Biology subjects

Nikolski, M.

Publications and source records attributed to Nikolski, M..

13 recordsLinked to original sources

bioRxiv2026

SpaceBlocks: an expert-supervised workflow for cohort-level spatial transcriptomics analysis

Spatial transcriptomics (ST) has become a central tool in tissue biology, with multiple technologies addressing the issue from a different angle. Each platform imposes its own raw data format conventions, and few workflows address reproducible, scalable, cohort-level analyses while integrating expert-supervised decisions to strengthen computational and biological provenance. In this work, we present SpaceBlocks, a Snakemake workflow for reproducible cohort-level analysis of single-cell resolution-ST data that prioritizes expert supervision over full automation. SpaceBlocks homogenizes heterogeneous platform outputs into a common data format, whose structure is validated to ensure downstream processing. Comprehensive expert supervision is facilitated through per sample quality control and cell type annotation diagnostic visualizations, after which downstream analyses automatically follow. SpaceBlocks also allows the ingestion of technology-agnostic externally formatted data and provides interoperability with existing tools and technologies. We demonstrate the use of SpaceBlocks through the reanalysis of 37 samples from public Visium HD, Xenium 5K and MERSCOPE datasets. SpaceBlocks exhibited processing scalability across the three platforms. This reanalysis identified previously missed cell types in Visium HD and heterogeneous marker gene expression in MERSCOPE, highlighting the need for expert-supervised annotations. Moreover, SpaceBlocks facilitated the identification of differentially expressed genes associated with previously defined multiple sclerosis (MS) spatial domains in MERSCOPE, and transcriptional differences between healthy and malignant colorectal tissues in Visium HD. Together, these findings establish SpaceBlocks as a flexible and reproducible workflow for expert-supervised spatial transcriptomics analysis across technologies and biological cohorts.

bioinformatics↗

bioRxiv2026

SpacePath enables pathway and molecular-family analysis beyond annotation ambiguity in spatial metabolomics and lipidomics

Mass spectrometry imaging (MSI) maps metabolites and lipids in tissue with spatial resolution, but molecular identity remains partly resolved: single ions match multiple candidate compounds, and most detected metabolites and lipids fall outside canonical biochemical pathways. We developed SpacePath, a computational framework that retains ion-to-compound ambiguity, scores canonical pathways across tissue locations, and groups assignments that were unmapped to pathways into de novo molecular families by chemical similarity. Pathway mappings covered on average only 1% of molecular assignments, whereas the combined SpacePath representation covered 24% of pooled assignments. Benchmarking on four public spatial lipidomics datasets (healthy mouse brain, cerebellum, a demyelination model, and human bladder cancer) recovered previously reported spatial patterns. In newly acquired glioblastoma xenograft datasets, SpacePath identified lipid programs specific to tumor and surrounding brain, and SpiderMass-MS/MS independently confirmed four tumor-associated features from a de novo family, refining the initial MS1 assignments. SpacePath extends the spatial interpretation of MSI data beyond canonical pathways while preserving annotation uncertainty.

bioinformatics↗

bioRxiv2026

Microbiota-gut-brain axis modulation drives glioblastoma progression and therapy resistance

Glioblastoma is a highly aggressive brain tumour with poor prognosis, whose aetiology, progression, and therapeutic resistance remain incompletely understood. While the microbiota-gut-brain axis has emerged as a key regulator of neurological disorders, its role in glioblastoma biology and treatment response is still largely unexplored. Using a clinically relevant immunocompetent murine model combining glioblastoma stem cell implantation, dextran sodium sulfate-induced gut inflammation, and a full Stupp-like therapeutic protocol, we investigated bidirectional gut-brain communication in glioblastoma. Tumour growth and recurrence were monitored by bioluminescence imaging, tumour transcriptomic profiles were analysed by RNA sequencing, and brain and colon tissues were subjected to histological and molecular analyses. Gut microbiota composition was assessed by 16S rRNA sequencing, while systemic metabolites and cytokines were quantified in plasma. Cross-compartment association bioinformatic analyses were performed to correlate multi-organ readouts. Gut inflammation enhanced glioblastoma growth and promoted tumour recurrence following therapy. Tumour progression was associated with increased infiltration of immunosuppressive macrophages, whereas recurrence correlated with elevated oxidative DNA damage. Remarkably, glioblastoma exerted systemic immunomodulatory effects, attenuating intestinal and systemic inflammatory responses, and induced profound remodelling of gut microbiota composition and predicted metabolic function, including enrichment of Akkermansia and depletion of Lactobacillus. Systemic metabolic profiling was investigated as a route of communication within the gut-brain axis and revealed adaptations in DSS-treated mice associated with tumour burden and therapeutic response. Multi-compartment correlation and multivariable association analyses identified specific bacterial genera and circulating metabolites associated with tumour volume, intestinal inflammation, and genomic instability. These findings uncover a dynamic, bidirectional microbiota-gut-brain axis in glioblastoma and identify intestinal inflammation as a critical determinant of tumour progression and therapeutic outcome. Targeting gut disturbances and microbiota-associated metabolic pathways may represent novel strategies to modulate glioblastoma aggressiveness and treatment response.

cancer biology↗

bioRxiv2026

Spatial and Bulk Transcriptomic Profiling Defines the Molecular Evolution of Cutaneous Squamous Cell Carcinoma and Reveals Stage-Specific Biomarkers of Clinical Relevance

Cutaneous squamous cell carcinoma (cSCC) is a common skin cancer associated with substantial morbidity and mortality in advanced stages. Despite its well-described stepwise progression from actinic keratosis to invasive disease, robust molecular markers for stage discrimination and clinical decision-making remain limited. We sought to define the transcriptional continuum underlying cSCC progression, identify stage-associated biomarkers, and assess the broader relevance of these programs across human malignancies. Bulk RNA sequencing (HTG EdgeSeq) and spatial transcriptomics (GeoMx) were performed on biopsies from eight patients, each presenting multiple disease stages (healthy skin, premalignant lesion, tumor core, and invasive front) within the same lesion field, enabling within-patient analysis of progression. Spatial transcriptomic analyses identified more than 2,000 differentially expressed genes whose expression varied across disease stages. These genes were organized into 18 coordinated expression programs reflecting progressive biological rewiring during tumor evolution. Proliferation, extracellular matrix remodeling, inflammation, and stress-response pathways were progressively upregulated, whereas epithelial differentiation and metabolic processes, including lipid and amino acid metabolism, were downregulated. Macrophages exhibited distinct metabolic reprogramming, with increased purine metabolism, glycolysis, and pyruvate metabolism across progression. To evaluate the broader clinical relevance of these progression-associated programs, we developed a reproducible Snakemake pipeline to systematically screen 32 solid and hematologic malignancies from The Cancer Genome Atlas (TCGA). A combined cSCC-progression signature was significantly associated with poor overall survival (P < 0.05) in 10 additional cancer types. Finally, we identified 12 stage-informative biomarkers, whose spatially restricted expression patterns were validated using Visium HD. This study provides a spatially resolved and stage-aware transcriptomic map of cSCC progression, identifies coordinated gene programs underlying disease evolution, and defines progression-associated signatures with prognostic relevance across multiple cancers, highlighting their potential translational value.

cancer biology↗

bioRxiv2026

Uncertainty-aware benchmarking reveals ambiguous transcripts in mRNA-lncRNA classification

BackgroundLong non-coding RNAs (lncRNAs) have gained significant attention in recent years, yet distinguishing them from protein-coding transcripts remains challenging. Indeed, many lncRNAs share mRNA-like processing and existing sequence-derived signals do not fully capture the coding/non-coding boundary. Recent GENCODE annotation efforts revealed tens of thousands of novel lncRNA sequences as well as the reclassification of some lncRNAs into the protein-coding class, highlighting the need to better characterize transcript features associated with classification uncertainty and errors. ResultsWe performed uncertainty-aware benchmarking by retraining and evaluating eight transcript classifiers under a controlled protocol on a label-stable GENCODE v46-v47 subset. Beyond conventional model evaluation metrics, we quantified inter-tool agreement and entropy-based uncertainty to stratify transcripts into consensus, discordant, and consensus-error groups. To expand standard sequence and ORF-derived signals, we incorporated repeat-derived features from mature transcripts and non-B DNA motif features across gene bodies. Although aggregate performance was high, [~]45% of transcripts showed inter-tool discordance, particularly among lncRNAs. Feature analyses linked low-uncertainty predictions to strong coding-like signals, whereas high-uncertainty profiles exhibited mixed signatures. Alongside classical predictors in global importance analyses, repeat-derived features appear as main contributors. ConclusionsBy combining controlled benchmarking with transcript-level agreement and uncertainty stratification, together with extended feature profiling, we identified patterns associated with classifier disagreement and misclassification. This novel framework provides practical guidance for interpreting predictions, motivating the development of more robust coding/non-coding classifiers, while also shedding light on the sequence properties that distinguish lncRNA sequences.

bioinformatics↗

bioRxiv2024

SeuratIntegrate: an R package to facilitate the use of integration methods with Seurat

MotivationIntegrating multiple datasets has become an increasingly common task in scRNA-seq analysis. The advent of single-cell atlases adds further complexity to this task, as they often involve combining data with complex, nested batch effects - such as those arising from multiple studies, organs or disease states. Accurate data integration is essential to distinguish cell types with sufficient granularity, thereby reflecting true biological patterns, and to create reliable reference datasets for the community. In this context, the latest version of Seurat (v5) introduced a multi-layered object structure to facilitate the integration of scRNA-seq datasets in a unified manner. However, the panel of available batch-correction methods remains limited to five algorithms within Seurat, restricting users from accessing a broader diversity of available tools, particularly Python-based methods. Furthermore, no existing R tool assists the user in making an informed decision in selecting the most appropriate integration approach. ResultsTo overcome these challenges, we developed SeuratIntegrate, an open source R package that extends Seurats functionality. SeuratIntegrate supports eight integration methods, incorporating both R- and Python-based tools, and enables performance evaluation of integration through several scoring methods. This functionality allows for a more versatile and informed integration process. AvailabilitySeuratIntegrate is available at https://github.com/cbib/Seurat-Integrate/. The package is released under the MIT License.

bioinformatics↗

bioRxiv2024

Exploiting a metabolic vulnerability in brain tumour stem cells using a brain-penetrant drug with safe profile

Glioblastoma (GB) remains one of the most treatment refractory and fatal tumour in humans. GB contains a population of self-renewing stem cells, the brain tumour stem cells (BTSC) that are highly resistant to therapy and are at the origin of tumour relapse. Here, we report, for the first time, that mubritinib potently impairs stemness and growth of patient-derived BTSCs harboring different oncogenic mutations. Mechanistically, by employing bioenergetic assays and rescue experiments, we provide compelling evidence that mubritinib acts on complex I of the electron transport chain to impair BTSC stemness pathways, self-renewal and proliferation. Global gene expression profiling revealed that mubritinib alters the proliferative, neural-progenitor-like, and the cell-cycling state signatures. We employed in vivo pharmacokinetic assays to establish that mubritinib crosses the blood-brain barrier. Using preclinical models of patient-derived and syngeneic murine orthotopic xenografts, we demonstrated that mubritinib delays GB tumourigenesis, and expands lifespan of animals. Interestingly, its combination with radiotherapy offers survival advantage to animals. Strikingly, thorough toxicological and behavioral studies in mice revealed that mubritinib does not induce any damage to normal cells and has a well-tolerated and safe profile. Our work warrants further exploration of this drug in in-human clinical trials for better management of GB tumours.

cancer biology↗

bioRxiv2022

TREM1+ regulatory myeloid cells expand in steatohepatitis-HCC and associate with poor prognosis and therapeutic resistance to anti-PD-1 blockade

Hepatocellular carcinoma (HCC) is an inflammation-associated cancer arising from viral and non-viral etiologies. Immune checkpoint blockade primarily benefits patients with viral HCC. Expansion of suppressive myeloid cells is a hallmark of chronic inflammation and cancer, but their heterogeneity in HCC is not fully resolved and might underlie immunotherapy resistance in the steatohepatitis setting. Here, we present a high resolution atlas of hepatic innate immune cells from patients with HCC that unravels a steatohepatitis contexture characterized by the emergence of high entropy myeloid cell states and myeloid-biased NK cell differentiation. We identify a discrete population of tumor-infiltrating myeloid cells, predominant in the steatohepatitis setting, that expresses a variety of myeloid lineage-affiliated genes, including granulocyte, macrophage and dendritic cell features, and can be identified in HCC tumors based on selective dual expression of TREM1 and CD163. Functional characterization reveals that TREM1+ CD163+ myeloid cells highly express TGF{beta} and IL-13RA, localize to HCC fibrotic lesions, and potently suppress T cell effector functions ex vivo, a function further potentiated by TREM1 engagement. We refer to this population as TREM1+ CD163+ regulatory myeloid cells (TREM1+CD163+ Mreg). Deconvolution analyses in large cohorts of patients with HCC and other solid tumors reveals that the density of TREM1+ CD163+ Mreg increases in advanced stages, associates with poor prognosis, and therapeutic resistance to PD-1 blockade. Our data support myeloid subset-targeted immunotherapies to treat HCC and identify TREM1 as a therapeutic target. HIGHLIGHTSO_LIAtlas of hepatic innate immune cells (100,000 transcriptomes) from patients with HCC C_LIO_LICore signatures to identify, discriminate and localize innate lymphoid and myeloid cells C_LIO_LIA population of TREM1+CD163+ myeloid cells, referred to as TREM1+CD163+ Mreg, expands in steatohepatitis HCC C_LIO_LITREM1+CD163+ Mreg express granulocyte- and macrophage/dendritic cell-lineage genes C_LIO_LITREM1+CD163+ Mreg potently suppress T cell effector functions, which is potentiated by TREM1 engagement by cognate ligands C_LIO_LITREM1+CD163+ Mreg produce high levels of TGF{beta} and populate fibrotic lesions C_LIO_LIThe density of TREM1+CD163+ Mreg increases in advanced HCC and associate with poor patient survival C_LIO_LIThe density of TREM1+CD163+ Mreg associates with resistance to immune checkpoint blockade in other solid tumors C_LI

immunology↗

bioRxiv2022

SalienceNet: an unsupervised Image-to-Image translation method for nuclei saliency enhancement in microscopy images

Automatic segmentation of nuclei in low-light microscopy images remains a difficult task, especially for high-throughput experiments where need for automation is strong. Low saliency of nuclei with respect to the background, variability of their intensity together with low signal-to-noise ratio in these images constitute a major challenge for mainstream algorithms of nuclei segmentation. In this work we introduce SalienceNet, an unsupervised deep learning-based method that uses the style transfer properties of cycleGAN to transform low saliency images into high saliency images, thus enabling accurate segmentation by downstream analysis methods, and that without need for any parameter tuning. We have acquired a novel dataset of organoid images with soSPIM, a microscopy technique that enables the acquisition of images in low-light conditions. Our experiments show that SalienceNet increased the saliency of these images up to the desired level. Moreover, we evaluated the impact of SalienceNet on segmentation for both Otsu thresholding and StarDist and have shown that enhancing nuclei with SalienceNet improved segmentation results using Otsu thresholding by 30% and using StarDist by 26% in terms of IOU when compared to segmentation of non-enhanced images. Together these results show that SalienceNet can be used as a common preprocessing step to automate nuclei segmentation pipelines for low-light microscopy images.

bioinformatics↗

bioRxiv2021

DeepSpot: a deep neural network for RNA spot enhancement in smFISH microscopy images

Detection of RNA spots in single molecule FISH microscopy images remains a difficult task especially when applied to large volumes of data. The small size of RNA spots combined with high noise level of images often requires a manual adaptation of the spot detection thresholds for each image. In this work we introduce DeepSpot, a Deep Learning based tool specifically designed to enhance RNA spots which enables spot detection without need to resort to image per image parameter tuning. We show how our method can enable the downstream accurate detection of spots. The architecture of DeepSpot is inspired by small object detection approaches. It incorporates dilated convolutions into a module specifically designed for the Context Aggregation for Small Object (CASO) and uses Residual Convolutions to propagate this information along the network. This enables DeepSpot to enhance all RNA spots to the same intensity and thus circumvents the need for parameter tuning. We evaluated how easily spots can be detected in images enhanced by our method, by training DeepSpot on 20 simulated and 1 experimental datasets, and have shown that more than 97% accuracy is achieved. Moreover, comparison with alternative deep learning approaches for mRNA spot detection (deepBlink) indicated that DeepSpot allows more precise mRNA detection. In addition, we generated smFISH images from mouse fibroblasts in a wound healing assay to evaluate whether DeepSpot enhancement can enable seamless mRNA spot detection and thus streamline studies of localized mRNA expression in cells.

bioinformatics↗

bioRxiv2020

FluoSim: simulator of single molecule dynamics for fluorescence live-cell and super-resolution imaging of membrane proteins

Fluorescence live-cell and super-resolution microscopy methods have considerably advanced our understanding of the dynamics and mesoscale organization of macro-molecular complexes that drive cellular functions. However, different imaging techniques can provide quite disparate information about protein motion and organization, owing to their respective experimental ranges and limitations. To address these limitations, we present here a unified computer program that allows one to model and predict membrane protein dynamics at the ensemble and single molecule level, so as to reconcile imaging paradigms and quantitatively characterize protein behavior in complex cellular environments. FluoSim is an interactive real-time simulator of protein dynamics for live-cell imaging methods including SPT, FRAP, PAF, and FCS, and super-resolution imaging techniques such as PALM, dSTORM, and uPAINT. The software, thoroughly validated against experimental data on the canonical neurexin-neuroligin adhesion complex, integrates diffusion coefficients, binding rates, and fluorophore photo-physics to calculate in real time the distribution of thousands of independent molecules in 2D cellular geometries, providing simulated data of protein dynamics and localization directly comparable to actual experiments.

biophysics↗

bioRxiv2020

A multi-layered systems approach for renal cell carcinoma

Renal cell carcinoma (RCC) still lacks prognostic and predictive biomarkers to monitor the disease and the response to therapy. The usual strategy in translational research is to start from human samples, to identify molecular markers and gene networks and then to functionally validate them in vitro and in animal models. We devised herein a completely opposite strategy from "mouse to man" by performing an aggressiveness screen and used functional genomics, imaging, clinical data and computational approaches in order to discover molecular pathways and players in renal cancer development and metastasis. Multiple cell lines for primary tumor growth, survival in the blood circulation and lung metastasis or metastatic spread from the primary tumor were generated and analyzed using a multi-layered approach which includes large-scale transcriptome, genome and methylome analyses. Transcriptome and methylome analyses demonstrated distinct clustering in three different groups. Remarkably, DNA sequencing did not show significant genomic variations in the different groups which indicates absence of clonal selection during the in vivo amplification process. Transcriptome analysis revealed distinct signatures of tumor aggressiveness which were validated in patient cohorts. Methylome analysis of full-length DNA allowed clustering of the same groups and revealed clinically relevant signatures. Furthermore, we identified SAA2 and CFB as soluble prognostic and predictive biomarkers of the therapeutic response. We also uncovered IL34 as another soluble prognostic biomarker and key regulator of renal cell carcinoma (RCC) progression. This was also functionally validated in vivo, and a mathematical model of IL34-dependent primary tumor growth and metastasis development was provided. These results indicate that such multilayered analysis in a RCC animal model leads to meaningful results that are of translational significance. One Sentence SummaryAn aggressiveness screen with multilayer systems analysis to identify signatures and biomarkers for renal cell carcinoma aggressiveness.

cancer biology↗

bioRxiv2019

DypFISH: Dynamic Patterned FISH to Interrogate RNA and Protein Spatial and Temporal Subcellular Distribution

Advances in single cell RNA sequencing have allowed for the identification and characterization of cellular subtypes based on quantification of the number of transcripts in each cell. However, cells may differ not only in the number of mRNA transcripts that they exhibit, but also in their spatial and temporal distribution, intrinsic to the definition of their cellular state. Here we describe DypFISH, an approach to quantitatively investigate the spatial and temporal subcellular localization of RNA and protein, by combining micropatterning of cells with fluorescence microscopy at high resolution. We introduce a range of analytical techniques for quantitatively interrogating single molecule RNA FISH data in combination with protein immunolabeling over time. Strikingly, our results show that constraining cellular architecture reduces variation in subcellular mRNA and protein distributions, allowing the characterization of their localization and dynamics with high reproducibility. Many tissues contain cells that exist in similar constrained architectures. Thus DypFISH reveals reproducible patterns of clustering, strong correlative influences of mRNA-protein localization on MTOC orientation when they are present and interdependent dynamics globally and at specific subcellular locations which can be extended to physiological systems.

systems biology↗