Biology subjects

Marschall, T.

Publications and source records attributed to Marschall, T..

At least 19 recordsLinked to original sources

bioRxiv2026

Scalable and rare-variant aware genome inference across the 1kGP cohort

Pangenome graphs built from haplotype-resolved de novo assemblies enable accurate analysis of genetic variation. The short-read-based tool PanGenie efficiently genotypes variants discovered in a pangenome across large cohorts and outper-forms linear reference-based methods for structural variants (SVs). However, it cannot detect novel variants absent from the graph, missing many rare SVs (allele frequency < 1%) and was limited to graphs with 254 haplotypes. First, we introduce a haplotype sampling step that reduces the number of haplotypes using sample-specific k-mers before genotyping, decreasing runtime twelvefold and memory usage 1.4-fold at 30x coverage. Second, we present a polishing work-flow that corrects residual errors in haplotypes inferred from PanGenie genotypes and incorporates rare and private mutations. We genotype 3,202 samples from the 1000 Genomes Project and use low-coverage ONT data (967 samples) for polishing. We achieve a median QV of 46 and provide the 1,934 polished haplotype sequences as a community resource.

bioinformatics↗

bioRxiv2026

Pangenome-based human genome analysis improves trait association and genomic prediction

The Human Pangenome Reference Consortium has generated 462 open-access reference genomes and a variation graph that represents differences among them, providing a substrate for pangenome-based analysis methods that overcome the longstanding limitation of comparing all genomic data to a single linear reference. A key unresolved question is the extent to which these approaches can improve trait mapping. We investigate this using the genetics of gene expression variation as a model. We developed a graph-based method (EdgeDepth) for associating sequence variation with traits using short-read genome sequencing data, and show that it captures complex forms of genetic variation missed by other methods. We evaluated trait mapping performance using 430 samples with deep RNA-seq data, and found that pangenomic methods enable the detection of expression quantitative trait loci involving multiallelic indels and structural variants, leading to increased power at a subset of genes. These include 812 genes (7.9% of total) with [≥]20% improvement in statistical significance relative to the 1000 Genomes Project callset, and 185 (1.8%) with a 50% improvement, 10 of which are candidates to explain prior GWAS results. Notably, these analyses implicate GBAP1 pseudogene copy number as a causal factor in Crohns disease, likely via miRNA-mediated regulation of GBA1, which explains prior GWAS results based on flanking SNPs. The inclusion of pangenome-specific variation also improved the performance of gene expression prediction models, with median variance explained increasing from 10.1% to 12.5%, and 14.6% of genes showing significant improvement ({Delta}r2>0.05). Taken together, these results suggest that integration of pangenomic methods into human genetic studies will improve trait association and genomic prediction at a meaningful subset of genes.

genetics↗

bioRxiv2025

Sequence-to-graph alignment based copy number calling using a network flow formulation

Variation of sequence copy number (CN) between individuals can be associated with phenotypical differences. Consequently, CN calling is an important step for disease association and identification, as well as for genome assembly validation. Traditionally, CN calling is done by mapping sequencing reads to a linear reference genome and estimating the CN from the observed read depth. This approach, however, is significantly hampered by sequences and rearrangements not present in a linear reference genome; at the same time simple CN prediction for individual graph nodes does not make use of the graph topology and can lead to inconsistent results. To address these issues, we propose Floco, a method for CN calling with respect to a genome graph using a network flow formulation. Given a graph and alignments against that graph, we calculate raw CN probabilities for every graph node based on the Negative Binomial distribution and the base pair coverage across the node, and then use integer linear programming to compute the CN flow through the whole graph. We tested this approach on 15 aligned datasets, involving three different graphs, as well as HiFi and ONT sequencing reads and linear assemblies split into reads. These results demonstrate that the addition of the network flow formulation increases the accuracy of CN predictions by up to 43% when compared with read depth based estimation alone. Additionally, we observed that concordance between predictions from the three different sequence sources was able to reach 93.2%. Floco fills a gap in CN calling tools specifically designed for genome graphs.

bioinformatics↗

bioRxiv2025

DiVerG: Scalable Distance Index for Validation of Paired-End Alignments in Sequence Graphs

Determining the distance between two loci within a genomic region is a recurrent operation in various tasks in computational genomics. A notable example of this task arises in paired-end read mapping as a form of validation of distances between multiple alignments. While straightforward for a single genome, graph-based reference structures render the operation considerably more involved. Given the sheer number of such queries in a typical read mapping experiment, an efficient algorithm for answering distance queries is crucial. In this paper, we introduce DiVerG, a compact data structure as well as a fast and scalable algorithm, for constructing distance indexes for general sequence graphs on multi-core CPU and many-core GPU architectures. DiVerG is based on PairG [28], but overcomes the limitations of PairG by exploiting the extensive potential for improvements in terms of scalability and space efficiency. As a consequence, DiVerG can process substantially larger datasets, such as whole human genomes, which are unmanageable by PairG. DiVerG offers faster index construction time and consistently faster query time with gains proportional to the size of the underlying compact data structure. We demonstrate that our method performs favorably on multiple real datasets at various scales. DiVerG achieves superior performance over PairG; e.g. resulting to 2.5-4x speed-up in query time, 44-340x smaller index size, and 3-50x faster construction time for the genome graph of the MHC region, as a particularly variable region of the human genome. The implementation is available at: https://github.com/cartoonist/diverg

bioinformatics↗

bioRxiv2024

EpiPAMPAS: Rapid detection of intra-protein epistasis via parsimonious ancestral state reconstruction and counting mutations

MotivationAn epistatic interaction is a non-linear combination of effects of individual mutations on fitness. This type of interaction is a known driver for evolution, as they alter the organisms fitness and adaptability. In this work we introduce EpiPAMPAS, a statistical method that is based on multiple sequence alignments (MSA) and detecting mutations in the same direction on a dendrogram instead of a phylogenetic tree using the Sankoff algorithm. ResultsWe tested EpiPAMPAS on both simulated and real sequencing data. On the simulated data, our method was able to detect the simulated epistatic pairs with very low p-value. In a real-world application, we tested the influenza proteins N1, N2, H1, H3 and HIV-1 envelope protein subtypes A, B and C. We observe that EpiPAMPAS detects fewer interacting pairs than comparable statistical approaches, although the overlap between detected positions is good. Moreover, some of the amino acids from the detected pairs are known to be deleterious for viral fitness. AvailabilityEpiPAMPAS is available under MIT license at https://github.com/kalininalab/EpiPAMPAS

bioinformatics↗

bioRxiv2024

gaftools: a toolkit for analyzing and manipulating pangenome alignments

Linear reference genomes are ubiquitously used in genomics research, despite known biases associated with their use. In recent years, there has a been a shift towards using graph-based reference genomes to address some of these biases, which has required new algorithms and file formats to be developed. This has created the need to develop new tools capable of utilizing these formats and performing operations similar to those carried out by traditional methods.In this paper we present "gaftools", a multi-purpose tool that introduces several utilities for processing graph alignments in GAF format. Gaftools enables users to index and sort alignments, with graph ordering serving as a necessary step for the sorting process. Additionally it allows users to view subsets of alignments and perform realignment using the wavefront alignment algorithm, among other useful features. Many of these functionalities are inspired by SAMtools, which provide similar operations for linear genomes, while gaftools adapts and extends them for pangenomes. Availabilitygaftools is available under MIT license at https://github.com/marschall-lab/gaftools.

bioinformatics↗

bioRxiv2024

Panacus: fast and exact pangenome growth and core size estimation

MotivationUsing a single linear reference genome poses a limitation to exploring the full genomic diversity of a species. The release of a draft human pangenome underscores the increasing relevance of pangenomics to overcome these limitations. Pangenomes are commonly represented as graphs, which can represent billions of base pairs of sequence. Presently, there is a lack of scalable software able to perform key tasks on pangenomes, such as quantifying universally shared sequence across genomes (the core genome) and measuring the extent of genomic variability as a function of sample size (pangenome growth). ResultsWe introduce Panacus (pangenome-abacus), a tool designed to rapidly perform these tasks and visualize the results in interactive plots. Panacus can process GFA files, the accepted standard for pangenome graphs, and is able to analyze a human pangenome graph with 110 million nodes in less than one hour. AvailabilityPanacus is implemented in Rust and is published as Open Source software under the MIT license. The source code and documentation are available at https://github.com/marschall-lab/panacus. Panacus can be installed via Bioconda at https://bioconda.github.io/recipes/panacus/README.html. ContactLuca Parmigiani (luca.parmigiani@uni-bielefeld.de), Daniel Doerr (daniel.doerr@hhu.de).

bioinformatics↗

bioRxiv2024

Locityper: targeted genotyping of complex polymorphic genes

The human genome contains numerous structurally-variable polymorphic loci, including several hundred disease-associated genes, almost inaccessible for accurate variant calling. Here we present Locityper, a tool capable of genotyping such challenging genes using short and long-read whole genome sequencing. For each target, Locityper recruits and aligns reads to locus haplotypes, for instance extracted from a pangenome, and finds the likeliest haplotype pair by optimizing read alignment, insert size and read depth profiles. Locityper accurately genotypes up to 194 of 256 challenging medically relevant loci (95% haplotypes at QV33), an 8.8-fold gain compared to 22 genes achieved with standard variant calling pipelines. Furthermore, Locityper provides access to hyperpolymorphic HLA genes and other gene families, including KIR, MUC and FCGR. With its low running time of 1h10m per sample at 8 threads, Locityper is scalable to biobank-sized cohorts, enabling association studies for previously intractable disease-relevant genes.

bioinformatics↗

bioRxiv2024

Long-read sequencing and structural variant characterization in 1,019 samples from the 1000 Genomes Project

Structural variants (SVs) contribute significantly to human genetic diversity and disease1-4. Previously, SVs have remained incompletely resolved by population genomics, with short-read sequencing facing limitations in capturing the whole spectrum of SVs at nucleotide resolution5-7. Here we leveraged nanopore sequencing8 to construct an intermediate coverage resource of 1,019 long-read genomes sampled within 26 human populations from the 1000 Genomes Project. By integrating linear and graph-based approaches for SV analysis via pangenome graph-augmentation, we uncover 167,291 sequence-resolved SVs in these samples, considerably advancing SV characterization compared to population-wide short-read sequencing studies3,4. Our analysis details diverse SV classes--deletions, duplications, insertions, and inversions--at population-scale. LINE-1 and SVA retrotransposition activities frequently mediate transductions9,10 of unique sequences, with both mobile element classes transducing sequences at either the 3'- or 5'-end, depending on the source element locus. Furthermore, analyses of SV breakpoint junctions suggest a continuum of homology-mediated rearrangement processes are integral to SV formation, and highlight evidence for SV recurrence involving repeat sequences. Our open-access dataset underscores the transformative impact of long-read sequencing in advancing the characterisation of polymorphic genomic architectures, and provides a resource for guiding variant prioritisation in future long-read sequencing-based disease studies.

genomics↗

bioRxiv2024

Structural and genetic diversity in the secreted mucins, MUC5AC and MUC5B

The secreted mucins MUC5AC and MUC5B play critical defensive roles in airway pathogen entrapment and mucociliary clearance by encoding large glycoproteins with variable number tandem repeats (VNTRs). These polymorphic and degenerate protein coding VNTRs make the loci difficult to investigate with short reads. We characterize the structural diversity of MUC5AC and MUC5B by long-read sequencing and assembly of 206 human and 20 nonhuman primate (NHP) haplotypes. We find that human MUC5B is largely invariant (5761-5762aa); however, seven haplotypes have expanded VNTRs (6291-7019aa). In contrast, 30 allelic variants of MUC5AC encode 16 distinct proteins (5249-6325aa) with cysteine-rich domain and VNTR copy number variation. We grouped MUC5AC alleles into three phylogenetic clades: H1 (46%, [~]5654aa), H2 (33%, [~]5742aa), and H3 (7%, [~]6325aa). The two most common human MUC5AC variants are smaller than NHP gene models, suggesting a reduction in protein length during recent human evolution. Linkage disequilibrium (LD) and Tajimas D analyses reveal that East Asians carry exceptionally large MUC5AC LD blocks with an excess of rare variation (p<0.05). To validate this result, we used Locityper for genotyping MUC5AC haplogroups in 2,600 unrelated samples from the 1000 Genomes Project. We observed signatures of positive selection in H1 and H2 among East Asians and a depletion of the likely ancestral haplogroup (H3). In Africans and Europeans, H3 alleles show an excess of common variation and deviate from Hardy-Weinberg equilibrium, consistent with heterozygote advantage and balancing selection. This study provides a generalizable strategy to characterize complex protein coding VNTRs for improved disease associations.

genomics↗

bioRxiv2024

SVarp: pangenome-based structural variant discovery

The linear human reference genome that we use today does not represent the haplotypic diversity of the global human population. This raises bias in genomic read alignment and limits our ability to call large structural variations (SV), especially at highly polymorphic loci. Thus, many SV alleles remain unresolved. Recent efforts to transition to a graph-based reference genome resulted in the generation of the first draft human pangenome reference, but tools to call SVs relative to the pangenome reference are presently lacking. In this study, we present the SVarp algorithm, aiming to discover haplotype resolved SVs on top of a pangenome reference using long sequencing reads. SVarp outputs local assemblies of SV alleles, termed svtigs, instead of a VCF file of SV breakpoints, which we propose as a general exchange format allowing for flexible downstream analyses. In order to assess the accuracy of svtigs, we used simulated and real human genomes. Simulations allowed us to make exact breakpoint comparisons against the true callsets. We observed [~]96% recall with deletions, insertions and duplications larger than 1,000bp, showing that SVarp can reliably detect genomic structural variants not yet represented in the graph. On the other hand, we compared SVarp output for ONT sequencing data at 20X coverage against independent genome assemblies of the same samples and found that [~]82% of our svtig predictions are validated by the assemblies by a match with more than 85% sequence identity. SVarp was implemented using C++ and its source code is available at https://github.com/asylvz/SVarp under MIT license.

bioinformatics↗

bioRxiv2024

Phasing Diploid Genome Assembly Graphs with Single-Cell Strand Sequencing

Haplotype information is crucial for biomedical and population genetics research. However, current strategies to produce de-novo haplotype-resolved assemblies often require either difficult-to-acquire parental data or an intermediate haplotype-collapsed assembly. Here, we present Graphasing, a workflow which synthesizes the global phase signal of Strand-seq with assembly graph topology to produce chromosome-scale de-novo haplotypes for diploid genomes. Graphasing readily integrates with any assembly workflow that both outputs an assembly graph and has a haplotype assembly mode. Graphasing performs comparably to trio-phasing in contiguity, phasing accuracy, and assembly quality, outperforms Hi-C in phasing accuracy, and generates human assemblies with over 18 chromosome-spanning haplotypes.

bioinformatics↗

bioRxiv2023

A comprehensive catalog of 3D genome organization in diverse human genomes facilitates understanding of the impact of structural variation on chromatin structure

The human genome is packaged into the three-dimensional (3D) nucleus and organized into functional units known as topologically associating domains (TADs) and chromatin loops. Recent studies show that the 3D genome can be modified by genome structural variants (SVs) through disrupting higher-order chromatin organizations such as TADs, which play an essential role in insulating genes from aberrant regulation by regulatory elements outside TADs. Here, we have developed an integrative Hi-C analysis pipeline to generate a comprehensive catalog of TADs, TAD boundaries, and loops in human genomes to fill the gap of limited resources. We identified 2,293 TADs and 6,810 sub-TADs missing in the previously released TADs of GM12878. We then quantified the impact of SVs overlapping with TAD boundaries and observed that two SVs could significantly alter chromatin architecture leading to abnormal expression and splicing of genes associated with human diseases.

genomics↗

bioRxiv2023

PanPA: generation and alignment of panproteome graphs

MotivationCompared to eukaryotes, prokaryote genomes are more diverse through different mechanisms, including a higher mutation rate and horizontal gene transfer. Therefore, using a linear representative reference can cause a reference bias. Graph-based pangenome methods have been developed to tackle this problem. However, comparisons in DNA space is still challenging due to this high diversity. In contrast, amino acids have higher similarity due to evolutionary constraints, resulting in conserved amino acids that, however, may be encoded by several synonymous codons. Coding regions cover the majority of the genome in prokaryotes. Thus, building panproteomes leverages the high sequence similarity while not losing much of the genome in non-coding regions. ResultsWe present PanPA, a method that takes a set of multiple sequence alignments (MSAs) of proteins or protein clusters, indexes them, and builds a graph for each MSA. In the querying step, it can align DNA or amino acid sequences back to these graphs. We first showcase that PanPA generates correct alignments on a panproteome from 1,350 E. coli. To demonstrate that panproteomes allow longer phylogenetic distance comparison, we compare DNA and protein alignments from 1,073 S. enterica assemblies against E. coli reference genome, pangenome, and panproteome using BWA, GraphAligner, and PanPA respectively, where PanPA was able to produce around 22% more alignments. We also aligned DNA short-reads WGS sample from S. enterica against the E. coli reference with BWA and the panproteome with PanPA, where PanPA was able to find alignment for 69% of the reads compared to 5% with BWA AvailabilityPanPA is available at https://github.com/fawaz-dabbaghieh/PanPA Contactfawaz@hhu.de, olga.kalinina@helmholtz-hzi.de Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2022

Assembly of 43 diverse human Y chromosomes reveals extensive complexity and variation

The prevalence of highly repetitive sequences within the human Y chromosome has led to its incomplete assembly and systematic omission from genomic analyses. Here, we present long-read de novo assemblies of 43 diverse Y chromosomes spanning 180,000 years of human evolution, including two from deep-rooted African Y lineages, and report remarkable complexity and diversity in chromosome size and structure, in contrast with its low level of base substitution variation. The size of the Y chromosome assemblies varies extensively from 45.2 to 84.9 Mbp and include, on average, 81 kbp of novel sequence per Y chromosome. Half of the male-specific euchromatic region is subject to large inversions with a >2-fold higher recurrence rate compared to inversions in the rest of the human genome. Ampliconic sequences associated with these inversions further show differing mutation rates that are sequence context-dependent and some ampliconic genes show evidence for concerted evolution with the acquisition and purging of lineage-specific pseudogenes. The largest heterochromatic region in the human genome, the Yq12, is composed of alternating arrays of DYZ1 and DYZ2 repeat units that show extensive variation in the number, size and distribution of these arrays, but retain a 1:1 copy number ratio of the monomer repeats, consistent with the notion that functional or evolutionary forces are acting on this chromosomal region. Finally, our data suggests that the boundary between the recombining pseudoautosomal region 1 and the non-recombining portions of the X and Y chromosomes lies 500 kbp distal to the currently established boundary. The availability of sequence-resolved Y chromosomes from multiple individuals provides a unique opportunity for identifying new associations of specific traits with Y-chromosomal variants and garnering novel insights into the evolution and function of complex regions of the human genome.

genomics↗

bioRxiv2022

Inversion polymorphism in a complete human genome assembly

The completion of the human genome significantly improved our ability to discover and interpret genome copy number variation. In order to understand its impact on the characterization of inversion polymorphisms, we remapped data from 41 human genomes and 10 new samples against the telomere-to-telomere (T2T) reference genome as compared to the standard GRCh38 reference. Our analysis shows a ~21% increase in sensitivity identifying and improving mapping of 63 inversions. We further identify 26 misorientations within GRCh38, and show that the T2T reference is three times more likely to represent the correct orientation of the major human allele. As a result, we report a significant bias for inversions accumulating within the pericentromeric regions of specific chromosomes and show that functional annotations around inverted regions, such as topological-associated domains, can be better interpreted.

genomics↗

bioRxiv2022

Gaps and complex structurally variant loci in phased genome assemblies

There has been tremendous progress in the production of phased genome assemblies by combining long-read data with parental information or linking read data. Nevertheless, a typical phased genome assembly generated by trio-hifiasm still generates more than ~140 gaps. We perform a detailed analysis of gaps, assembly breaks, and misorientations from 77 phased and assembled human genomes (154 unique haplotypes). We find that trio-based approaches using HiFi are the current gold standard although chromosome-wide phasing accuracy is comparable when using Strand-seq instead of parental data. We find two-thirds of defined contig ends cluster near the largest and most identical repeats [including segmental duplications (35.4%) or satellite DNA (22.3%) or to regions enriched in GA/AT rich DNA (27.4%)]. As a result, 1513 protein-coding genes overlap assembly gaps in at least one haplotype and 231 are recurrently disrupted or missing from five or more haplotypes. In addition, we estimate that 6-7 Mbp of DNA are incorrectly orientated per haplotype irrespective of whether trio-free or trio-based approaches are employed. 81% of such misorientations correspond to bona fide large inversion polymorphisms in the human species, most of which are flanked by large identical segmental duplications. In addition, we also identify large-scale alignment discontinuities consistent with an 11.9 Mbp deletion and 161.4 Mbp of insertion per human haploid genome. While 99% of this variation corresponds to satellite DNA, we identify 230 regions of the euchromatic DNA with frequent expansions and contractions, nearly half of which overlap with 197 protein-coding genes. Although not completely resolved, these regions include copy number polymorphic and biomedically relevant genic regions where complete resolution and a pangenome representation will be most useful, yet most challenging, to realize.

genomics↗

bioRxiv2022

Constructing founder sets under allelic and non-allelic homologous recombination

Homologous recombination between the maternal and paternal copies of a chromosome is a key mechanism for human inheritance and shapes population genetic properties of our species. However, a similar mechanism can also act between different copies of the same sequence, then called non-allelic homologous recombination (NAHR). This process can result in genomic rearrangements--including deletion, duplication, and inversion--and is underlying many genomic disorders. Despite its importance for genome evolution and disease, there is a lack of computational models to study genomic loci prone to NAHR. In this work, we propose such a computational model, providing a unified framework for both (allelic) homologous recombination and NAHR. Our model represents a set of genomes as a graph, where human haplotypes correspond to walks through this graph. We formulate two founder set problems under our recombination model, provide flow-based algorithms for their solution, and demonstrate scalability to problem instances arising in practice.

bioinformatics↗