Carsten Olivier Daub

Background and Purpose: Multimodal integration of clinical and omics data could clarify the hepatic mechanisms of lipid‐lowering therapies, but interpretable methods for heterogeneous data remain limited. The objective was to apply interpretable multimodal networks (IMNs), a topology‐aware framework that condenses heterogeneous omics and clinical data into compact, cross‐modal modules, to characterize treatment‐specific hepatic responses to simvastatin and ezetimibe. Experimental Approach: In the Stockholm Study, patients scheduled for cholecystectomy were randomized for 4 weeks to simvastatin, ezetimibe, their combination or placebo. We integrated liver transcriptomics and DNA methylation with plasma/bile biochemistry, including lipoprotein functionality, on the human protein–protein interactome to build treatment‐specific IMNs and evaluated selected network‐derived predictions in a human hepatocyte‐like model (SOAT2‐only‐HepG2). Key Results: IMNs reduced dimensionality while preserving network topology, producing distinct molecular signatures for simvastatin, ezetimibe and combination therapy that mapped to different biological processes; the placebo network was less multimodal and more fragmented. IMNs highlighted apolipoprotein‐centred modules and predicted drug effects on APOA2, PPARA and PTPRS, linked to plasma C‐peptide. In SOAT2‐only‐HepG2 cells, selected predictions were supported experimentally, including APOA2 induction and ezetimibe‐associated PTPRS upregulation. Conclusions and Implications: IMNs provide a practical, interactome‐based route from multimodal patient data to focused, testable hypotheses about lipid‐lowering therapies, yielding compact features suitable for downstream predictive modelling while retaining biological interpretability.

Spot‐based spatial transcriptomics (ST) allows for unbiased gene expression analysis within tissue architecture, overcoming the limitations of single‐cell RNA sequencing (scRNA‐seq) by preserving spatial context. However, the high spatial resolution in ST leads to cellular heterogeneity within spots, requiring computational deconvolution to infer cellular compositions. While scRNA‐seq serves as a key reference for deconvolution, the impact of reference composition on its accuracy is still unclear. In this study, we systematically evaluate the impact of reference selection for cellular deconvolution and provide helpful guidelines for researchers. Pseudospots mimicking 55 μm Visium spots were generated from spatial transcriptomics data to evaluate global and cell type‐specific deconvolution in primary (Xenium) and metastatic (MERFISH) breast cancer samples. Focusing on state‐of‐the‐art deconvolution tools Cell2location and RCTD, we assess the influence of varying reference sizes, cell type distributions, and reference‐ST pairings, as well as the usage of large breast cancer and cross‐cancer atlases. Our findings demonstrate that even small references can yield accurate deconvolution, with RCTD and Cell2location exhibiting similar results. Spatial domains of prominent cell types like cancer and stromal cells were detected, although their contributions were systematically under‐ or over‐estimated. Additionally, Reference‐ST sample matching enhances accuracy compared to the usage of cross‐patient references, while large diverse breast cancer atlases also provided reliable results. This study concluded that reference selection had modest effects on RCTD and Cell2location, but matching samples and large atlases stabilize outcomes. Nonetheless, performance varies between cell types and annotation levels, and deconvolution results should always be interpreted with caution.

The domestic dog is a powerful genetic model for complex traits, disease and behaviours relevant to human biology, yet the canine genome still lacks a systematic transcription-defined atlas of regulatory elements. Existing resources have improved annotation but largely infer regulatory activity from chromatin features rather than directly mapping transcription initiation at promoters and enhancers. Here, we address this gap using 114 CAGE-seq libraries spanning 56 tissues and developmental stages from 9 dogs. We identify 68,446 promoters and 46,661 active enhancers, and define their tissue-enriched activity across the canine body. Regulatory programmes are organized around shared transcription factor motif infrastructures, with prominent enrichment of KLF family motifs across tissues. The cerebellum emerges as a major regulatory hub, showing the highest density of enhancer-promoter interactions among brain regions. During embryogenesis, enhancer activity shifts from early neurodevelopmental patterning programmes, including OTX2 and ZIC1/4, towards regulators of functional maturation and synaptic organization, including AQP4 and NLGN3. Comparative analyses revealed 69 enhancers with highly similar regulatory structures linked to potential orthologous genes in dog and human. Together, these data establish a tissue-resolved transcription-based atlas of the canine regulatory genome that advances functional annotation and improves interpretation of non-coding variation in comparative and disease genomics.

The domestic dog is a powerful genetic model for complex traits, disease and behaviours relevant to human biology, yet the canine genome still lacks a systematic transcription-defined atlas of regulatory elements. Existing resources have improved annotation but largely infer regulatory activity from chromatin features rather than directly mapping transcription initiation at promoters and enhancers. Here, we address this gap using 114 CAGE-seq libraries spanning 56 tissues and developmental stages from 9 dogs. We identify 68,446 promoters and 46,661 active enhancers, and define their tissue-enriched activity across the canine body. Regulatory programmes are organized around shared transcription factor motif infrastructures, with prominent enrichment of KLF family motifs across tissues. The cerebellum emerges as a major regulatory hub, showing the highest density of enhancer-promoter interactions among brain regions. During embryogenesis, enhancer activity shifts from early neurodevelopmental patterning programmes, including OTX2 and ZIC1/4, towards regulators of functional maturation and synaptic organization, including AQP4 and NLGN3. Comparative analyses revealed 69 enhancers with highly similar regulatory structures linked to potential orthologous genes in dog and human. Together, these data establish a tissue-resolved transcription-based atlas of the canine regulatory genome that advances functional annotation and improves interpretation of non-coding variation in comparative and disease genomics.

Plasma cell-free RNA (cfRNA) reflects tissue- and cell-type-specific activity across pathological states and is a promising biomarker for organ injury and disease. Computational deconvolution methods are widely used to infer organ and cell-type contributions to cfRNA profiles. However, most were originally developed for single-tissue bulk transcriptomes and their performance in body-wide cfRNA settings, where any tissue or cell type can contribute, remains poorly characterised. Here, we present a systematic benchmarking of tissue- and cell type-of-origin deconvolution for plasma cfRNA that considers both methodological and reference-related sources of variability under realistic cfRNA simulation settings. We evaluated seven commonly used deconvolution methods across distinct algorithmic classes and multi-organ reference configurations derived from bulk and single-cell atlases. We assessed performance using simulation frameworks that model multi-organ mixtures, technical noise, and transcript degradation. We further examined deconvolution methods across multiple previously published clinical cfRNA cohorts spanning diverse disease contexts. Across both tissue- and cell-type-level analyses, deconvolution performance was strongly influenced by both method choice and reference parameters. Tissue-of-origin inference was comparatively robust across simulated and clinical datasets, recovering disease-associated organ signals and concordance with biochemical markers. In contrast, cell type-of-origin inference showed greater variability and reduced consistency across analytical settings, leading to divergent interpretations in both simulations and published clinical cfRNA cohorts. Together, these findings demonstrate that methodological and reference-related variability are major sources of uncertainty in cfRNA deconvolution, with tissue-level inference being more robust than cell-type-level inference. Our benchmarking framework provides guidance for reference selection and comparative interpretation in cfRNA deconvolution.

Background Spot-based Spatial transcriptomics (ST) allows for unbiased gene expression analysis within tissue architecture, overcoming the limitations of single-cell RNA sequencing (scRNA-seq) by preserving spatial context. However, the high spatial resolution in ST leads to cellular heterogeneity within spots, requiring computational deconvolution to infer cellular compositions. While scRNA-seq serves as a key reference for deconvolution, the impact of reference composition on its accuracy is still unclear. In this study, we systematically evaluate the impact of reference selection for cellular deconvolution and give helpful guide-lines to researchers to address this task. Methods We systematically evaluate the impact of single-cell reference selection for global and cell type-specific deconvolution performances in primary and metastatic breast cancer ST datasets. Focusing on state-of-the-art deconvolution tools Cell2location and Robust Cell Type Decomposition (RCTD), we assess the influence of varying reference sizes, cell-type distributions, reference-ST pairings,as well as the usage of large breast cancer and cross-cancer atlases. Results Our findings demonstrate that even small reference datasets can yield accurate deconvolution results, with RCTD and Cell2location exhibiting similar performances across cell-types. Heterogeneous entities, such as myeloids, presented greater challenges for accurate deconvolution compared to more distinct ones, such as epithelial cells. Spatial domains associated with prominent cell types like cancer and stromal cells were detected, although their contributions were systematically under- or over-estimated Additionally, we found that perfect reference-ST matching enhances deconvolution accuracy compared to the usage of cross-patient references. Finally, large breast cancer single-cell atlases composed of a diverse set of patients were also able to provide reliable deconvolution results. Conclusion This study provides important insights into optimizing reference selection for ST deconvolution, highlighting the strengths and limitations of current computational tools in addressing cellular heterogeneity within spatial transcriptomics datasets.

STUDY QUESTION: What is the effect of different oxygen (O ₂ ) levels on the transcriptomic profile of bovine embryos during the in vitro culture? SUMMARY ANSWER: Embryos grown in hypoxia (6% O ₂ ) from zygotes until the blastocyst stage had the highest blastocyst formation rate, whereas normoxia (20% O ₂ ) delayed transcriptomic reprogramming and embryonic genome activation, and induced changes in energy metabolism gene expression. WHAT IS KNOWN ALREADY: Mammalian preimplantation embryo development is a complex sequence of events where, within a week, the zygote is reprogrammed to totipotency and subsequently diverges to embryonic and extraembryonic cell lineages for post-implantation embryo development. This period of development is sensitive to oxygen levels that can affect various cellular processes. STUDY DESIGN, SIZE, DURATION: In this study, we used triplicate bovine embryos as a model for human embryogenesis to compare the influence of O ₂ levels on preimplantation embryonic development by culturing embryos either in normoxic (20% O ₂ ) or physiological hypoxic (6% O ₂ ) conditions, or sequential hypoxia until 16-cell stage and then switching to ultrahypoxic culture (2% O ₂ ). PARTICIPANTS/MATERIALS, SETTING, METHODS: As the readout for varied O ₂ effects, we performed RNA sequencing using 5’ targeted STRT-N method on single embryos. We compared zygotes, 4-, 8-, 16-cell and blastocyst stage embryos grown in either normoxic or hypoxic condition, adding ultrahypoxia for blastocyst stage embryos as the third condition. MAIN RESULTS AND THE ROLE OF CHANCE: We found that the initial cleavage rate was not affected by O ₂ levels but there was a clear difference in blastocyst formation rate. In hypoxia, 36% of embryos reached blastocyst stage while in normoxia the blastocyst formation rate was 13%. In final ultrahypoxia condition only 4.6% of embryos reached blastocyst stage. Transcriptomic profiles showed that normoxic conditions slowed down oocyte transcript degradation and embryonic genome activation. Key metabolic enzyme genes were also altered between hypoxic and normoxic conditions at the blastocyst stage. Both hypoxic and ultrahypoxic conditions induced energy production by upregulating genes involved in glycolysis and lipid metabolism typical to in vivo embryos. In contrast, normoxic conditions failed to upregulate glycolysis genes and only depended on primitive oxidative phosphorylation metabolism. We conclude that constant hypoxia culture of in vitro embryos provided the highest blastocyst formation rate and appropriate energy metabolism. Normoxia altered the energy metabolism and decreased the blastocyst formation rate. Even though ultrahypoxia at blastocyst stage resulted in a drop of blastocyst formation, the transcriptional profile of surviving embryos was normal. LARGE SCALE DATA: The raw data (BCL files) are available at Zenodo: XXXXX FASTQ files generated are available in the EMBL’s European Bioinformatics Institute (EMBL-EBI)-BioStudies with accession number X-XXXX. LIMITATIONS, REASONS FOR CAUTION: The limitation of this study is the use of bovine as an animal model instead of human embryos. Due to this, the direct translation of the results to human should be taken with caution. WIDER IMPLICATIONS OF THE FINDINGS: This study supports previous literature on hypoxic culture conditions being the most suitable for in vitro embryo culture. In addition, we provide new insights on why embryos grown in normoxia do not have the same success rate as embryos grown in hypoxia. We did not observe any benefits of lowering the oxygen levels to 2%, calling for caution of switching to this culture system. STUDY FUNDING/COMPETING INTEREST(S): This project has received funding from the European Union’s Horizon 2020 Research and Innovation Programme under the Marie Sklodowska-Curie grant agreement No. 813707.; Work in the JK laboratory is supported by Jane and Aatos Erkko Foundation, Sigrid Jusélius Foundation, Liv och Hälsa (Finland), Swedish Brain Foundation and Swedish Research Council. The study was also supported by the Estonian Research Council (grant no. PRG1076) and the Horizon Europe NESTOR project (grant no. 101120075). What this means for patients?: During assisted reproduction treatment (ART) oocytes are fertilized in vitro by sperm to form an embryo. The embryos are then exposed to environmental effects during the 5 days of culture before implantation. Embryo culture takes place inside the incubator, where different oxygen levels can be used. In this study, we compared 3 different oxygen concentrations for embryo culture: atmospheric or normoxia (20%), physiologic or hypoxia (6%) and combined low and ultra-low concentration with initial culture at 6% oxygen until day 3 followed by 2% (ultrahypoxia) until day 5 of culture. We conducted this study because many ART clinics use different oxygen concentrations for embryo culture, and we aimed to provide insight on how these different concentrations impact embryo development by monitoring gene expression. As using human embryos for this type of experiments is not commonly approved, we used bovine embryos whose early development resembles that of humans. We found that culturing embryos for 5 days at constant low oxygen concentration (hypoxia) gives the highest number of embryos that reach the blastocyst stage (day 5) and have expected gene expression that provides the physiological path of embryo development. Culturing embryos at atmospheric oxygen concentration showed that only 13% of fertilised eggs reached the blastocyst stage at day 5, while culturing at the final ultra-low oxygen concentration had the lowest percentage of matured embryos, and we did not observe any benefits that would justify the use of such a culture system. We suggest that in vitro culturing of embryos at constant 5-6% oxygen gives the best results.

The dog, Canis lupus familiaris, is an important model for studying human diseases. Unlike many model organisms, the dog genome has a comparatively poor functional annotation, which hampers gene discovery for development, morphology, disease, and behavior. To fill this gap, we established a comprehensive tissue biobank for both the dog and wolf samples. The biobank consists of 5485 samples representing 132 tissues from 13 dogs, 12 dog embryos, and 24 wolves. In a subset of 100 tissues from nine dogs and 12 embryos, we characterized gene expression activity for each promoter, including alternative and novel, i.e., previously not annotated, promoter regions, using the 5’ targeting RNA sequencing technology STRT2-seq. We identified over 100,000 promoter region candidates in the recent canine genome assembly, CanFam4, including over 45,000 highly reproducible sites with gene expression and respective tissue enrichment levels. We provide a promoter and gene expression atlas with interactive, open data resources, including a data coordination center and genome browser track hubs. We demonstrated the applicability of Dog Genome Annotation (DoGA) data and resources using multiple examples spanning canine embryonic development, morphology and behavior, and diseases across species.

Zebrafish, a popular organism for studying embryonic development and for modeling human diseases, has so far lacked a systematic functional annotation program akin to those in other animal models. To address this, we formed the international DANIO-CODE consortium and created a central repository to store and process zebrafish developmental functional genomic data. Our data coordination center (https://danio-code.zfin.org) combines a total of 1,802 sets of unpublished and re-analyzed published genomic data, which we used to improve existing annotations and show its utility in experimental design. We identified over 140,000 cis-regulatory elements throughout development, including classes with distinct features dependent on their activity in time and space. We delineated the distinct distance topology and chromatin features between regulatory elements active during zygotic genome activation and those active during organogenesis. Finally, we matched regulatory elements and epigenomic landscapes between zebrafish and mouse and predicted functional relationships between them beyond sequence similarity, thus extending the utility of zebrafish developmental genomics to mammals.