genes highly expressed are unlikely to be methylated (p < 0.01), indicating that the lrGS methylome may be used as a proxy for gene expression. Overall, we highlight the usefulness of GS technologies in pediatric genomics, moving us towards more effective, personalized therapeutic interventions for these children.

Diagnostic work-up of CNS tumors often requires extensive molecular analyses to conclude in a CNS WHO (2021) concordant diagnosis. Diagnostically relevant features comprise the epigenetic profile, amplifications, point mutations, deletions, gene fusions and duplications. While in routine diagnostic work-up several analyses have to be performed separately, comprising NGS tools, FISH, pyrosequencing, Sanger sequencing and methylome analyses, we were aiming for a WGS approach using nanopore sequencing to receive all diagnostically relevant information from a single analysis to reduce the hands on work and to expand the information on the tumor tissue as well as to speed up the molecular analyses. In this study we have tested about 50 CNS tumors using Promethion flow cells and compared the molecular findings with the results from the routine diagnostic work-up. In most of the samples the similar diagnosis was reached using methylation profiling. In 3 cases the tumor cell content was too low, leading to a discordant methylation based diagnosis and molecular alteration frequency below automatized detection. In turn, 2 diagnoses could be refined with confirmation and adaptation of the diagnoses obtained from routine diagnostic work-up.

Nearly all our genes undergo alternative splicing, the process by which different transcripts are produced from a single gene. Alternative splicing is fundamental to brain development and physiology, and its disruption has been strongly associated with neuropsychiatric disorders. Genes expressed in the brain tend to be complex having hundreds of currently annotated isoforms. To understand their regulation across development, brain regions, and across cell types we developed and applied novel protocols for long read RNA Sequencing.

As many of these genes often have low to medium abundance in the brain leading to a significant power loss in isoforms quantification, we combined CaptureSeq target enrichment nanopore sequencing to comprehensively characterise and quantify the RNA isoform profiles of 1221 protein-coding and lncRNA genes in human post-mortem tissue. We first demonstrate that this approach preserves the quantitative aspect of transcript expression. We then profiled 52 prefrontal cortex, caudate and hippocampus samples from 20 and identified 44,624 unique isoforms, of which over 88% were previously unannotated. This approach enabled the identification of differentially expressed transcripts between tissues and during development, many of these isoforms switches are predicted to impact the protein structure and coding potential. To further understand transcript regulation during neuronal development we also applied single cell long read. These results suggest that the human brain transcriptome is far more complex and dynamic than current annotations indicate.