Journal: bioRxiv

Article Title: DOMINO: diffusion-optimised graph learning identifies domain structures with enhanced accuracy and scalability

doi: 10.64898/2025.12.15.694536

Figure Lengend Snippet: a-d . Visualisation of spatial domain identification across representative datasets from different platforms, with ground truth annotations shown alongside results from DOMINO, GraphST, STAGATE, Banksy, and UTAG. Visualisation with only dotted outlines indicates that the method failed to process the ST data. ST data are a Xenium breast cancer, b CosMx non-small cell lung cancer, c Stereoseq mouse embryo and d Visium human dorsolateral prefrontal cortex. e . Radar plots summarising six clustering accuracy metrics across five platforms (CosMx, MERFISH, Visium, Xenium, Stereo-seq), including adjusted Rand index (ARI), normalised mutual information (NMI), adjusted mutual information (AMI), Fowlkes–Mallows index (FMI), inverted normalised variation of information (NVIi), and Purity. f-g . Summarised benchmarking scores combining multiple accuracy metrics (see Methods) across all datasets. f. penalised methods that failed to process the datasets by assigning the lowest score, while g. Ignore all failure events when calculating the score. h . Boundary purity scores, where higher values indicate fewer mixed labels at predicted domain boundaries, reflecting sharper and more distinct boundaries. Sores are normalised to the ground truth labels. i . Boundary consistency scores, where higher values indicate more uniform and stable domain assignments within predicted regions, reflecting greater internal coherence. Sores are normalised to the ground truth labels.

Article Snippet: In cancer research specifically, spatial transcriptomics (ST) has become increasingly important as commercial platforms such as Xenium (10x Genomics), Stereoseq (BGI), and CosMx (NanoString) now enable simultaneous measurement of gene expression and cell position within native tissue architecture from both fresh and archival Formalin Fixed Paraffin Embedded (FFPE) tissues at single-cell resolution – .

Techniques:

Journal: bioRxiv

Article Title: Integrating 12 Spatial and Single Cell Technologies to Characterise Tumour Neighbourhoods and Cellular Interactions in three Skin Cancer Types

doi: 10.1101/2025.07.25.666708

Figure Lengend Snippet: (a) Simplified cross-section of the human epidermis, highlighting squamous cells, melanocytes and basal cells. Coloured regions represent cSCC (green), which originates from squamous cells, melanoma (orange), which originates from melanocytes, and BCC (blue), which originates from basal cells. Two orange melanocytes are shown in the dermal region as occurs in invasive melanoma; other cells in the lower dermis layer are not depicted. (b) Overview of sample design and technologies used to generate data for this project. ROI - region of interest; FOV - field of view; S - cSCC; B - BCC; M - melanoma; HC - healthy (cancer patient); HNC - healthy (non-cancer patient donor). Technologies included are single cell RNA sequencing for fresh samples, single nuclei sequencing for formalin-fixed samples, Visium, Xenium, CosMX, GeoMX DSP for whole transcriptome, GeoMX DSP for proteins, Polaris, RNAscope, the proximal ligation assay, spatial glycomics and CODEX.

Article Snippet: Each biopsy was measured by up to 12 technologies: Chromium single-cell RNA sequencing (scRNASeq), FLEX single nuclei sequencing (snRNAseq), whole transcriptome Visium spatial transcriptomics, single-cell resolution Xenium spatial transcriptomics, NanoString CosMx Spatial Molecular Imaging (CosMx), NanoString GeoMx Digital Spatial Profiling for proteins (Immune-oncology panel, GeoMx), GeoMx cancer transcriptome atlas (GeoMX CTA), Opal Multiplex Polaris protein assay, RNAScope RNA in situ hybridisation, Proximal Ligation Assay (PLA), MALDI-TOF spatial glycomics, and CODEX spatial proteomics ( ).

Techniques: RNA Sequencing, Sequencing, RNAscope, Ligation

Journal: bioRxiv

Article Title: Integrating 12 Spatial and Single Cell Technologies to Characterise Tumour Neighbourhoods and Cellular Interactions in three Skin Cancer Types

doi: 10.1101/2025.07.25.666708

Figure Lengend Snippet: (a) Dot plot showing the percentage of each Level 2 cell type within patient samples. Dots are coloured by cell type category and dot size indicates their percentage within each sample; all columns sum to 100. Results of differential abundance statistical tests are shown to the right, comparing abundance in cSCC vs melanoma, cSCC vs healthy skin, and melanoma vs healthy skin. Asterisks indicate the sample in which the cell type was found to be more abundant, either healthy skin (pink), cSCC-BCC (blue) or melanoma (yellow). (b) A venn diagram of the top significant upregulated genes across cancerous and non-cancerous KCs and melanocytes. (red) Upregulated in cSCC/BCC KC Cancer cells compared to Malignant Melanocytes from melanoma samples, (green) Upregulated in Malignant Melanocytes from melanoma samples compared to cSCC/BCC KC Cancer cells, (yellow) Upregulated in Malignant melanocytes compared to other melanocytes in melanoma samples, (blue) Upregulated in cancer KCs compared to other KCs in cSCC/BCC sample. (c) Heatmaps showing top 50 differentially expressed genes across Cancer vs Normal KCs (top left), Melanocytes vs Melanoma (bottom). Each column of the heatmap indicates a pseudo-bulked pool. (d) Integrative, multiple platform analysis of differentially expressed genes. From left to right, the Venn diagram shows the overlap between DE genes between cSCC cancer KCs vs normal KCs across scRNAseq and for KCs in cancerous tissues compared to those from the normal tissues from non-cancer donors with spatial datasets of Visium, Xenium and CosMX. e) UMAP plot for scRNAseq data showing the expression of SOX2 in cancer vs non-cancer samples, which matches the location of KC cancer cells in UMAP shown in . f) Tissue gene expression plot of CosMX data showing two of the five shared markers SOX2 and LAMP3. Pathological annotation of the region is shown on the left.

Article Snippet: Each biopsy was measured by up to 12 technologies: Chromium single-cell RNA sequencing (scRNASeq), FLEX single nuclei sequencing (snRNAseq), whole transcriptome Visium spatial transcriptomics, single-cell resolution Xenium spatial transcriptomics, NanoString CosMx Spatial Molecular Imaging (CosMx), NanoString GeoMx Digital Spatial Profiling for proteins (Immune-oncology panel, GeoMx), GeoMx cancer transcriptome atlas (GeoMX CTA), Opal Multiplex Polaris protein assay, RNAScope RNA in situ hybridisation, Proximal Ligation Assay (PLA), MALDI-TOF spatial glycomics, and CODEX spatial proteomics ( ).

Techniques: Expressing, Gene Expression

Journal: bioRxiv

Article Title: Integrating 12 Spatial and Single Cell Technologies to Characterise Tumour Neighbourhoods and Cellular Interactions in three Skin Cancer Types

doi: 10.1101/2025.07.25.666708

Figure Lengend Snippet: (a) Cross-modality comparison of the ten communities identified for each of Visium, CosMx and Xenium. The 4 colored bars represent super-communities (or meta-communities), which group the 10 finer communities based on their dominant cell type composition. Each row shows a community identified from one of the three spatial platforms. The left heatmap shows similarity across communities within and between technologies, measured by pairwise Pearson correlation values between communities based on their cell type composition. This allows similar communities across technology platforms and samples to be grouped to form meta-communities. The right heatmap shows the cellular makeup of each community (i.e. proportion of each cell type per community), providing information to label the groups of communities. The central annotation shows the broad classification of communities into immune, KC, stromal or tumour-related communities, based on the cellular makeup of each. (b) Spatial localisation of cells belonging to communities CosMx_6 (left) and Xenium_2 and Xenium_7 (right). Together with Visium_2, these communities form a meta-community that is enriched for melanocytes. (c) Inter-community communication within melanoma CosMx_6. The chord plot visualises cell-cell communication mediated by Collagen signaling pathways, using the CellChat pathway database. Lines connect communicating cell types; line thickness represents greater communication between cell pairs. (d) Ligand-receptor interactions between pairs of cell types within the melanoma community CosMx_6. Top significant L-R pairs and corresponding cell type pairs are shown. (e) Cell type co-occurrence in CosMx samples between melanocytes and either other melanocytes (brown), Treg cells (blue), fibroblasts (green) or other cells (black). Each line plots the co-occurrence score (y-axis) between melanocytes and the test cell type calculated over increasing spatial distances (x-axis). The samples from left to right are melanoma 23346-105P, 30037-07BR and 6475-07FC. (f-g) Cell type proportions of communities identified in Xenium (f) and CODEX (g) for adjacent sections from the same sample (48974-2B). The melanoma community in both datasets is enriched with melanocytes. (h) Joint pathway analysis using upregulated genes or proteins of the melanocyte communities in Xenium and CODEX data (shown in f and g), and highly expressed glycans of the melanocyte community in MALDI data (shown in Fig S12a). The proteins, genes, metabolites are mapped to KEGG metabolic pathways. The X-axis shows the number of genes/proteins from Xenium and CODEX data found in the pathway, while the Y-axis shows glycans in the same pathway.

Article Snippet: Each biopsy was measured by up to 12 technologies: Chromium single-cell RNA sequencing (scRNASeq), FLEX single nuclei sequencing (snRNAseq), whole transcriptome Visium spatial transcriptomics, single-cell resolution Xenium spatial transcriptomics, NanoString CosMx Spatial Molecular Imaging (CosMx), NanoString GeoMx Digital Spatial Profiling for proteins (Immune-oncology panel, GeoMx), GeoMx cancer transcriptome atlas (GeoMX CTA), Opal Multiplex Polaris protein assay, RNAScope RNA in situ hybridisation, Proximal Ligation Assay (PLA), MALDI-TOF spatial glycomics, and CODEX spatial proteomics ( ).

Techniques: Comparison, Protein-Protein interactions

Journal: bioRxiv

Article Title: Integrating 12 Spatial and Single Cell Technologies to Characterise Tumour Neighbourhoods and Cellular Interactions in three Skin Cancer Types

doi: 10.1101/2025.07.25.666708

Figure Lengend Snippet: (a) Exemplar spatial plots showing the LR score for IL34_CSF1R from patient 48974. The black box indicates a region highlighted below the main image. Here, zoomed-in boxes show the IL34_CSF1R LR score (left) and IL34 (middle) and CSF1R (right) gene expression for the same tissue region. (b) Melanoma high resolution spatial transcriptomics samples from STOmics and Curio-Seeker shows cells expressing IL34 and CSF1R. (c-d) Heatmaps indicating grouped GO terms and associated genes that are enriched in IL34_CSF1R-positive spots in melanoma samples compared to IL34_CSF1R-negative spots. GO term groups were calculated by k-means clustering (k = 3) of GO semantic similarity scores; two such groups are shown here. The full heatmap is shown in Fig S9b . (e) Proximal ligation assay (PLA) for validating CD44 interactions in melanoma (top). A merged image of signal for the ligand and the receptor and a zoom-in window highlighting the interaction on the cell membrane. A positive PLA signal is visible if two interacting proteins are in a proximity less than 20 nm. The bottom panels show signals for positive (E-Cadherin-b-Catenin) and negative (CD31-AQP1) controls.

Article Snippet: Each biopsy was measured by up to 12 technologies: Chromium single-cell RNA sequencing (scRNASeq), FLEX single nuclei sequencing (snRNAseq), whole transcriptome Visium spatial transcriptomics, single-cell resolution Xenium spatial transcriptomics, NanoString CosMx Spatial Molecular Imaging (CosMx), NanoString GeoMx Digital Spatial Profiling for proteins (Immune-oncology panel, GeoMx), GeoMx cancer transcriptome atlas (GeoMX CTA), Opal Multiplex Polaris protein assay, RNAScope RNA in situ hybridisation, Proximal Ligation Assay (PLA), MALDI-TOF spatial glycomics, and CODEX spatial proteomics ( ).

Techniques: Gene Expression, Expressing, Ligation, Membrane

Journal: bioRxiv

Article Title: Integrating 12 Spatial and Single Cell Technologies to Characterise Tumour Neighbourhoods and Cellular Interactions in three Skin Cancer Types

doi: 10.1101/2025.07.25.666708

Figure Lengend Snippet: (a) Heatmap of LR scores for LR pairs enriched per cancer type, with a consistent trend across samples and the two Visium and CosMx platforms. Differentially expressed LR pairs were calculated comparing each cancer type vs the others using a pseudobulked LR scores with 3 pools per sample. Each heatmap row is a distinct CosMx or Visium sample. The two L-R pairs specific for melanoma IL34-CSF1R and FGF2-CD44 were used for experimental validations. (b) Differential interaction analysis based on LR pairs and cell type pairs. The Venn diagram compares differential LR pair results between Melanoma and the combined BCC + cSCC datasets, calculated using edgeR with pseudobulked LR scores. The diagram highlights consistent and unique results between CosMX and Visium, where Up indicates a higher LR score in Melanoma and Down indicates a lower score in BCC + cSCC. Cell-to-cell communication between the LR pairs that are up- and downregulated in melanoma in both CosMx and Visium is shown in the two Network plots flanking the Venn diagram. In both Network plots, the purple arrows show pairs of cell types that have interactions higher in Melanoma and green arrows show interactions between cell type pairs more in the BCC + cSCC than in Melanoma. The number displayed for each arrow shows the integrated p-value across all biological replicates (the thicker arrows indicate more interactions). Interactions between the two cell types can still be significantly upregulated in melanoma even if the set of LR pairs were downregulated. (c) Spatial mapping of cancer type-enriched LR pairs in CosMx data. One of the LR pairs that was significantly different between cancer types across technologies in Panel a, namely IL34-CSF1R (higher in melanoma) is shown. It is visualised in FOVs from melanoma sample (top) and BCC sample (bottom). For each cancer type, the cell type annotation of the FOV is shown (top left) with orange and black boxes indicating the highlighted regions (top right). Magnified boxes (top right) show the presence of the ligand (pink) and receptor (red), with white arrows showing the connections between ligands and receptors of nearby cells. An overview of interactions at tissue level is shown by large coloured arrows, representing cumulative interactions between two cell types in the tissue, with the location of the arrow root as the centroid coordinate of all cells in one cell type (bottom left). (d) Melanoma drug target graph integrating multiple biological and pharmacological knowledge types. Nodes represent genes, drugs, and biological functions. Level 1 connections show melanoma-associated genes and drugs targeting melanoma. Level 2 links display drugs targeting the melanoma-associated genes from Level 1 and a broader gene set targeted by drugs in the network. All genes in the graph are either upregulated or have high ligand-receptor scores. Clusters 1, 2, and 3 are pathways enriched with genes shown in the graph.

Article Snippet: Each biopsy was measured by up to 12 technologies: Chromium single-cell RNA sequencing (scRNASeq), FLEX single nuclei sequencing (snRNAseq), whole transcriptome Visium spatial transcriptomics, single-cell resolution Xenium spatial transcriptomics, NanoString CosMx Spatial Molecular Imaging (CosMx), NanoString GeoMx Digital Spatial Profiling for proteins (Immune-oncology panel, GeoMx), GeoMx cancer transcriptome atlas (GeoMX CTA), Opal Multiplex Polaris protein assay, RNAScope RNA in situ hybridisation, Proximal Ligation Assay (PLA), MALDI-TOF spatial glycomics, and CODEX spatial proteomics ( ).

Techniques:

Journal: bioRxiv

Article Title: Integrated spatial morpho-transcriptomics predicts functional traits in pancreatic cancer

doi: 10.1101/2025.03.12.642933

Figure Lengend Snippet: (A) Schematic of Spatial Morphology and RNA Transcript (SMART) analysis framework. (B) Representative images of cell lines profiled with phase holotomography following 48 hours of KRAS inhibitor (RMC-6236) treatment and at baseline conditions. Scale bar = 20 µm. Panc0203 MR refers to the Panc0203 parental cell line grown to resistance to MRTX1133 in increasing drug concentration . (C) Quantification of cell object area and density throughout the treatment course with RMC-6236, stratified by RMC-6236 sensitivity. (D) Holotomography and clustering analysis of an additional KRAS inhibitor sensitive cell line, AsPC1, in KRAS inhibitor treated and untreated conditions. (E) Quantification of cell death of AsPC1 cells in treated and untreated conditions. Cycle length is 20 minutes. (F-G) Quantification of AsPC1 cells in clusters versus singlets in KRAS inhibitor treated and untreated conditions.

Article Snippet: We pioneered a new single-cell framework, Spatial Morphology and RNA Transcript Analysis (SMART), to integrate cell morphology and transcriptomic state using a combination of holotomography (Nanolive 3D Cell Explorer 96focus), CellPainting , and spatial molecular imaging (SMI; Bruker/Nanostring CosMx) ( ).

Techniques: Concentration Assay

Journal: bioRxiv

Article Title: Synthetic rescue of XPC phenotype via PIK3C3 downregulation

doi: 10.1101/2023.08.08.552431

Figure Lengend Snippet: The expression of XPC in both cell lines was examined as well as the cells’ photosensitivity and repair capacity. a) XPC expression. RNA and protein extraction was carried out for both XP-C and WT cells to analyze the expression of XPC. RT-PCR quantification of XPC mRNA revealed a highly significant five-fold decrease in XPC mRNA in XP-C cells compared to WT (p-value<0.0001). At the protein level, western blot analysis revealed the total absence of XPC protein expression in XP-C cells unlike the WT cells (p-value<0.001) unpaired t test. b) Viability of fibroblasts 24hrs post UVB. XP-C cells manifest significantly increased photosensitivity compared to WT cells. Both XP-C and WT cells were seeded in 96-well plates to be irradiated at 80% confluency with increasing UVB doses then their viability was quantified 24hours later by the incubation with PrestoBlue. XP-C cells show a sharper significant decrease in viability as a function of increased UVB dose compared to WT cells. Viability was calculated by means of percent of control with 100% control being non-irradiated cells. * p< 0.05, unpaired t-test. c) 6-4PP repair in Normal and XP-C cells. 6-4 PP quantification was conducted on irradiated or non-irradiated WT and XP-C cells. XP-C cells show elevated levels of DNA damage 24hrs post UV while normal cells show significant repair at the same time point. Single cell analysis was carried out via the quantification of nuclear DNA damage in several individual cells per condition. Hoechst staining was utilized for the identification of the nuclei that will be set as the region of interest (ROI) for the quantification of DNA damage. DNA damage at time zero was set as 100% while that of non-irradiated cells was set as 0% damage. *** p<0.001 paired t test.

Article Snippet: Post saturation the cells were incubated overnight with 6-4PP antibody (Cosmo Bio, California, USA).

Techniques: Expressing, Protein Extraction, Reverse Transcription Polymerase Chain Reaction, Western Blot, Irradiation, Incubation, Single-cell Analysis, Staining

Journal: bioRxiv

Article Title: Synthetic rescue of XPC phenotype via PIK3C3 downregulation

doi: 10.1101/2023.08.08.552431

Figure Lengend Snippet: XP-C-KO and WT keratinocytes cells were transfected with either siLATS1, siPIK3C3 or siAS then subjected to increased UVB doses. Their viability and DNA damage was further on assessed post UV by the incubation with PrestoBlue or staining with 6-4PP DNA damage antibody. a) PIK3C3 siRNAs showed UV-protection in XP-C cells viability compared to siAS (p-value<0.05). WT cells UVB-protection was evident in cells with LATS1 knockdown (p-value<0.05) while siPIK3C3 transfection had no effect. Cell survival was calculated by normalizing the fluorescence intensity of Presto Blue into a RZscore. b) The knock down of both kinases enabled a significant decrease in DNA damage in XP-C keratinocytes compared to siAS cells. * p-value <0.05, *** p-value <0.001. Student T test.

Article Snippet: Post saturation the cells were incubated overnight with 6-4PP antibody (Cosmo Bio, California, USA).

Techniques: Transfection, Incubation, Staining, Fluorescence