Biology subjects

Lancaster, J.

Publications and source records attributed to Lancaster, J..

8 recordsLinked to original sources

bioRxiv2026

Mapping immune cellular landscapes and vaccine responses across a spectrum of health and immunodeficiency.

Immune responses to infection and vaccination exhibit diversity between individuals that can be shaped by differences in their immune cell landscapes and the signalling, transcriptional, and genetic mechanisms that coordinate immune cell function. Specific antibody deficiency (SAD) and common variable immunodeficiency (CVID) are common forms of predominantly antibody deficiencies that result in poor responses to vaccination. While molecular and cellular causes of the immune dysfunction and poor vaccination responses for individuals with CVID have been reported, immune cell or molecular defects have not yet been identified in SAD. Here, we have used single-cell multi-omics to define the cellular landscapes, transcriptional states, adaptive immune repertoires and protein expression of patients with SAD and CVID before and after polysaccharide vaccination. We discovered that while SAD and CVID exhibit overlapping immune defects, including accumulation of exhausted NK memory cells and dysregulated expression of genes that mediate lipopolysaccharide sensing and clearance by monocytes, individuals with SAD have a unique expansion of cytotoxic CD4+ T cells that correlates with reduced regulatory T cells. In response to vaccination, we observed rapid changes in gene expression associated with lipopolysaccharide responses by monocytes and NF-kB pathway activation in B cells, and an apparent expansion of a CD95+ class-switched memory B cell population that does not occur in patients with lower antigen-specific responses. Together, our findings reveal cellular and molecular factors that underpin variability in vaccine responses and define SAD in a broader spectrum of immune dysfunction.

immunology↗

bioRxiv2026

Single-cell CRISPR activation screens in primary B cells discover gene regulatory mechanisms for hundreds of autoimmune risk loci.

Genome-wide association studies (GWAS) have discovered thousands of genetic variants linked to autoimmune disease, and yet the molecular pathways underlying autoimmunity have remained elusive. A key challenge is that >90% of identified GWAS risk loci are in non-coding genomic regions making it difficult to predict their relevance to disease. Here, we have curated fine-mapped non-coding risk variants from over 30 different autoimmune traits including common conditions such as systemic lupus erythematosus (SLE), Crohns disease, and multiple sclerosis, and reveal shared genetic signatures between diverse autoimmune diseases. We subsequently performed a high-throughput single-cell multi-omic CRISPR activation screen targeting 763 autoimmune risk loci in primary human B cells (a highly relevant cell type to autoimmune diseases) and discover 524 cis-regulatory target gene effects for 378 risk loci, with many risk loci regulating multiple gene targets. This Single Cell Analysis of Non-coding Distal Autoimmune Loci (SCANDAL) provides a powerful experimental resource linking non-coding risk loci to many disease-relevant genes, including lowly-expressed cytokines and transcription factors for which perturbation effects can be difficult to quantify with other CRISPR-based strategies. We reveal how increased transcriptional activity at one non-coding risk locus can drive transcription at other risk loci within the same regulatory landscape that may be relevant to understand genetic pleiotropy of autoimmune diseases. Finally, we quantified allele-specific effects on target gene expression with massive parallel reporter assays and prime editing to discover a gain-of-function variant associated with SLE that controls expression of the transcription factor REL/cREL which subsequently binds dozens of risk loci and target genes associated with different autoimmune diseases. Our study provides a valuable resource linking non-coding risk loci with their cis-regulatory target genes and advances our understanding of the shared genetic networks and mechanisms involved in autoimmunity.

genomics↗

bioRxiv2026

Pan-cell-type prediction of splicing patterns from sequence and splicing factor expression

AO_SCPLOWBSTRACTC_SCPLOWAlternative splicing is a core determinant of cell-type-specific gene expression in humans, and its dysregulation contributes to many diseases including neurodegeneration, autoimmunity, and cancer. However, current deep learning models for predicting RNA expression from sequence are limited in how they handle cellular context dependence. These models typically achieve cell-type specificity by training separate models or heads for each tissue or cell-type, which assumes discrete, predefined cell types. This design not only prevents learning from pathological or experimentally perturbed transcriptomes, but also prevents generalization to new cellular contexts. Here we introduce PanExonNet, a deep learning framework that integrates cis- and trans-regulation by conditioning sequence-based splicing predictions on an inferred "splicing state" derived from the expression of RNA-binding proteins (RBPs) and spliceosome components. PanExonNet is trained on diploid, individual-specific gene sequences containing indels to predict splicing profiles derived from short-read RNA-seq, and it outputs multiple tracks at single-nucleotide resolution as well as donor-acceptor junction usage. Compared to multi-headed baselines such as Borzoi and Pangolin, PanExonNet exhibits substantially higher context specificity--as quantified by a {Delta}PSI correlation metric designed to isolate context-specific variation--and, crucially, generalizes to unseen cell types. Adding perturbational knockdown followed by RNA sequencing (KD-RNA-seq) data to the training set further improves generalization. This performance is enabled in part by our introduction of contextualizable convolutions, a modular layer that may broadly benefit genomic sequence modeling. This framework provides a scalable foundation for future DNA-to-RNA models that could improve variant effect prediction, the design of oligonucleotide therapeutics, and biomarker discovery across diverse cellular contexts.

genomics↗

bioRxiv2025

Benchmarking long-read RNA-sequencing technologies with LongBench: a cross-platform reference dataset profiling cancer cell lines with bulk and single-cell approaches

Long-read RNA sequencing enables full-length transcript profiling and improved isoform resolution, but variable platforms and evolving chemistries demand careful benchmarking for reliable application. We present LongBench, a matched, multi-platform reference dataset spanning bulk, single-cell, and single-nucleus transcriptomics across eight human lung cancer cell lines with synthetic spike-in controls. LongBench in-corporates three state-of-the-art long-read protocols alongside Illumina short reads: Oxford Nanopore Technologies (ONT) PCR-cDNA, ONT direct RNA, and PacBio Kinnex. We systematically evaluate transcript capture, quantification accuracy, differential expression, isoform usage, variant detection, and allele-specific analyses. Our results show high concordance in gene-level differential analyses across protocols, but reduced consistency for transcript-level and isoform analyses due to length- and platform-dependent biases. Single-cell long-read data are highly concordant with bulk for high-confidence features, though single-nuclei data show reduced feature detection. LongBench provides one of the largest publicly available long-read benchmarking resources, enabling rigorous cross-platform evaluation and guiding technology selection for transcriptomic research.

genomics↗

bioRxiv2025

Towards accurate, reference-free differential expression: A comprehensive evaluation of long-read de novo transcriptome assembly

Long-read RNA sequencing has significantly advanced transcriptomics by enabling the full length of transcripts to be assessed. However, current analysis methods often depend on a high-quality reference genome and gene annotation. Recently, de novo assembly methods have been developed to utilise long-read data in cases where a reference genome is unavailable, such as in non-model organisms. Despite the potential of these tools, there remains a lack of benchmarking and established protocols for optimal reference-free, long-read transcriptome assembly and differential expression analysis. Here, we comprehensively evaluate the current state-of-the-art long-read de novo transcriptome assembly tools, RATTLE, RNA-Bloom2 and isONform, and compare their performance to one of the leading short-read assemblers, Trinity. We assess various metrics, including assembly quality and computational efficiency, across a range of datasets, which include simulated data and spike-in sequin transcripts, where ground truth is known, and real data from human and pea (Pisum sativum) samples, using a reference-based approach to define truth. To represent contemporary analysis scenarios, the datasets cover depths from 6 million to 60 million reads, Oxford Nanopore Technologies (ONT) cDNA, ONT direct RNA and Pacific Biosciences (PacBio) 10x single-cell sequencing. Critically, we assessed the downstream impact of assembly choice on the detection of differential gene and transcript expression. Our results confirm that long reads generate longer assembled transcripts than short-reads for reference-free analysis, though limitations remain compared to reference-guided approaches, and suggest scope for improved accuracy and reduced redundancy. Of the de novo pipelines, RNA-Bloom2, coupled with Corset for transcript clustering, was the best performing in terms of both accuracy and computational efficiency. Our findings offer guidance when selecting the most effective strategy for long-read differential expression analysis, when a high-quality reference genome is unavailable.

bioinformatics↗

bioRxiv2024

Transcriptomic alterations including p53 pathway dysregulation prime DNMT3A mutant cells for transformation

DNMT3A mutations are prevalent in haematologic malignancies. Our mouse model introduced the murine homologue (R878H) of the human hotspot R882H mutation into the mouse Dnmt3a locus, resulting in globally reduced DNA methylation in all tissues. Mice with heterozygous R878H mutations developed {gamma}-radiation induced thymic lymphoma more rapidly than control mice, suggesting a vulnerability to stress stimuli in Dnmt3aR878H/+ cells. In competitive transplantations, Dnmt3aR878H/+ Lin-Sca-1+Kit+ (LSK) cells had a competitive advantage over wt cells, indicating a self-renewal phenotype at the expense of differentiation. RNA-sequencing of Dnmt3aR878H/+ LSKs exposed to low dose {gamma}-radiation showed downregulation of the p53 pathway. Accordingly, reduced PUMA expression was observed by flow cytometry in the bone marrow of {gamma}-irradiated Dnmt3aR878H/+ mice due to altered p53 signalling. These findings provide new insights as to how DNMT3A mutations cause subtle changes in the transcriptome of LSK cells which contribute to their increased self-renewal and propensity for malignant transformation. SIGNIFICANCEHotspot DNMT3A R882H mutations are overrepresented in leukaemia suggesting that they confer a susceptibility to disease if further mutations are acquired. To advance therapies for DNMT3A mutant disease, it is essential to understand the mechanisms by which this mutation primes cells for malignant transformation.

cancer biology↗

bioRxiv2024

Measuring X inactivation skew for retinal diseases with adaptive nanopore sequencing

X-linked genetic disorders typically affect females less severely than males due to the presence of a second X chromosome not carrying the deleterious variant. However, the phenotypic expression in females is highly variable, which may be explained by an allelic skew in X chromosome inactivation. Accurate measurement of X inactivation skew is crucial to understand and predict disease phenotype in carrier females, with prediction especially relevant for degenerative conditions. We propose a novel approach using nanopore sequencing to quantify skewed X inactivation accurately. By phasing sequence variants and methylation patterns, this single assay reveals the disease variant, X inactivation skew, its directionality, and is applicable to all patients and X-linked variants. Enrichment of X-chromosome reads through adaptive sampling enhances cost-efficiency. Our study includes a cohort of 16 X-linked variant carrier females affected by two X-linked inherited retinal diseases: choroideremia and RPGR-associated retinitis pigmen-tosa. As retinal DNA cannot be readily obtained, we instead determine the skew from peripheral samples (blood, saliva and buccal mucosa), and correlate it to phenotypic outcomes. This revealed a strong correlation between X inactivation skew and disease presentation, confirming the value in performing this assay and its potential as a way to prioritise patients for early intervention, such as gene therapy currently in clinical trials for these conditions. Our method of assessing skewed X inactivation is applicable to all long-read genomic datasets, providing insights into disease risk and severity and aiding in the development of individualised strategies for X-linked variant carrier females.

genomics↗

bioRxiv2017

Bayesian Optimisation for Neuroimaging Pre-processing in Brain Age Prediction

Neuroimaging-based age predictions using machine learning have been shown to relate to cognitive performance, health outcomes and progression of neurodegenerative disease. However, even leading age-prediction algorithms contain measurement error, motivating efforts to improve experimental pipelines. T1-weighted MRI is commonly used for age prediction, and the pre-processing of these scans involves normalisation to a common template and resampling to a common voxel size, followed by spatial smoothing. Resampling parameters are often selected arbitrarily. Here, we sought to improve brain-age prediction accuracy by optimising resampling parameters using Bayesian optimisation.\n\nUsing data on N=2001 healthy individuals (aged 16-90 years) we trained support vector machines to i) distinguish between young (<50 years) and old (>50 years) brains and ii) predict chronological age, with accuracy assessed using cross-validation. We also evaluated model generalisability to the Cam-CAN dataset (N=648, aged 18-88 years). Bayesian optimisation was used to identify optimal voxel size and smoothing kernel size for each task. This procedure adaptively samples the parameter space to evaluate accuracy across a range of possible parameters, using independent sub-samples to iteratively assess different parameter combinations to arrive at optimal values.\n\nWhen distinguishing between young and old brains a classification accuracy of 96.25% was achieved, with voxel size = 11.5mm3 and smoothing kernel = 2.3mm. For predicting chronological age, a mean absolute error (MAE) of 5.08 years was achieved, with voxel size = 3.73mm3 and smoothing kernel = 3.68mm. This was compared to performance using default values of 1.5mm3 and 4mm respectively, which gave a MAE = 5.48 years, a 7.3% improvement. When assessing generalisability, best performance was achieved when applying the entire Bayesian optimisation framework to the new dataset, out-performing the parameters optimised for the initial training dataset.\n\nOur study demonstrates the proof-of-principle that neuroimaging models for brain age prediction can be improved by using Bayesian optimisation to select more appropriate pre-processing parameters. Our results suggest that different parameters are selected and performance improves when optimisation is conducted in specific contexts. This motivates use of optimisation techniques at many different points during the experimental process, which may result in improved statistical sensitivity and reduce opportunities for experimenter-led bias.

neuroscience↗