Genome sequencing experiments have revolutionized molecular biology by allowing researchers to identify important DNA-encoded elements genome wide. Regions where these elements are found appear as peaks in the analog signal of an assay’s coverage track, and despite the ease with which humans can visually categorize these patterns, the size of many genomes necessitates algorithmic implementations. Commonly used methods focus on statistical tests to classify peaks, discounting that the background signal does not completely follow any known probability distribution and reducing the information-dense peak shapes to simply maximum height. Deep learning has been shown to be highly accurate for many pattern recognition tasks, on par or even exceeding human capabilities, providing an opportunity to reimagine and improve peak calling. We present the peak calling framework LanceOtron, which combines deep learning for recognizing peak shape with multifaceted enrichment calculations for assessing significance. In benchmarking ATAC-seq, ChIP-seq and DNase-seq, LanceOtron outperforms long-standing, gold-standard peak callers through its improved selectivity and near-perfect sensitivity. A fully featured web application is freely available from LanceOtron.molbiol.ox.ac.uk, command line interface via python is pip installable from PyPI at https://pypi.org/project/lanceotron/, and source code and benchmarking tests are available at https://github.com/LHentges/LanceOtron. Supplementary data are available at Bioinformatics online.
Haemoglobin E (HbE) β-thalassaemia causes approximately 50% of all severe thalassaemia worldwide; equating to around 30,000 births per year. HbE β-thalassaemia is due to a point mutation in codon 26 of the human HBB gene on one allele (GAG; glutamatic acid → AAG; lysine, E26K), and any mutation causing severe β-thalassaemia on the other. When inherited together in compound heterozygosity these mutations can cause a severe thalassaemic phenotype. However, if only one allele is mutated individuals are carriers for the respective mutation and have an asymptomatic phenotype (β-thalassaemia trait). Here we describe a base editing strategy which corrects the HbE mutation either to wildtype (WT) or a normal variant haemoglobin (E26G) known as Hb Aubenas and thereby recreates the asymptomatic trait phenotype. We have achieved editing efficiencies in excess of 90% in primary human CD34 + cells. We demonstrate editing of long-term repopulating haematopoietic stem cells (LT-HSCs) using serial xenotransplantation in NSG mice. We have profiled the off-target effects using a combination of circularization for in vitro reporting of cleavage effects by sequencing (CIRCLE-seq) and deep targeted capture and have developed machine-learning based methods to predict functional effects of candidate off-target mutations.
ATAC-seq, ChIP-seq, and DNase-seq have revolutionized molecular biology by allowing researchers to identify important DNA-encoded elements genome-wide. Regions where these elements are found appear as peaks in the analog signal of an assay’s coverage track, and despite the ease with which humans can visually categorize these regions, meaningful peak calls from whole genome datasets require complex analytical techniques. Current methods focus on statistical tests to classify peaks, reducing the information-dense peak shapes to simply maximum height, and discounting that background signals do not completely follow any known probability distribution for significance testing. Deep learning has been shown to be highly accurate for image recognition, on par or exceeding human ability, providing an opportunity to reimagine and improve peak calling. We present the peak calling framework LanceOtron, which combines multifaceted enrichment measurements with deep learning image recognition techniques for assessing peak shape. In benchmarking transcription factor binding, chromatin modification, and open chromatin datasets, LanceOtron outperforms the long-standing, gold-standard peak caller MACS2 through its improved selectivity and near perfect sensitivity. In addition to command line accessibility, a graphical web application was designed to give any researcher the ability to generate optimal peak calls and interactive visualizations in a single step.
Tracking and understanding data quality, analysis and reproducibility are critical concerns in the biological sciences. This is especially true in genomics where next generation sequencing (NGS) based technologies such as ChIP-seq, RNA-seq and ATAC-seq are generating a flood of genome-scale data. However, such data are usually processed with automated tools and pipelines, generating tabular outputs and static visualisations. Interpretation is normally made at a high level without the ability to visualise the underlying data in detail. Conventional genome browsers are limited to browsing single locations and do not allow for interactions with the dataset as a whole. Multi Locus View (MLV), a web-based tool, has been developed to allow users to fluidly interact with genomics datasets at multiple scales. The user is able to browse the raw data, cluster, and combine the data with other analysis and annotate the data. User datasets can then be shared with other users or made public for quick assessment from the academic community. MLV is publically available athttps://mlv.molbiol.ox.ac.uk.
Resolving the gene targets of non-coding genetic variation is the major bottleneck in translating genome wide association studies into mechanistic understanding of complex diseases such as coronary artery disease (CAD). Combining new Transformer-based Machine Learning (ML) approaches trained on cardiovascular epigenetics with high-resolution, allele-specific genomic and transcriptomic technologies we create a highly scalable platform to simultaneously resolve causal variants, cell-type of action, output gene, and direction of effect. When applied to CAD genetics, our ML predicts causal variants from 20,747 candidate SNPs across 9 vessel cell-types and identifies disrupted transcription factor binding motifs using ML feature attributions. We investigate 94 of the top predictions in endothelial cells using Micro Capture-C, revealing the importance of fluid shear stress and TGF-β signaling pathways. We exploit allelic skew in heterozygous cells to demonstrate both variant causality and effect direction, demonstrating this platform can be used to rapidly resolve non-coding genetics in complex disease.