Leveraging cell-cell similarity for high-performance spatial and temporal cellular mappings from gene expression data

Integrative single-cell RNA-seq and ATAC-seq analysis of the evolutionary trajectory features of adipose-derived stem cells induced into astrocytes

This study employs single-cell RNA sequencing (scRNA-seq) and assay for transposase-accessible chromatin with high-throughput sequencing technologies (scATAC-seq) to perform joint sequencing on cells at various time points during the induction of adipose-derived stem cells (ADSCs) into astrocytes. We applied bioinformatics approaches to investigate the differentiation trajectories of ADSCs during their induced differentiation into astrocytes. Pseudotemporal analysis was used to infer differentiation trajectories. Additionally, we assessed chromatin accessibility patterns during the differentiation process. Key transcription factors driving the differentiation of ADSCs into astrocytes were identified using motif and footprint methods. Our analysis revealed significant shifts in gene expression during the induction process, with astrocyte-related genes upregulated and stem cell-related genes downregulated. ADSCs first differentiated into neural stem cell-like cells with high plasticity, which further matured into astrocytes via two distinct pathways. Marked changes in chromatin accessibility were observed during ADSC-induced differentiation, affecting transcription regulation and cell function. Transcription factors analysis identified NFIA/B/C/X and CEBPA/B/D as key regulators in ADSCs differentiation into astrocytes. We observed a correlation between chromatin accessibility and gene expression, with ADSCs exhibiting broad chromatin accessibility prior to lineage commitment, where chromatin opening precedes transcription initiation. In summary, we found that ADSCs first enter a neural stem cell-like state before differentiating into astrocytes. ADSCs also display extensive chromatin accessibility prior to astrocyte differentiation, although transcription has not yet been initiated. These findings offer a theoretical framework for understanding the molecular mechanisms underlying this process.

Revealing hidden patterns in deep neural network feature space continuum via manifold learning

Deep neural networks (DNNs) extract thousands to millions of task-specific features during model training for inference and decision-making. While visualizing these features is critical for comprehending the learning process and improving the performance of the DNNs, existing visualization techniques work only for classification tasks. For regressions, the feature points lie on a high dimensional continuum having an inherently complex shape, making a meaningful visualization of the features intractable. Given that the majority of deep learning applications are regression-oriented, developing a conceptual framework and computational method to reliably visualize the regression features is of great significance. Here, we introduce a manifold discovery and analysis (MDA) method for DNN feature visualization, which involves learning the manifold topology associated with the output and target labels of a DNN. MDA leverages the acquired topological information to preserve the local geometry of the feature space manifold and provides insightful visualizations of the DNN features, highlighting the appropriateness, generalizability, and adversarial robustness of a DNN. The performance and advantages of the MDA approach compared to the existing methods are demonstrated in different deep learning applications.