Biology subjects

Rasmussen, S.

Publications and source records attributed to Rasmussen, S..

At least 19 recordsLinked to original sources

bioRxiv2026

multiomeR: a pipeline for reproducible analysis of large-scale single-nucleus Epi Multiome datasets

The 10x Genomics Epi Multiome assay measures gene expression and chromatin accessibility in the same nuclei, and studies increasingly profile many samples. Yet analysis after Cell Ranger ARC is usually assembled by hand from separate tools, which makes it slow to revise, hard to reproduce and memory-intensive for large studies. multiomeR addresses these problems with a configurable R pipeline for quality control, processing and integration of both data types, and cell-type annotation across many samples. Plain-text configuration files record every analytical choice, each step is tracked so that a change reruns only what it affects, and data stay on disk rather than in memory. Optional modules compare groups of donors, link accessible regions to the genes they may regulate, and test which cell types are enriched for genetic associations with a trait. In a benchmark on 31 public samples (375,757 nuclei), multiomeR had a shorter estimated run time than a Seurat v5.6/Signac v2 workflow (2.2 versus 5.6 h), because it runs independent steps in parallel. It also used fewer memory gigabyte-hours (87 versus 114 GB-hours), with similar total CPU time. multiomeR thus makes large multi-sample multiome studies practical to analyze, revise and extend.

bioinformatics↗

bioRxiv2026

SLIM: A small linear model with STRING embeddings for single-cell genetic perturbation prediction

Predicting cellular responses to genetic perturbations is central to understanding gene function and prioritizing therapeutic targets, but experimental screens cannot exhaustively cover genes, cell types, and perturbation combinations. Recent benchmarks have shown that simple baselines can match or outperform substantially more complex models, suggesting that informative biological priors may be as important as model capacity. Here we present SLIM, a lightweight extension of the bilinear model of Ahlmann-Eltze et al. SLIM represents perturbations with 64-dimensional embeddings derived from the STRING protein network and predicts mean transcriptional responses through a closed-form ridge-regression estimator. It then constructs single-cell populations by retrieving training cells and rescaling each gene to match the predicted mean. We evaluated SLIM against four deep learning models and two simple baselines on four single-gene perturbation datasets and one combinatorial perturbation dataset. Across these within-dataset benchmarks, SLIM achieved competitive mean-response accuracy, ranked first in eight of twelve single-gene dataset-metric comparisons, and produced substantially lower maximum mean discrepancy values than the evaluated alternatives. The model has 640 trainable parameters and fitted each benchmark dataset in under 10 seconds on a CPU. These results show that compact biological representations can support accurate and computationally efficient perturbation prediction. Code is available at https://github.com/RasmussenLab/SLIM. Key PointsO_LISLIM combines a closed-form bilinear predictor with STRING-derived perturbation embeddings. C_LIO_LIAcross five within-dataset benchmarks, SLIM achieved competitive mean-response prediction with only 640 trainable parameters. C_LIO_LISLIM builds cell populations by rescaling retrieved training cells to the predicted mean, so they inherit realistic cell-to-cell variation and gene-gene covariation. C_LIO_LIThe results highlight the importance of perturbation representations and population-construction procedures in low-data benchmarks. C_LIO_LISLIM fits each benchmark dataset in under 10 seconds on a standard CPU. C_LI

bioinformatics↗

bioRxiv2025

Molecular maps of diseases from omics data and network embeddings

Identifying disease-relevant proteins and pathways remains a fundamental challenge in understanding disease mechanisms and supporting therapeutic development. While omics analyses can provide valuable insights, they typically consider each gene/protein separately rather than at the level of biological systems. This can be addressed by combining the omics data with protein networks. We integrate disease-specific omics data with a universal functional association network from STRING, which we represent using node2vec embedding. This way, we constructed disease maps for seven diseases spanning inflammatory, oncological, neurological, and vascular diseases based on genetics, transcriptomics, somatic mutation, and proteomics data. Compared to omics analysis alone, the use of a simple linear model on top of network embedding enabled us to identify 2-4 times as many known disease-relevant proteins at the same specificity. Clustering of the resulting disease maps revealed both functional modules shared by many diseases, such as inflammatory pathways and cancer hallmarks, and disease-specific modules, such as keratinization in atopic dermatitis and extracellular matrix remodeling in aortic aneurysm. Together, these results highlight the value of protein network embedding when analyzing omics data to understand diseases.

bioinformatics↗

bioRxiv2025

Unsupervised learning of multi-omics data enables disease risk prediction in the UK Biobank

The size and complexity of biomedical datasets continue to grow, driving the development of methods that reduce dimensionality while preserving biological signals. Yet, when deep learning is applied to such data, the impact of preprocessing choices and dataset properties on model behavior is often overlooked. Here, we applied our framework Multi-Omics Variational autoEncoder (MOVE) to multiomics data from 452,026 UK Biobank participants, aiming to both evaluate the power of the learned representations for disease risk prediction and critically analyze how non-biological factors, like dataset properties and preprocessing decisions, can shape and influence the results. We show that reducing the dimensionality of the data by a factor of 80 still yields comparable prediction performance across 15 different diseases. We further demonstrate how dataset properties and preprocessing choices impact the model performance, latent representation and downstream results, and our findings strongly underline the need for thorough analysis and understanding of a models behavior before drawing conclusions from its results.

bioinformatics↗

bioRxiv2025

On the use of variational autoencoders for biomedical data integration

Variational Autoencoders (VAEs) are a widely used framework to integrate diverse biomedical data modalities, create representations that capture the underlying structure of the datasets, and obtain insights about the relations between variables. Here we describe how this is achieved from an empirical point of view in our previously developed VAE-based framework MOVE, providing an intuitive perspective on the inner workings of multimodal VAEs in biomedical contexts. We explore how the models emerging dynamics shape their performance and how in silico perturbations can be leveraged to identify potential associations between variables. To do that, we extend our framework to handle perturbations of continuous variables, introduce a new approach to better capture associations between them, and create synthetic datasets to benchmark the proposed methods against well-defined ground truth associations. We finally showcase our findings in real biomedical scenarios, namely a multimodal dataset of inflammatory bowel disease and a dataset containing genetic knockdowns in K562 and RPE1 cells.

bioinformatics↗

bioRxiv2025

Body Fluid Proteomic Landscape of Acute Exercise

Physical activity improves health, yet the molecular mechanisms remain partially understood. This study presents a high-resolution, time-resolved atlas profiling 10,127 proteins across plasma, saliva, and urine from healthy adults post-acute exercise. Exercise regulated over 3,000 proteins, revealing distinct, fluid-specific temporal dynamics. By integrating fluid-specific exercise signatures with tissue and disease atlases, we delineated the contribution of tissues and associations to various diseases. Network analysis across body fluids elucidated coordinated remodeling in the extracellular matrix and immune activation orchestrating exercise-induced networks. Many exercise-responsive plasma proteins were robust across age, sex, and exercise modalities, indicating a conserved systemic signature. Integration with genetic data established exercise-regulated proteins as modulators of metabolic traits and identified over 200 targeted by approved drugs, highlighting their impact on disease-relevant pathways. This comprehensive atlas, available as an open-access resource https://cbmr.ku.dk/research/research-groups/deshmukh-group/shiny-apps/, advances our molecular insight into exercise adaptations and enables exerkine discovery, biomarker development, and pharmacological exercise-mimetic strategies. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/656705v1_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@bf3baforg.highwire.dtl.DTLVardef@e79b11org.highwire.dtl.DTLVardef@1757b9borg.highwire.dtl.DTLVardef@5fddc_HPS_FORMAT_FIGEXP M_FIG C_FIG HighlightsO_LIExercise induces robust and distinct changes across body fluid proteomes C_LIO_LITissue remodeling and immune activation drive exercise-induced network expansion C_LIO_LI[~]1,000 exercise-regulated plasma proteins are age, exercise mode, or sex-specific C_LIO_LIGenetic inference identifies druggable exerkines that regulate health and disease C_LI

physiology↗

bioRxiv2025

Accurate plasmid reconstruction from metagenomics data using assembly-alignment graphs and contrastive learning

Plasmids are extrachromosomal DNA molecules that enable horizontal gene transfer in bacteria, often conferring advantages such as antibiotic resistance. Despite their significance, plasmids are underrepresented in genomic databases due to challenges in assembling them, caused by mosaicism and micro-diversity. Current plasmid assemblers rely on detecting circular paths in single-sample assembly graphs, but face limitations due to graph fragmentation and entanglement, and low coverage. We introduce PlasMAAG (Plasmid and organism Metagenomic binning using Assembly Alignment Graphs), a framework to recover plasmids and organisms from metagenomic samples that leverages an approach that we call "assembly-alignment graphs" alongside common binning features. On synthetic benchmark datasets, PlasMAAG reconstructed 50-121% more near-complete plasmids than competing methods and improved the Matthews Correlation Coefficient of geNomad contig classification by 28-106%. On hospital sewage samples, PlasMAAG outperformed all other methods, reconstructing 33% more plasmid sequences. PlasMAAG enables the study of organism-plasmid associations and intra-plasmid diversity across samples, offering state-of-the-art plasmid reconstruction with reduced computational costs.

bioinformatics↗

bioRxiv2024

Binning meets taxonomy: TaxVAMB improves metagenome binning using bi-modal variational autoencoder

A common procedure for studying the microbiome is binning the sequenced contigs into metagenome-assembled genomes. Currently, unsupervised and self-supervised deep learning based methods using co-abundance and sequence based motifs such as tetranucleotide frequencies are state-of-the-art for metagenome binning. Taxonomic labels derived from alignment based classification have not been widely used. Here, we propose TaxVAMB, a metagenome binning tool based on semi-supervised bi-modal variational autoencoders, combining tetranucleotide frequencies and contig co-abundances with contig annotations returned by any taxonomic classifier on any taxonomic rank. TaxVAMB outperforms all other binners on CAMI2 human microbiome datasets, returning on average 40% more near-complete assemblies than the next best binner. On real long-read datasets TaxVAMB recovers on average 13% more near-complete bins and 14% more species. When used in a single-sample setup, TaxVAMB on average returns 83% more high quality bins than VAMB. TaxVAMB bins incomplete genomes drastically better than any other tool, returning 255% more high quality bins of incomplete genomes than the next best binner. Our method has immediate research and industrial applications, as well as methodological novelty which can be translated to other biological problems with semi-supervised multimodal datasets.

bioinformatics↗

bioRxiv2024

Modeling nascent transcription from chromatin landscape and structure

BackgroundDifferent cell types and their associated functionalities emerge from a single genomic sequence when certain regions are expressed while others remain silenced. Modeling gene expression and its potential malfunctioning in different cellular contexts is hence pivotal to understand both development and disease. ResultsWe present the Chromatin Landscape and Structure to Expression Regressor (CLASTER), an epigenetic-based deep neural network that can integrate different data modalities describing the chromatin landscape and its 3D structure. CLASTER effectively translates them into nascent transcription levels measured by EU-seq at a kilobasepair resolution. Our predictions reached a Pearson correlation with targets above r=0.86 at both bin and gene levels, without relying on DNA sequence nor explicitly extracted chromatin features. The model mostly used the information found within 10 kbp of the predicted locus, even when a wide genomic region of 1 Mbp was available. Explicit modeling of long-range interactions using multi-headed attention and high-resolution chromatin contact maps had little impact on model performance, despite the model correctly identifying elements in these inputs influencing nascent transcription. The trained model served then as a platform to predict the transcriptional impact of simulated epigenetic silencing perturbations. ConclusionsOur results point towards a rather local, integrative and combinatorial paradigm of gene regulation, where changes in the chromatin environment surrounding a gene shape its context-specific transcription. We conclude that the predominant locality and limitations of current machine learning approaches might emerge as a genuine signature of genomic organization, having broad implications for future modeling approaches.

bioinformatics↗

bioRxiv2024

BinBencher: Fast, flexible and meaningful benchmarking suite for metagenomic binning

New methods for metagenomic binning are typically evaluated using benchmarking software, and become tuned to maximize whatever criterion is measured by the benchmark. Subtleties in benchmarking procedures can cause misleading evaluations, derailing method development. Differences between procedures used to evaluate binning tools make them hard to compare, which slows progress in the field. We introduce BinBencher, a free software suite for benchmarking, and show how BinBencher produces evaluations that are more biologically meaningful than alternative benchmarking approaches.

bioinformatics↗

bioRxiv2023

Taxometer: Improving taxonomic classification of metagenomics contigs

For taxonomy based classification of metagenomics assembled contigs, current methods use sequence similarity to identify their most likely taxonomy. However, in the related field of metagenomics binning contigs are routinely clustered using information from both the contig sequences and their abundance. We introduce Taxometer, a neural network based method that improves the annotations and estimates the quality of any taxonomic classifier by combining contig abundance profiles and tetra-nucleotide frequencies. When applied to five short-read CAMI2 datasets, it increased the average share of correct species-level contig annotations of the MMSeqs2 tool from 66.6% to 86.2% and reduced the share of wrong species-level annotations in the CAMI2 Rhizosphere dataset two-fold on average for Metabuli, Centrifuge, and Kraken2. Finally, we applied Taxometer to two complex long-read metagenomics data sets for benchmarking taxonomic classifiers. Taxometer is available as open-source software and can enhance any taxonomic annotation of metagenomic contigs.

bioinformatics↗

bioRxiv2023

N-terminal toxin signal peptides efficiently load therapeutics into a natural nano-injection system

Targeted delivery of therapeutics to specific cells is a major bottleneck towards personalized medicine. The extracellular injection system (eCIS) of Serratia entomophila, the antifeeding prophage (Afp), promises potential for drug delivery purposes. However, the precise mechanism of action, toxin location, and Afp loading remain unclear. Here, we reveal a minimal N-terminal signal peptide (NtSP) of the toxin Afp18, that plays a key role in toxin packing. By engineering fusion proteins, we demonstrate that Afp18s NtSP can shuttle effectors for Afp loading. We packed non-eCIS effectors, including CRISPR-Cas protein Cas{Phi}-2 from Biggiephage, and a human antimicrobial peptide, LL37, into Afp. Additionally, NtSPs from eCIS effectors of other species facilitate loading of Cas{Phi}-2 into Afp. We observed cargo being packed inside the Afp tail tube through cryo-EM single particle analysis. The presented results enhance our understanding of eCIS toxin packing and contribute to their development as targeted delivery systems. TeaserA novel use of the Afp nano injection systems N-terminal signal peptide in targeted therapeutics delivery

molecular biology↗

bioRxiv2023

Adversarial and variational autoencoders improve metagenomic binning

Assembly of reads from metagenomic samples is a hard problem, often resulting in highly fragmented genome assemblies. Metagenomic binning allows us to reconstruct genomes by regrouping the sequences by their organism of origin, thus representing a crucial processing step when exploring the biological diversity of metagenomic samples. Here we present Adversarial Autoencoders for Metagenomics Binning (AAMB), an ensemble deep learning approach that integrates sequence co-abundances and tetranucleotide frequencies into a common denoised space that enables precise clustering of sequences into microbial genomes. When benchmarked, AAMB presented similar or better results compared with the state-of-the-art reference-free binner VAMB, reconstructing [~]7% more near-complete (NC) genomes across simulated and real data. In addition, genomes reconstructed using AAMB had higher completeness and greater taxonomic diversity compared with VAMB. Finally, we implemented a pipeline integrating VAMB and AAMB that enabled improved binning, recovering 20% and 29% more simulated and real NC genomes, respectively, compared to VAMB with moderate additional runtime. AAMB is freely available at https://github.com/RasmussenLab/VAMB.

bioinformatics↗

bioRxiv2023

Mass spectrometry-based proteomics imputation using self supervised deep learning

Imputation techniques provide means to replace missing measurements with a value and are used in almost all downstream analysis of mass spectrometry (MS) based proteomics data using label-free quantification (LFQ). Here we demonstrate how collaborative filtering, denoising autoencoders, and variational autoencoders can impute missing values in the context of LFQ at different levels. We applied our method, proteomics imputation modeling mass spectrometry (PIMMS), to an alcohol-related liver disease (ALD) cohort with blood plasma proteomics data available for 358 individuals. Removing 20 percent of the intensities we were able to recover 15 out of 17 significant abundant protein groups using PIMMS-VAE imputations. When analyzing the full dataset we identified 30 additional proteins (+13.2%) that were significantly differentially abundant across disease stages compared to no imputation and found that some of these were predictive of ALD progression in machine learning models. We, therefore, suggest the use of deep learning approaches for imputing missing values in MS-based proteomics on larger datasets and provide workflows for these.

bioinformatics↗

bioRxiv2021

Genome binning of viral entities from bulk metagenomics data

Despite the accelerating number of uncultivated virus sequences discovered in metagenomics and their apparent importance for health and disease, the human gut virome and its interactions with bacteria in the gastrointestinal are not well understood. In addition, a paucity of whole-virome datasets from subjects with gastrointestinal diseases is preventing a deeper understanding of the viromes role in disease and in gastrointestinal ecology as a whole. By combining a deep-learning based metagenomics binning algorithm with paired metagenome and metavirome datasets we developed the Phages from Metagenomics Binning (PHAMB) approach for binning thousands of viral genomes directly from bulk metagenomics data. Simultaneously our methodology enables clustering of viral genomes into accurate taxonomic viral populations. We applied this methodology on the Human Microbiome Project 2 (HMP2) cohort and recovered 6,077 HQ genomes from 1,024 viral populations and explored viral-host interactions. We show that binning can be advantageously applied to existing and future metagenomes to illuminate viral ecological dynamics with other microbiome constituents.

genomics↗

bioRxiv2021

Deep integrative models for large-scale human genomics

Polygenic risk scores (PRSs) are expected to play a critical role in achieving precision medicine. Currently, PRS predictors are generally based on linear models using summary statistics, and more recently individual-level data. However, these predictors mainly capture additive relationships and are limited in data modalities they can use. Here, we developed a deep learning framework (EIR) for PRS prediction which includes a model, genome-local-net (GLN), specifically designed for large scale genomics data. The framework supports multi-task (MT) learning, automatic integration of other clinical and biochemical data, and model explainability. When applied to individual level data in the UK Biobank, we found that GLN outperformed LASSO for a wide range of diseases and in particularly autoimmune diseases. Furthermore, we show that this was likely due to modelling epistasis, and we showcase this by identifying widespread epistasis for Type 1 Diabetes. Furthermore, we trained PRS by integrating genotype, blood, urine and anthropometrics and found that this improved performance for 93% of 290 diseases and disorders considered. Finally, we found that including genotype data provided better calibrated PRS models compared to using measurements alone. EIR is available at https://github.com/arnor-sigurdsson/EIR.

bioinformatics↗

bioRxiv2020

A paired liver biopsy and plasma proteomics study reveals circulating biomarkers for alcohol-related liver disease

Existing tests for detecting liver fibrosis, inflammation and steatosis, three stages of liver disease that are still reversible are severely hampered by limited accuracy or invasive nature. Here, we present a paired liver-plasma proteomics approach to infer molecular pathophysiology and to identify biomarkers in a cross-sectional alcohol-related liver disease cohort of nearly 600 individuals. Metabolic functions were downregulated whereas fibrosis-associated signaling and novel immune responses were upregulated, but only half of tissue proteome changes were transmitted to the circulation. Machine learning models based on our biomarker panels outperformed existing tests, laying the foundation for a generic proteomic liver health assessment.

biochemistry↗

bioRxiv2020

APOL1 Variant-Expressing Endothelial Cells Exhibit Autophagic Dysfunction and Mitochondrial Stress

Apolipoprotein L1 (APOL1) gene risk variants (RV) associate with renal and cardiovascular disease particularly in SLE. We hypothesized that in RV-carrying human umbilical vein endothelial cells (HUVECs) cytokine-induced APOL1 expression compromises mitochondrial respiration, lysosome integrity, and autophagic flux. HUVEC cultures of each APOL1 genotype were generated. APOL1 was expressed using IFN{gamma}; HUVEC mitochondrial function, lysosome integrity, and autophagic flux were measured. IFN{gamma} increased APOL1 expression across all genotypes 20-fold (p=0.001). Compared to the homozygous G0 (ancestral) allele (0RV), high risk (2RV) HUVECs showed both depressed baseline and maximum mitochondrial oxygen consumption (p<0.01), and impaired mitochondrial networking on MitoTracker assays. These cells also demonstrated a contracted lysosome compartment (p<0.001), and an accumulation of autophagosomes suggesting a defect in autophagic flux. Treatment of 0RV HUVECs with a non-selective lysosome inhibitor, hydroxychloroquine, produced autophagosome accumulations similar to the 2RV cells, thus implicating lysosome dysfunction in blocking autophagy. Compared to 0RV and 2RV HUVECs, 1 RV cells demonstrated an intermediate autophagy defect which was exacerbated by IFN{gamma}. Our findings implicate dysfunction of mitochondrial respiration, lysosome, and autophagy in APOL1 RV-mediated endothelial cytotoxicity. IFN{gamma} amplified this phenotype even in variant heterozygous cells-a potential underpin of the APOL1/inflammation interaction. This is the first description of APOL1 pathobiology in variant heterozygous cell cultures.

cell biology↗