OBJECTIVE: To identify OA subtypes based on cartilage transcriptomic data in cartilage tissue and characterize their underlying pathophysiological processes and/or clinically relevant characteristics. METHODS: This study includes n = 66 primary OA patients (41 knees and 25 hips), who underwent a joint replacement surgery, from which macroscopically unaffected (preserved, n = 56) and lesioned (n = 45) OA articular cartilage were collected [Research Arthritis and Articular Cartilage (RAAK) study]. Unsupervised hierarchical clustering analysis on preserved cartilage transcriptome followed by clinical data integration was performed. Protein-protein interaction (PPI) followed by pathway enrichment analysis were done for genes significant differentially expressed between subgroups with interactions in the PPI network. RESULTS: Analysis of preserved samples (n = 56) resulted in two OA subtypes with n = 41 (cluster A) and n = 15 (cluster B) patients. The transcriptomic profile of cluster B cartilage, relative to cluster A (DE-AB genes) showed among others a pronounced upregulation of multiple genes involved in chemokine pathways. Nevertheless, upon investigating the OA pathophysiology in cluster B patients as reflected by differentially expressed genes between preserved and lesioned OA cartilage (DE-OA-B genes), the chemokine genes were significantly downregulated with OA pathophysiology. Upon integrating radiographic OA data, we showed that the OA phenotype among cluster B patients, relative to cluster A, may be characterized by higher joint space narrowing (JSN) scores and low osteophyte (OP) scores. CONCLUSION: Based on whole-transcriptome profiling, we identified two robust OA subtypes characterized by unique OA, pathophysiological processes in cartilage as well as a clinical phenotype. We advocate that further characterization, confirmation and clinical data integration is a prerequisite to allow for development of treatments towards personalized care with concurrently more effective treatment response.

Background: Single-cell transcriptomics is rapidly advancing our understanding of the cellular composition of complex tissues and organisms. A major limitation in most analysis pipelines is the reliance on manual annotations to determine cell identities, which are time-consuming and irreproducible. The exponential growth in the number of cells and samples has prompted the adaptation and development of supervised classification methods for automatic cell identification. Results: Here, we benchmarked 22 classification methods that automatically assign cell identities including single-cell-specific and general-purpose classifiers. The performance of the methods is evaluated using 27 publicly available single-cell RNA sequencing datasets of different sizes, technologies, species, and levels of complexity. We use 2 experimental setups to evaluate the performance of each method for within dataset predictions (intra-dataset) and across datasets (inter-dataset) based on accuracy, percentage of unclassified cells, and computation time. We further evaluate the methods' sensitivity to the input features, number of cells per population, and their performance across different annotation levels and datasets. We find that most classifiers perform well on a variety of datasets with decreased accuracy for complex datasets with overlapping classes or deep annotations. The general-purpose support vector machine classifier has overall the best performance across the different experiments. Conclusions: We present a comprehensive evaluation of automatic cell identification methods for single-cell RNA sequencing data. All the code used for the evaluation is available on GitHub (https://github.com/tabdelaal/scRNAseq-Benchmark). Additionally, we provide a Snakemake workflow to facilitate the benchmarking and to support the extension of new methods and new datasets.