Biology subjects

Flicek, P.

Publications and source records attributed to Flicek, P..

At least 19 recordsLinked to original sources

bioRxiv2026

Explainable machine learning relates histological to genomic pathology

Background & AimsHaematoxylin and eosin (H&E) staining remains the diagnostic gold standard for solid cancers, including hepatocellular carcinoma, and is increasingly complemented by genomic profiling for precision medicine. Inferring genomic alterations directly from H&E images could streamline testing, but heterogeneity and biases in human training data limit interpretation of genotype-phenotype associations. Here, we aimed to relate histologic to genomic pathology to provide biological explainability for mutation prediction models and assess the impact of germline variation on model performance. MethodsWe analysed 597 murine liver tumours with matched whole-genome sequencing and histopathology (163,835 image tiles; 22.9 million nuclei). Our controlled in vivo design accounted for germline variation, biological sex, and causal mutagen (N-diethylnitrosamine), removing confounding factors present in human cohorts. We trained and evaluated deep learning and supervised machine learning models to predict germline variation and cancer driver alterations from H&E. ResultsModelling accurately predicted germline and somatic alterations from histology, at both locus-specific and genome-wide scales. Quantitative image analysis revealed an unexpected association between Egfr driver mutations and hepatic steatosis, linking genotype to an interpretable morphological phenotype. While model performance declined when applied to tumours from unrepresented genetic backgrounds, this limitation was biologically informative, revealing strain-dependent differences in tumour evolution, notably the prevalence of whole-genome duplication. ConclusionsMachine learning integration of histological and genomic pathology enables accurate, interpretable inference of genetic alterations from H&E, potentially reducing reliance on costly ancillary molecular assays. Our predictions are supported by human-interpretable biological features, addressing concerns around "black-box" technologies. However, caution is required when applying such methods to samples with a genetic background that, even if closely related, is beyond the genetic horizon of training data.

pathology↗

bioRxiv2025

Germline and somatic variation influence nuclear-mitochondrial crosstalk in tumourigenesis

Hepatocellular carcinoma is the most prevalent primary liver cancer and arises from hepatocytes, which are very metabolically active and contain abundant mitochondria. Despite extensive characterisation of nuclear driver genes and pathways, the role of crosstalk between the mitochondrial and nuclear genomes in mitochondrial mutagenesis remains poorly understood. To address this, we leveraged a chemical carcinogenesis model of liver cancer in four mouse strains. We analysed hundreds of tumours with paired whole-genome sequencing and total RNA sequencing to comprehensively characterise mitochondrial mutations and expression. Following de novo assembly of strain-specific mitochondrial genomes, we devised an accurate and efficient heteroplasmy detection approach and analysed mitochondrial DNA (mtDNA) content and transcription. Across strains, there was high concordance of the number of heteroplasmies, mutation signatures, and variant allele frequencies. In contrast, the heteroplasmy rate and genic locations differed between strains, suggesting strain-specificity in DNA repair mechanisms in early tumour development. Independent of age or environmental factors, tumours had lower mtDNA content than adjacent normal liver, suggesting mitochondrial loss during tumourigenesis. Additionally, within individual strains, mtDNA content differences were associated with nuclear driver gene choice. Our comprehensive characterisation of mitochondrial genomes and expression patterns demonstrates that both germline and somatic genetic variation influence tumour mtDNA content, mutation patterns, and gene expression.

cancer biology↗

bioRxiv2025

Genetic background sets the trajectory of cancer evolution

Human cancers are heterogeneous. Their genomes evolve from genetically diverse germlines in complex and dynamic environments, including exposure to potential carcinogens. This heterogeneity of humans, our environmental exposures, and subsequent tumours makes it challenging to understand the extent to which cancer evolution is predictable. Addressing this limitation, we re-ran early tumour evolution hundreds of times in diverse, inbred mouse strains, capturing genetic variation comparable to and beyond that found in human populations. The sex, environment, and carcinogenic exposures were all controlled and tumours comprehensively profiled with whole genome and transcriptome sequencing. Within a strain, there was a high degree of consistency in the mutational landscape, a limited range of driver mutations, and all strains converged on the acquisition of a MAPK activating mutation with similar transcriptional disruption of that pathway. Despite these similarities in the phenotypic state of tumours, different strains took markedly divergent paths to reach that state. This included pronounced biases in the precise driver mutations, the strain specific occurrence of whole genome duplication, and differences in subclonal selection that reflected both cancer susceptibility and tumour growth rate. These results show that interactions between the germline genome and the environment are highly deterministic for the trajectory of tumour genome evolution, and even modest genetic divergence can substantially alter selection pressures during cancer development, influencing both cancer risk and the biology of the tumour that develops.

genomics↗

bioRxiv2024

GENCODE: massively expanding the lncRNA catalog through capture long-read RNA sequencing

Accurate and complete gene annotations are indispensable for understanding how genome sequences encode biological functions. For more than twenty years, the GENCODE consortium has developed reference annotations for the human and mouse genomes, becoming a foundation for biomedical and genomics communities worldwide. Nevertheless, collections of important yet poorly-understood gene classes like long non-coding RNAs (lncRNAs) remain incomplete and scattered across multiple, uncoordinated catalogs. To address this, GENCODE has undertaken the most comprehensive lncRNA annotation effort to date. This is founded on the manually supervised computational annotation of full-length targeted long-read sequencing, on matched embryonic and adult tissues, of orthologous regions in human and mouse. Altogether 17,931 human genes (140,268 transcripts) and 22,784 mouse genes (136,169 transcripts) have been added to the GENCODE catalog representing a 2-fold and 6-fold growth in transcripts, respectively - the greatest increase in the number of annotated human genes since the sequencing of the human genome. Our targeted design assigned human-mouse orthologs at a rate beyond previous studies, tripling the number of human disease-associated lncRNAs that have mouse orthologs. Novel lncRNA genes consistently exhibit biological signals of functionality, and they greatly enhance the functional interpretability of the human genome. While poorly expressed in bulk RNA-Seq samples, many of them are highly expressed in specific cell populations, maybe even contributing to cell-type determination. The expanded GENCODE lncRNA annotations mark a critical step toward deciphering the human and mouse genomes.

genomics↗

bioRxiv2023

Transcriptomics and chromatin accessibility in multiple African population samples

Mapping the functional human genome and impact of genetic variants is often limited to European-descendent population samples. To aid in overcoming this limitation, we measured gene expression using RNA sequencing in lymphoblastoid cell lines (LCLs) from 599 individuals from six African populations to identify novel transcripts including those not represented in the hg38 reference genome. We used whole genomes from the 1000 Genomes Project and 164 Maasai individuals to identify 8,881 expression and 6,949 splicing quantitative trait loci (eQTLs/sQTLs), and 2,611 structural variants associated with gene expression (SV-eQTLs). We further profiled chromatin accessibility using ATAC-Seq in a subset of 100 representative individuals, to identity chromatin accessibility quantitative trait loci (caQTLs) and allele-specific chromatin accessibility, and provide predictions for the functional effect of 78.9 million variants on chromatin accessibility. Using this map of eQTLs and caQTLs we fine-mapped GWAS signals for a range of complex diseases. Combined, this work expands global functional genomic data to identify novel transcripts, functional elements and variants, understand population genetic history of molecular quantitative trait loci, and further resolve the genetic basis of multiple human traits and disease.

genomics↗

bioRxiv2023

Building pangene sets from plant genome alignments confirms presence-absence variation

Consistent gene annotation in crops is becoming harder as genomes for new cultivars are frequently published. Gene sets from recently sequenced accessions have different gene identifiers to those on the reference accession, and might be of higher quality due to technical advances. For these reasons there is a need to define pangenes, which represent all known syntenic orthologues for a gene model and can be linked back to the original annotation sources. A pangene set effectively summarizes our current understanding of the coding potential of a crop and can be used to inform gene model annotation in new cultivars. Here we present an approach (get_pangenes) to identify and analyze pangenes that is not biased towards the reference annotation. The method involves computing Whole Genome Alignments (WGA), which are used to estimate gene model overlaps. After a benchmark on Arabidopsis, rice, wheat and barley datasets, we find that minimap2 performs better than the GSAlign WGA algorithm. Our results show that pangenes recapitulate known phylogeny-based orthologies while adding extra core gene models in rice. More importantly, get_pangenes can also produce clusters of genome segments (gDNA) that overlap with gene models annotated in other cultivars. By lifting-over CDS sequences, gDNA clusters can help refine gene models across individuals and confirm or reject observed gene Presence-Absence Variation. A collection of flowering-related genes from the barley pangenome are discussed in detail. Documentation and source code are available at https://github.com/Ensembl/plant-scripts.

genomics↗

bioRxiv2022

Large-scale benchmarking of circRNA detection tools reveals large differences in sensitivity but not in precision

The detection of circular RNA molecules (circRNAs) is typically based on short-read RNA sequencing data processed by computational detection tools. During the last decade, a plethora of such tools have been developed, but a systematic comparison with orthogonal validation is missing. Here, we set up a circRNA detection tool benchmarking study, in which 16 tools were used and detected over 315,000 unique circRNAs in three deeply sequenced human cell types. Next, 1,516 predicted circRNAs were empirically validated using three orthogonal methods. Generally, tool-specific precision values are high and similar (median of 98.8%, 96.3%, and 95.5% for qPCR, RNase R, and amplicon sequencing, respectively) whereas the sensitivity and number of predicted circRNAs (ranging from 1,372 to 58,032) are the most significant tool differentiators. Furthermore, we demonstrate the complementarity of tools through the increase in detection sensitivity by considering the union of highly-precise tool combinations while keeping the number of false discoveries low. Finally, based on the benchmarking results, recommendations are put forward for circRNA detection and validation.

genomics↗

bioRxiv2022

The complete sequence of a human Y chromosome

The human Y chromosome has been notoriously difficult to sequence and assemble because of its complex repeat structure including long palindromes, tandem repeats, and segmental duplications1-3. As a result, more than half of the Y chromosome is missing from the GRCh38 reference sequence and it remains the last human chromosome to be finished4, 5. Here, the Telomere-to-Telomere (T2T) consortium presents the complete 62,460,029 base pair sequence of a human Y chromosome from the HG002 genome (T2T-Y) that corrects multiple errors in GRCh38-Y and adds over 30 million base pairs of sequence to the reference, revealing the complete ampliconic structures of TSPY, DAZ, and RBMY gene families; 41 additional protein-coding genes, mostly from the TSPY family; and an alternating pattern of human satellite 1 and 3 blocks in the heterochromatic Yq12 region. We have combined T2T-Y with a prior assembly of the CHM13 genome4 and mapped available population variation, clinical variants, and functional genomics data to produce a complete and comprehensive reference sequence for all 24 human chromosomes.

genomics↗

bioRxiv2022

DNA methylation patterns of transcription factor binding regions characterize their functional and evolutionary contexts

BackgroundDNA methylation is an important epigenetic modification which has numerous roles in modulating genome function. Its levels are spatially correlated across the genome, typically high in repressed regions but low in transcription factor (TF) binding sites and active regulatory regions. However, the mechanisms establishing genome-wide and TF binding site methylation patterns are still unclear. ResultsWe used a comparative approach to investigate the association of DNA methylation to TF binding evolution in mammals. Specifically, we experimentally profiled DNA methylation and combined this with published occupancy profiles of five distinct TFs (CTCF, CEBPA, HNF4A, ONECUT1, FOXA1) in the liver of five mammalian species (human, macaque, mouse, rat, dog). TF binding sites were lowly methylated, but they often also had intermediate methylation levels. Employing a classification and clustering approach, we extracted distinct and species conserved patterns of DNA methylation levels at TF bound regions. CEBPA, HNF4A, ONECUT1 and FOXA1 shared the same methylation patterns, while CTCFs differed. These patterns characterize alternative functions and chromatin landscapes of TF bound regions. Leveraging our phylogenetic framework, we found DNA methylation gain upon evolutionary loss of TF occupancy, indicating coordinated evolution. Furthermore, each methylation pattern has its own evolutionary trajectory reflecting its genomic contexts. ConclusionsOur epigenomic analyses found that specific DNA methylation profiles characterize TF binding, and are associated to their regulatory activity, chromatin contexts, and evolutionary trajectories.

genomics↗

bioRxiv2022

Strand-resolved mutagenicity of DNA damage and repair

DNA base damage is a major source of oncogenic mutations1. Such damage can produce strand-phased mutation patterns and multiallelic variation through the process of lesion segregation2. Here, we exploited these properties to reveal how strand-asymmetric processes, such as replication and transcription, shape DNA damage and repair. Despite distinct mechanisms of leading and lagging strand replication3,4, we observe identical fidelity and damage tolerance for both strands. For small DNA adducts, our results support a model in which the same translesion polymerase is recruited on-the-fly to both replication strands, starkly contrasting the strand asymmetric tolerance of bulky adducts5. We find that DNA damage tolerance is also common during transcription, where RNA-polymerases frequently bypass lesions without triggering repair. At multiple genomic scales, we show the pattern of DNA damage induced mutations is largely shaped by the influence of DNA accessibility on repair efficiency, rather than gradients of DNA damage. Finally, we reveal specific genomic conditions that can corrupt the fidelity of nucleotide excision repair and actively drive oncogenic mutagenesis. These results provide insight into how strand-asymmetric mechanisms underlie the formation, tolerance, and repair of DNA damage, thereby shaping cancer genome evolution.

genomics↗

bioRxiv2021

Multiple adaptive solutions to face climatic constraints: novel insights in the debate over the role of convergence in local adaptation

The extent to which genomic convergence shapes locally adapted phenotypes in different species remains a fundamental question in evolutionary biology. To bring new insights to this debate we set up a framework which aimed to compare the adaptive trajectories of two domesticated mammal species co-distributed in diversified landscapes. We sequenced the genomes of 160 sheep and 161 goats extensively managed along environmental gradients, including temperature, rainfall, seasonality and altitude, to identify genes and biological processes shaping local adaptation. Allele frequencies at adaptive loci were rarely found to vary gradually along environmental gradients, but rather displayed a discontinuous shift at the extremities of environmental clines. Of the more than 430 adaptive genes identified, only 6 were orthologous between sheep and goats and those responded differently to environmental pressures, suggesting different adaptive mechanisms in these two closely related species. Such diversity of adaptive pathways may result from a high number of biological functions involved in adaptation to multiple eco-climatic gradients, and provides more arguments for the role of contingency and stochasticity in adaptation rather than repeatability.

evolutionary biology↗

bioRxiv2021

A minimally morphologically destructive approach for DNA retrieval and whole genome shotgun sequencing of pinned historic Dipteran vector species

Museum collections contain enormous quantities of insect specimens collected over the past century, covering a period of increased and varied insecticide usage. These historic collections are therefore incredibly valuable as genomic snapshots of organisms before, during, and after exposure to novel selective pressures. However, these samples come with their own challenges compared to present-day collections, as they are fragile and retrievable DNA is low yield and fragmented. In this paper we tested several DNA extraction procedures across pinned historic Diptera specimens from four disease vector genera: Anopheles, Aedes, Culex and Glossina. We identify an approach that minimizes morphological damage while maximizing DNA retrieval for Illumina library preparation and sequencing that can accommodate the fragmented and low yield nature of historic DNA. We identify several key points in retrieving sufficient DNA while keeping morphological damage to a minimum: an initial rehydration step, a short incubation without agitation in a low salt Proteinase K buffer, and critical point drying of samples post-extraction to prevent tissue collapse caused by air drying. The suggested method presented here provides a solid foundation for exploring the genomes and morphology of historic Diptera collections. Significance statementLarge museum collections of pinned insects could provide important snapshots of genomes through time, but unfortunately DNA retrieval from such fragile samples often leads to severe morphological damage, especially in delicate species such as disease transmitting Diptera. In this study we have worked on a combined method that minimizes morphological damage while maximizing the retrieval of DNA from dry pinned Diptera species. We identified the importance of tissue rehydration, gentle DNA lysis buffer incubation, and critical point drying to restore collapsed tissues. We hope this approach will make it possible for more historic insect specimens to become available for genomic research while ensuring they remain intact for morphological studies.

evolutionary biology↗

bioRxiv2021

A community-driven roadmap to advance research on translated open reading frames detected by Ribo-seq

Ribosome profiling (Ribo-seq) has catalyzed a paradigm shift in our understanding of the translational vocabulary of the human genome, discovering thousands of translated open reading frames (ORFs) within long non-coding RNAs and presumed untranslated regions of protein-coding genes. However, reference gene annotation projects have been circumspect in their incorporation of these ORFs due to uncertainties about their experimental reproducibility and physiological roles. Yet, it is indisputable that certain Ribo-seq ORFs make stable proteins, others mediate gene regulation, and many have medical implications. Ultimately, the absence of standardized ORF annotation has created a circular problem: while Ribo-seq ORFs remain unannotated by reference biological databases, this lack of characterisation will thwart research efforts examining their roles. Here, we outline the initial stages of a community-led effort supported by GENCODE / Ensembl, HGNC and UniProt to produce a consolidated catalog of human Ribo-seq ORFs.

genomics↗

bioRxiv2021

Efficient masking of plant genomes by combining kmer counting and curated repeats

Ii.Summary/AbstractThe annotation of repetitive sequences within plant genomes can help in the interpretation of observed phenotypes. Moreover, repeat masking is required for tasks such as whole-genome alignment, promoter analysis or pangenome exploration. While homology-based annotation methods are computationally expensive, k-mer strategies for masking are orders of magnitude faster. Here we benchmark a two-step approach, where repeats are first called by k-mer counting and then annotated by comparison to curated libraries. This hybrid protocol was tested on 20 plant genomes from Ensembl, using the kmer-based Repeat Detector (Red) and two repeat libraries (REdat and nrTEplants, curated for this work). We obtained repeated genome fractions that match those reported in the literature, but with shorter repeated elements than those produced with conventional annotators. Inspection of masked regions overlapping genes revealed no preference for specific protein domains. Half of Red masked sequences can be successfully classified with nrTEplants, with the complete protocol taking less than 2h on a desktop Linux box. The repeat library and the scripts to mask and annotate plant genomes can be obtained at https://github.com/Ensembl/plant-scripts.

genomics↗

bioRxiv2021

High coverage whole genome sequencing of the expanded 1000 Genomes Project cohort including 602 trios

The 1000 Genomes Project (1kGP) is the largest fully open resource of whole genome sequencing (WGS) data consented for public distribution of raw sequence data without access or use restrictions. The final release of the 1kGP included 2,504 unrelated samples from 26 populations and was based primarily on low coverage WGS. Here, we present a new, high coverage 3,202-sample WGS 1kGP resource, sequenced to a targeted depth of 30X using the Illumina NovaSeq 6000 system, which now includes 602 complete trios. We performed SNV/INDEL calling against the GRCh38 reference using GATKs HaplotypeCaller, and generated a comprehensive set of SVs by integrating multiple analytic methods through a sophisticated machine learning model. We make all the data generated as part of this project publicly available and we envision it to become the new de facto public resource for the worldwide genomics and genetics community.

genomics↗

bioRxiv2020

The Ensembl COVID-19 resource: Ongoing integration of public SARS-CoV-2 data

The COVID-19 pandemic has seen unprecedented use of SARS-CoV-2 genome sequencing for epidemiological tracking and identification of emerging variants. Understanding the potential impact of these variants on the infectivity of the virus and the efficacy of emerging therapeutics and vaccines has become a cornerstone of the fight against the disease. To support the maximal use of genomic information for SARS-CoV-2 research, we launched the Ensembl COVID-19 browser, incorporating a new Ensembl gene set, multiple variant sets (including novel variation calls), and annotation from several relevant resources integrated into the reference SARS-CoV-2 assembly. This work included key adaptations of existing Ensembl genome annotation methods to model ribosomal slippage, stringent filters to elucidate the highest confidence variants and utilisation of our comparative genomics pipelines on viruses for the first time. Since May 2020, the content has been regularly updated and tools such as the Ensembl Variant Effect Predictor have been integrated. The Ensembl COVID-19 browser is freely available at https://covid-19.ensembl.org.

bioinformatics↗

bioRxiv2020

De novo assembly of 64 haplotype-resolved human genomes of diverse ancestry and integrated analysis of structural variation

Long-read and strand-specific sequencing technologies together facilitate the de novo assembly of high-quality haplotype-resolved human genomes without parent-child trio data. We present 64 assembled haplotypes from 32 diverse human genomes. These highly contiguous haplotype assemblies (average contig N50: 26 Mbp) integrate all forms of genetic variation across even complex loci such as the major histocompatibility complex. We focus on 107,590 structural variants (SVs), of which 68% are inaccessible by short-read sequencing. We identify new SV hotspots (spanning megabases of gene-rich sequence), characterize 130 of the most active mobile element source elements, and find that 63% of all SVs arise by homology-mediated mechanisms--a twofold increase from previous studies. Our resource now enables reliable graph-based genotyping from short reads of up to 50,340 SVs, resulting in the identification of 1,525 expression quantitative trait loci (SV-eQTLs) as well as SV candidates for adaptive selection within the human population.

genetics↗

bioRxiv2020

Cell-type, single-cell, and spatial signatures of brain-region specific splicing in postnatal development

Alternative RNA splicing varies across brain regions, but the single-cell resolution of such regional variation is unknown. Here we present the first single-cell investigation of differential isoform expression (DIE) between brain regions, by performing single cell long-read transcriptome sequencing in the mouse hippocampus and prefrontal cortex in 45 cell types at postnatal day 7 (www.isoformAtlas.com). Using isoform tests for brain-region specific DIE, which outperform exon-based tests, we detect hundreds of brain-region specific DIE events traceable to specific cell-types. Many DIE events correspond to functionally distinct protein isoforms, some with just a 6-nucleotide exon variant. In most instances, one cell type is responsible for brain-region specific DIE. Cell types indigenous to only one anatomic structure display distinctive DIE, where for example, the choroid plexus epithelium manifest unique transcription start sites. However, for some genes, multiple cell-types are responsible for DIE in bulk data, indicating that regional identity can, although less frequently, override cell-type specificity. We validated our findings with spatial transcriptomics and long-read sequencing, yielding the first spatially resolved splicing map in the postnatal mouse brain (www.isoformAtlas.com). Our methods are highly generalizable. They provide a robust means of quantifying isoform expression with cell-type and spatial resolution, and reveal how the brain integrates molecular and cellular complexity to serve function.

neuroscience↗