BMC Genomics

Generating high-quality genome assemblies for small animals with large genomes is complex due to their small body size, DNA contamination, and repetitive elements. Ticks exemplify these complexities, while also being a global health threat to humans, domestic animals, and wildlife. Advances in long-read sequencing platforms now make it feasible to obtain large amounts of raw sequence data from individual specimens, but challenges remain. Key genome workflow challenges include error correction, assembly, transposable element annotation, and chromosome assignment and synteny. Here we examine three individual Dermacentor reticulatus ticks using deep Oxford Nanopore sequencing. Comparing and contrasting bioinformatic tools for raw read and assembly manipulation allows us to identify the parameters that provide a high-quality haploid genome assembly among 38 assemblies. We find Dorado error corrected raw read data from approximately two flow cells are needed, but limiting assembly coverage to 40x produces the highest quality assemblies. Then we examined the effects of optimal workflow parameters on downstream analyses of gene synteny and manual transposable element annotation of the three tick assemblies in comparison to an independent assembly of a fourth D. reticulatus individual from a different country. Gene synteny analysis allows chromosome assignment of scaffolds and transposable element identification was improved markedly with manual curation. Finally, we compared the genetic variation of D. reticulatus across two populations and found similar genetic diversity. Our study provides a clear workflow to obtain high-quality assemblies from Oxford Nanopore sequences and genetic characterization of a species with a large, complex, and repetitive genome.

The sheep research community benefits from a genome assembly that has been refined over the last 15 years. However, the rich diversity of sheep breeds suggests that a single individual genome is not enough to represent the species full genomic content. Third-generation sequencing technologies have recently made it possible to generate high quality assemblies that should enable to elucidate unique and/or complex phenotype determinisms depending on genetic structural variations. In the SeqOccin project, twelve breeds were selected to represent (mostly French) sheep genetic diversity: Basco-Bearnaise, Black Belly, Blanche du Massif Central, Corse, Ile de France, Lacaune, Manech Tete Noire, Noire du Velay, Ouessant, Rambouillet, Vendeen, Soay. High-molecular-weight DNA was extracted from blood samples and subjected to long read whole-genome sequencing. The reads were used to generate de novo assemblies and were aligned onto the public reference to call structural variants. PacBio HiFi and parental Illumina reads were used to assemble phased haplotypes of the Lacaune breed. For the other breeds, PacBio raw (CLR) or ONT reads were generated. Our work demonstrates the feasibility of assembling genomes with a reasonable amount of data and appropriate software packages, delivering one high-quality assembly and 11 standard sheep assemblies that pave the way for pangenome construction and structural variant functional studies.

Background Comparisons of single-cell and single-nucleus RNA sequencing (scRNA-seq and snRNA-seq) data have been described in some mammalian tissues and, subsequently, in Drosophila, but remain unexplored in most invertebrate lineages. The xenacoelomorphs occupy key phylogenetic positions, yet they differ anatomically from mammals. They have a reduced extracellular matrix, high-salt body fluid, and no circulatory system. Despite these differences, they possess a well-developed nervous system. One of the xenacoelomorphs, the photosymbiotic acoel (Praesagittifera naikaiensis) also harbours symbiotic Tetraselmis algae, whose RNA can be co-captured with host RNA. Results We compared scRNA-seq and snRNA-seq data from whole P. naikaiensis specimens. Both methods yielded high-quality data with comparable gene detection but a larger share of scRNA-seq reads derived from symbiotic algae. Gene-level analyses revealed that neural genes were enriched in snRNA-seq relative to non-neural genes. Cross-method label transfer and integration-based validation identified six snRNA-seq clusters, including some neural populations, that lacked a clear scRNA-seq counterpart. In contrast, three cell populations, including muscle and metabolically active clusters, were reciprocally validated as captured by both methods. Conclusions Our results show that key snRNA-seq advantages, particularly the enhanced recovery of neural transcripts, are recapitulated in our dataset, which is consistent with previous reports in mammals. Furthermore, snRNA-seq reduces symbiont-derived reads and recovers several cell populations underrepresented in scRNA-seq. These findings provide practical guidance for cell atlas construction in non-model, symbiotic invertebrates.

Candida auris is an emerging multidrug-resistant fungal pathogen. However, little is known about the role of long noncoding RNAs (lncRNAs) in the human immune response to this fungus. We reanalyzed RNA-sequencing data from peripheral blood mononuclear cells obtained from three healthy donors. The cells were exposed to live C. auris, purified mannan, purified {beta} glucan, or RPMI control for 4 h and 24 h. We identified 607 putative novel polyadenylated lncRNA loci that were absent from GENCODE Release 48. Of the 449 novel loci tested for differential expression, 102 responded to at least one stimulus. Novel-lncRNA responses were limited at 4 h but increased markedly at 24 h. At 24 h, 78 novel lncRNAs responded to live C. auris, 34 to mannan, and 11 to {beta}-glucan. Many of these responses were specific to a particular stimulus. Among the 102 responsive lncRNAs, 94 were significantly co-expressed with protein coding genes. The coding-gene partners of the most highly connected lncRNAs were associated with innate immunity, defense responses, lysosomal functions, and signaling pathways. Temporal analysis identified 19 novel lncRNAs whose responses changed between 4 and 24 h. Fourteen were also differentially expressed and were selected for further analysis. Combining temporal response, co-expression, and genomic proximity identified 4 candidate cis-associated pairs: MSTRG.32011-TLR2, MSTRG.5763-DUSP5, MSTRG.25627-HCK, and MSTRG.18159 ITGB3.These findings reveal a predominantly late and stimulus-specific novel-lncRNA response to C. auris and provide prioritized candidates for experimental validation.

Carbohydrate-active enzymes (CAZymes) drive the turnover of polysaccharides in the oceans. However, the scale of diversity in polysaccharide utilization and genome organization strategies has yet to be characterized, particularly in marine ecosystems. In this paper, we introduce mpcgcdb.com as an interactive web catalog for marine CAZyme gene clusters, their associated metagenome-assembled genomes, and relevant enzyme families found in marine metagenomic samples. This database contains nearly 290,000 marine CAZyme gene clusters from more than 22,000 genomes, enabling user-initiated exploratory visualizations for genome organization networks and enzyme phylogeny. This tool will assist researchers in developing hypotheses about gene function, connecting species to metabolic niches, and identifying gaps where future work is needed to understand different pathways in marine glycan cycling.

Cell-type-specific molecular responses are obscured in bulk RNA sequencing, whereas generating matched single-cell datasets across environmental conditions remains costly and technically challenging, particularly in non-model organisms. Here, we present a workflow that integrates existing bulk and single-cell RNA-seq resources to prioritize cell-type-associated environmental-response genes. We demonstrate the workflow by examining heat-stress-associated neuronal genes in the reef-building coral Stylophora pistillata. Neuronal genes were first identified based on cell type-specific expression in single-cell transcriptomic data and then intersected with differentially expressed genes from bulk RNA-seq datasets under heat-stress. In parallel, we applied a deconvolution approach to infer cell type-specific expression changes directly from bulk data, independently capturing neuronal transcriptional responses. This integrated pipeline reduced thousands of heat-responsive genes previously identified in bulk RNA-seq studies to nine heat-associated, neuron-enriched candidates. Eight of the nine candidates showed their highest mean expression in the same neuronal subcluster. Targeted sequence-based characterization supported the identification of a TRPA-family ion channel, two calmodulin/centrin-family Ca2+-binding proteins, and a Class A/rhodopsin-like GPCR of unresolved ligand specificity. Two additional candidates were predicted secreted proteins belonging to conserved coral protein families, while the molecular functions of the remaining candidates remained unresolved. Their coordinated regulation suggests heat-associated remodeling of neuronal sensory and intracellular signaling processes. Our workflow provides a scalable framework for extracting cell type-associated molecular responses from existing transcriptomic resources and prioritizing candidates for further investigation. More broadly, it offers a generalizable strategy for investigating cell type-specific gene expression responses across organisms, cell types, and environmental conditions without requiring new single-cell datasets.

Genetic rearrangement in the absence of selection is poorly understood. Long-read, high-depth sequencing has enabled analysis of the genetic landscape of spontaneous structural variants. We have applied PacBio Revio sequencing to identify the major classes of genetic alterations during long-term vegetative growth in Aspergillus flavus. These classes include a) precise deletions frequently associated with the flanking region of homology, b) clusters of short deletions that are heavily enriched within centromeric and subtelomeric regions that are typically 94% A+T, and c) deletions and expansions within a subset of homopolymer tracts. We find 70 homopolymer tracts longer than 44 bp, dispersed throughout the genome in noncoding regions and unassociated with transposable elements. A broader analysis of third-generation sequenced datasets from fungi found within the Phylum Ascomycota showed a high but variable prevalence of homopolymeric tracts across species. The adventitious identification of a putative flbA mutation, responsible for an altered conidial phenotype, by the Deep Variant AI system also demonstrates the power of this approach. These studies show that high-depth, long-read sequencing can probe SNPs, indels, homopolymeric tracts, and large-scale rearrangements in an unbiased fashion, and demonstrate that short deletions and homopolymer tract expansions arise spontaneously in A. flavus and other filamentous fungi.

A complete reference genome for species with chromosomally-determined separate sexes should contain scaffolds for all sex chromosome homologs. However, sex chromosomes present distinct computational challenges compared to autosomes. Here we present a k-mer based analysis that utilizes whole-genome sequencing of a few sex-identified isolates: Cytogenetics-By-Sequencing (CBS) tools. Unlike other approaches that typically address one aspect of the sex chromosomes, CBS-tools strives to guide users from the discovery of the heterogametic sex through identifying the sex-determination region (SDR). The core of CBS-tools is automated quantification of sex-specific k-mers in order to predict the heterogametic sex. Using publicly-available datasets, CBS-tools correctly identified the known sex-system of the 31 species tested. Additionally, we used these k-mers to verify and correct phasing of sex chromosomes between haplotypes in species representing different sex-systems. Finally, we used these k-mers to delimit the SDR boundary using an interactive web platform. CBS-tools was developed with previously unexplored sex chromosome systems in mind, but is also suitable for well-examined sex chromosome pairs.

Structural variants (SVs) can strongly influence evolutionary processes by suppressing recombination to create jointly inherited regions and by altering gene expression and the 3D organization of DNA. Recent improvements in sequencing technologies have led to an influx of high-resolution genomic data, bolstering research on SVs. When detecting and characterizing SVs, researchers face trade-offs in data and methods concerning scale, resolution, and computational intensity. Here, we evaluate such trade-offs by quantifying how data type, analytical approach, and SV attributes affect the ability to detect and characterize SVs in Timema cristinae stick insects, with a focus on inversions. We specifically evaluate inversions characterized by (i) a population genomic approach based on patterns of local population structure inferred from genotyping-by-sequencing data, (ii) pairwise alignments of eight de novo genome assemblies and SV calling with SyRI, and (iii) a pangenome constructed from the alignment of the eight genome assemblies. Our results suggest that while broad genomic regions harboring SVs can be detected by any of the approaches, the approaches differ in how they characterize inversions, especially small or complex ones. The local population structure approach is exploratory and generally only detects the largest inversions. Pairwise comparative alignments identify the greatest number of inversions, but downstream analyses are influenced by identification of shared inversions across genome pairs. Pangenome approaches are scalable and produce network graphs that describe complex SVs but require a conceptual shift in visualization and interpretation. The collective results highlight limitations and benefits of different approaches in a burgeoning research field.

Background: Pangenome variations graphs (PVG) are promising structures to study diversity and reduce reference bias in genomics. However, it is not currently possible include complete annotation of the PVG. In a previous work, we introduced GrAnnoT, a tool for transferring the annotation from a reference genome to a PVG, and from a PVG to its embedded genomes. However, GrAnnoT was limited to transferring one annotation to the graph, and thus could not include all genes. Results: We implemented new options to GrAnnoT, that can now handle annotations of multiple genomes, and integrate them in the graph. We also developed GrAPEx, a pipeline that identifies, extracts and annotates non-annotated regions in the graph. We tested those two tools to attest their performances. Our benchmarks revealed that both GrAnnoT and GrAPEx are reliable. GrAnnoT is now able to integrate correctly several annotations, increasing the amount of biological information available in the graph. Additionally, it offers useful options to study gene distribution in the PVG. GrAPEx allows rapid annotation of the whole PVG, with no reduction in the annotation quality. Conclusion: Overall, the two tools presented here allow to annotate thoroughly the whole PVG, and integrate biological annotation to the pangenome. It also reduces the previous reference bias, where only one annotation (often the reference one) could be included in the graph. This constitutes a precious addition to the pangenomic toolkit.

Genomic structural variants (SVs) are important sources of genetic diversity and contribute to the adaptation and evolution of plant-pathogenic fungi. Colletotrichum graminicola, the causal agent of maize (Zea mays L.) anthracnose, causes significant yield losses and serves as an important model for genetic studies. To investigate genomic plasticity in this pathogen, long-read genome assemblies were generated for four C. graminicola field strains collected in France, Croatia, Brazil, and Argentina and compared with the chromosome-level reference genome from the USA. Comparative analyses revealed extensive structural variation across core chromosomes, including large-scale translocations, inversions, duplications, and insertions and deletions, whereas minichromosomes showed markedly reduced collinearity among strains. Secreted proteins and predicted effectors were located significantly closer to regions affected by SVs. Approximately 174 kb of segmental duplications lacking genes were shared between core chromosome 04 and minichromosomes 11 and 13 in the reference genome but were absent from the other field strains, revealing a strain-specific structural link between core and minichromosomes. Protein clustering identified 733 candidate strain-specific genes across the reference genome. Approximately 72% of the genes annotated on the minichromosomes were strain-specific, indicating that most of their annotated genes lacked orthologous gene models in the other strains. Of the 733 candidate strain-specific genes, 492 retained homologous genomic sequences in at least one other strain and many also showed evidence of gene disruption. This pattern is consistent with lineage-specific pseudogenization and suggests that apparent gene presence-absence variation may arise through several evolutionary processes, including gene gain, gene loss and loss of gene function. Insertions and deletions in potentially regulatory flanking regions were associated with plant-pathogen interaction genes. In addition, interpretable machine learning analyses revealed a co-occurrence of duplications and LTR retrotransposons in genomic regions associated with secreted genes. This study shows that structural variation, transposable elements, and lineage-specific pseudogenization are important features of genome evolution in C. graminicola and identifies associations between genomic variation and genes involved in plant-pathogen interactions.

BackgroundN6-Methyladenosine (m6A) regulates RNA fate in flowering plants, but the repertoire and expression patterns of m6A-related genes remain incompletely characterized in liverworts. Here, we characterized these gene families in the liverwort Marchantia polymorpha within a comparative framework spanning diverse plant lineages. ResultsA comparative survey of 119 plant genomes using sequence similarity and conserved-domain criteria identified 18 candidate m6A-related loci in M. polymorpha, comprising seven writer-related genes, five AlkB homolog genes, and six YTH- or CPSF30-related genes. Across the surveyed genomes, canonical writer components generally had low copy numbers, whereas AlkB homolog and reader-related families showed greater copy-number variation. Phylogenetic and domain analyses supported the family assignments of the M. polymorpha candidates. Eight of the 18 candidate loci showed detectable collinearity with Marchantia quadrata. Putative promoter analysis identified motifs annotated as hormone-, light-, and stress-responsive, with hormone-responsive motifs accounting for the largest share of predicted sites. Public transcriptome analysis revealed variation in candidate gene expression across developmental stages and tissues. Quantitative reverse-transcription PCR showed higher transcript abundance for all six YTH- or CPSF30-related genes in thalli than in gemmae or gemma cups. Cold, heat, osmotic, salt, and dark treatments elicited gene-specific expression responses. Darkness increased transcript abundance of several candidates, including MpMTA, MpALKBH1A, and MpECT2. A predicted protein-association network suggested links between subsets of the candidates and RNA-processing proteins. ConclusionsThese findings define a candidate m6A-related gene repertoire in M. polymorpha and identify tissue- and stress-responsive genes for targeted functional studies of RNA methylation in this liverwort.

Prospero-related homeobox (Prox) genes encode a family of transcription factors that play essential roles in the development of several organs and systems, including the central nervous system, lymphatic endothelium, musculature, and liver. Despite their developmental importance, the evolutionary history of the vertebrate Prox gene family remains poorly understood. In this study we combined phylogenetic and synteny analysis to characterise the evolution of the Prox family in vertebrates. Our results reveal that two to three Prox subfamilies were already present in the last common ancestor of jawed vertebrates. We clarify the identity and evolutionary relationships of well-studied members of this family and identify multiple independent losses of Prox2 and Prox3 genes in specific vertebrate lineages. Furthermore, we uncover evidence for the existence of a fourth Prox gene in the ancestral vertebrate genome, which was subsequently lost. Overall, this study provides the first comprehensive analysis of the evolutionary history of the vertebrate Prox gene family and establishes a foundations for future studies on the functional roles of these genes.

Mouse models are widely used in translational neuroscience, but whether a gene has the same partners in mouse and human prefrontal cortex (PFC) is unknown. The genes whose expression rises and falls with a given gene across samples (its co-expression partners) indicate the molecular program to which it belongs. We compared the co-expression of 13,214 genes between mouse medial PFC and human dorsolateral PFC in 246,211 single-nucleus RNA-sequencing profiles. These two regions are not homologous, so part of any measured difference may reflect region and not species. The study has five main findings. First, the similarity of partners for single genes was modest (median 0.169; 0, unrelated; 1, identical). This median was below that of every comparison among human datasets but within the range of comparisons among mouse datasets, and correction for measurement noise raised it to 0.349 at most. Second, groups of co-expressed genes (modules) were strongly preserved. Third, neuropsychiatric disease genes were modestly more conserved than other genes, but no more than control genes matched for mutational constraint and expression, consistent with general gene properties and with how these genes are discovered. Fourth, no disease gene was shown to have different partners in the two species. Fifth, four of five human modules recurred in an independent human dataset. We also identified two sources of error: analyzing only the most variable genes excludes most disease genes, and networks built from few samples, as for glial cells, underestimate the similarity between species. Scores and code are released for cross-species studies.

Chromatin profiling methods, such as ChIP-seq (chromatin immunoprecipitation followed by sequencing), are used to characterize the genomic localization of DNA-associated proteins. While ChIP and ChIP-seq have been used for decades, an orthogonal approach, CUT&Tag (Cleavage Under Targets and Tagmentation), is gaining popularity as an efficient and cost-effective alternative. Although previous comparative studies note discrepancies in signal-to-noise ratios and detection bias at certain genomic regions, many differences between ChIP-seq and CUT&Tag results remain largely unreconciled. Here, we systematically investigate the disagreeing signals captured by these two methods across well annotated genomic regions. We assess multiple histone mark profiles generated by different groups in two cell lines (K562 and MCF-7). Overall, our analysis indicates that compared to ChIP-seq, CUT&Tag may have limited sensitivity in low-GC environments and, as previously observed, increased signal in hyper-accessible chromatin. Notably, within GC-poor regions of active gene bodies and Polycomb-repressed domains, CUT&Tag exhibits a loss of H3K36me3 and H3K27me3 signal, respectively, where occupancy of these histone marks is otherwise expected. Further, active promoters with discrepant H3K4me3 and H3K27ac signal between the two assays differ systematically in GC content and chromatin accessibility. Promoters differentially enriched for CUT&Tag signal relative to ChIP-seq generally show higher, broader GC-content profiles and higher DNase-seq and ATAC-seq signal, while promoters enriched for ChIP-seq signal harbor the opposite characteristics. A local bias for high GC content and/or chromatin accessibility in CUT&Tag may also explain its differing signal patterns at nucleosome-depleted regions (NDRs) compared to ChIP-seq and MNase-seq promoter profiles. Altogether, our results highlight that studies focused on profiling the intensity and structure of histone modification occupancy can be sensitive to potential biases of CUT&Tag to GC content and chromatin accessibility. These characteristics should be accounted for when selecting a chromatin profiling approach, analyzing and interpreting data as well as drawing biological conclusions.

Environmental exposures can induce epigenetic modifications that persist across generations, potentially contributing to the transmission of environmentally-induced phenotypes. However, the extent to which such persistence of molecular changes are maintained independently of genetic mechanisms remains unclear. In this study, we investigated the evolution and genetic determinism of DNA methylation (DNAm) variation across three successive generations of Japanese quails (G0 to G2) belonging to two epilines, defined according to whether their female ancestor had received genistein supplementation (epi+) or not (epi-). Using reduced representation bisulfite sequencing (RRBS), we characterised CpG methylation patterns and identified differentially methylated cytosines (DMCs) and regions (DMRs) between control and genistein-supplemented epilines. Out of 112,745 CpG sites analysed, no DMCs or DMRs were detected between epilines in the first generation following exposure (G0), whereas 621 DMCs (36 DMRs) were identified in G1, and 1,381 DMCs (101 DMRs) in G2. Despite this progressive increase of differential methylation sites, mean differences in methylation rate between epilines were globally small. Similarly, genome-wide genetic differentiation between epilines was very limited. Overall, SNP-based heritability of DNA methylation was 0.19 indicating that genetic variation accounted for only a small part of the global variation of DNAm levels. By contrast, DNAm at differentially methylated CpG sites was extensively under genetic control, with a mean SNP-based heritability > 0.5, and meQTL analysis further identified significant associations between SNPs and methylation levels. Together, these results indicate that local DNAm differentiation in quail is under substantial genetic regulation, while the limited genetic differentiation between groups suggests that genetic variation alone can not fully explain the progressive accumulation of methylation differences following an ancestral exposure to genistein. These results support DNAm as a plausible molecular candidate mechanism in the multigenerational response to environmental exposure and highlight the complex interplay between genetic and epigenetic regulations in the inheritance of phenotypes.

The goat research community benefits from one of the most comprehensive livestock genome assemblies, ARS1. However, the rich diversity of goat breeds suggests that a single individual genome is not enough to represent the species full genomic content. Third-generation sequencing technologies have recently made it possible to generate very high quality assemblies that should enable to elucidate unique and/or complex phenotype determinisms depending on genetic structural variations. In the SeqOccin project, nine goat breeds were selected to represent French goat genetic diversity: Alpine, Boer, Creole, Fosses, Poitevine, Provencale, Pyreneenne, Rove and Saanen. High-molecular-weight DNA was extracted from blood samples and subjected to long read whole-genome sequencing. There reads were used to generate de novo assemblies and were aligned onto the public reference to call structural variants. PacBio HiFi and parental Illumina reads were used to assemble phased haplotypes of the Alpine breed. For the other breeds, PacBio raw (CLR) or ONT reads were generated. Our work demonstrates the feasibility of assembling genomes with a reasonable amount of data and appropriate software packages, delivering one high-quality assembly and 8 standard goat assemblies that pave the way for pangenome construction and structural variant functional studies.

Premise:: Conifers comprise one-sixth of global forests and landscape genomic studies are urgently needed to understand their biology, yet relatively few conifer species have reference genomes. Targeted exon-capture is a practical alternative to whole-genome sequencing, but selecting loci without a reference genome remains a barrier. We developed a multi-tissue reference transcriptome for pi non pine (Pinus edulis) to ask how multi-tissue sampling improves probe design, and what is lost under a conventional single-tissue approach. Methods: We assembled a reference transcriptome from four diploid tissues (needle, phloem, root, vegetative bud) and one haploid tissue (megagametophyte), using the latter as a pseudo-guide. We designed a target-capture probe-set by prioritizing highly expressed P. edulis transcripts and transcripts matching a conifer exon database. Results: We recovered 76% of conserved plant genes with few duplicated transcripts. Filtering yielded 15,664 transcripts totaling 31.4 Mb, ~0.1% of the genome. Tissue-specific expression was substantial: 27.4% of transcripts occurred exclusively in non-needle tissues. Functional annotation confirmed complementary contributions, including seed-desiccation and germination genes from megagametophytes, and defense- and stress-associated genes from roots and phloem. Discussion: We built a transcriptome and efficient probe set using a roadmap suitable for other non-model plants with large genomes; keys to success were prioritizing complementary tissues and using expression-based filtering.

Genomic variant calling is a central computational task in disease research and diagnostics, where somatic mutations such as JAK2 V617F serve as diagnostic markers in myeloid neoplasms. VarDict, originally written in Perl, and its reimplementation VarDictJava are widely used variant callers in the bioinformatics community. The Perl implementation requires run times of up to several hours; VarDictJava reduces this to a few minutes but can consume large amounts of main memory, which limits the number of jobs that can be run in parallel on a given machine. We present vardictcpp, a functionally identical reimplementation of VarDictJava in C++17. vardictcpp is approximately four times faster in single-core execution and 9-11 times faster using eight cores, while requiring up to 17 times less main memory. Variant calls are identical to those of VarDictJava on the evaluated whole-exome samples. The reduced memory footprint allows more concurrent jobs on existing hardware and thereby increases sample throughput, which is particularly relevant for clinical sequencing laboratories with fixed compute resources. vardictcpp is implemented in C++17 and freely available at https://github.com/MHH-Bioinformatics-Hematology/vardictcpp

Direct RNA nanopore sequencing offers the capacity to directly profile native RNAs, enabling detection of diverse epitranscriptomic modifications based on current changes as the RNA passes through a pore. Innovations in chemistry, hardware and bioinformatic tools have increased sequencing capacity and accuracy, creating demand for sequencing multiple samples within individual runs. Two recent solutions enable barcoding and demultiplexing of samples, but independent comparisons of their performance is limited. Here we benchmark WarpDemuX and SeqTagger using orthogonal ground truths derived from species identity and matched IVT spike-ins across four RNA004 flow cells. Both achieve high demultiplexing precision but differ in their trade-offs in precision and retention. SeqTagger provided strong performance at its recommended settings, whilst WarpDemuX performance varied by model, barcode and read quality. IVT-derived score cutoffs allowed dataset-specific calibration and increased WarpDeMux assignment yield, although precision achieved on IVTs was not consistently reproduced in species reads. These results support matched IVTs as internal controls for guiding choice of demultiplexing settings, while demonstrating the need to validate demultiplexing precision in biological RNA populations.