Prior knowledge-guided multilevel graph neural network for tumor risk prediction and interpretation via multi-omics data integration.

The interrelation and complementary nature of multi-omics data can provide valuable insights into the intricate molecular mechanisms underlying diseases. However, challenges such as limited sample size, high data dimensionality and differences in omics modalities pose significant obstacles to fully harnessing the potential of these data. The prior knowledge such as gene regulatory network and pathway information harbors useful gene-gene interaction and gene functional module information. To effectively integrate multi-omics data and make full use of the prior knowledge, here, we propose a Multilevel-graph neural network (GNN): a hierarchically designed deep learning algorithm that sequentially leverages multi-omics data, gene regulatory networks and pathway information to extract features and enhance accuracy in predicting survival risk. Our method achieved better accuracy compared with existing methods. Furthermore, key factors nonlinearly associated with the tumor pathogenesis are prioritized by employing two interpretation algorithms (i.e. GNN-Explainer and IGscore) for neural networks, at gene and pathway level, respectively. The top genes and pathways exhibit strong associations with disease in survival analyses, many of which such as SEC61G and CYP27B1 are previously reported in the literature.

AllenDigger, a Tool for Spatial Expression Data Visualization, Spatial Heterogeneity Delineation, and Single-Cell Registration Based on the Allen Brain Atlas.

Spatial transcriptomics can be used to capture cellular spatial organization and has facilitated new insights into different biological contexts, including developmental biology, cancer, and neuroscience. However, its wide application is still hindered by its technical challenges and immature data analysis methods. Allen Brain Atlas (ABA) provides a great source for spatial gene expression throughout the mouse brain at various developmental stages with in situ hybridization image data. To the best of our knowledge, the portal developed to access spatial expression data is not very useful to biologists. Here, we developed a toolkit to collect and preprocess expression data from the ABA and allow a friendlier query to visualize the spatial distribution of genes of interest, characterize the spatial heterogeneity of the brain, and register cells from single-cell transcriptomics data to fine anatomical brain regions via machine learning methods with high accuracy. AllenDigger will be very helpful to the community in precise spatial gene expression queries and add extra spatial information to further interpret the scRNA-seq data in a cost-effective manner.

Comprehensive analyses of B-cell compartments across the human body reveal novel subsets and a gut-resident memory phenotype.

Although human B cells have been extensively studied, most reports have used peripheral blood as a source. Here, we used a unique tissue resource derived from healthy organ donors to deeply characterize human B-cell compartments across multiple tissues and donors. These datasets revealed that B cells in the blood are not in homeostasis with compartments in other tissues. We found striking donor-to-donor variability in the frequencies and isotype of CD27+ memory B cells (MBCs). A comprehensive antibody-based screen revealed markers of MBC and allowed identification of novel MBC subsets with distinct functions defined according to surface expression of CD69 and CD45RB. We defined a tissue-resident MBC phenotype that was predominant in the gut but absent in blood. RNA-sequencing of MBC subsets from multiple tissues revealed a tissue-resident MBC gene signature as well as gut- and spleen-specific signatures. Overall, these studies provide novel insights into the nature and function of human B-cell compartments across multiple tissues.

Tissue Determinants of Human NK Cell Development, Function, and Residence.

Immune responses in diverse tissue sites are critical for protective immunity and homeostasis. Here, we investigate how tissue localization regulates the development and function of human natural killer (NK) cells, innate lymphocytes important for anti-viral and tumor immunity. Integrating high-dimensional analysis of NK cells from blood, lymphoid organs, and mucosal tissue sites from 60 individuals, we identify tissue-specific patterns of NK cell subset distribution, maturation, and function maintained across age and between individuals. Mature and terminally differentiated NK cells with enhanced effector function predominate in blood, bone marrow, spleen, and lungs and exhibit shared transcriptional programs across sites. By contrast, precursor and immature NK cells with reduced effector capacity populate lymph nodes and intestines and exhibit tissue-resident signatures and site-specific adaptations. Together, our results reveal anatomic control of NK cell development and maintenance as tissue-resident populations, whereas mature, terminally differentiated subsets mediate immunosurveillance through diverse peripheral sites. VIDEO ABSTRACT.

High Dimensional Functionomic Analysis of Human Hematopoietic Stem and Progenitor Cells at a Single Cell Level.

The ability to conduct investigation of cellular transcription, signaling, and function at the single-cell level has opened opportunities to examine heterogeneous populations at unprecedented resolutions. Although methods have been developed to evaluate high-dimensional transcriptomic and proteomic data (relating to cellular mRNA and protein), there has not been a method to evaluate corresponding high-dimensional functionomic data (relating to cellular functions) from single cells. Here, we present a protocol to quantitatively measure the differentiation potentials of single human hematopoietic stem and progenitor cells, and then cluster the cells according to these measurements. High dimensional functionomic analysis of cell potential allows cell function to be linked to molecular mechanisms within the same progenitor population.

Lineage specification of human dendritic cells is marked by IRF8 expression in hematopoietic stem cells and multipotent progenitors.

The origin and specification of human dendritic cells (DCs) have not been investigated at the clonal level. Through the use of clonal assays, combined with statistical computation, to quantify the yield of granulocytes, monocytes, lymphocytes and three subsets of DCs from single human CD34+ progenitor cells, we found that specification to the DC lineage occurred in parallel with specification of hematopoietic stem cells (HSCs) to the myeloid and lymphoid lineages. This started as a lineage bias defined by specific transcriptional programs that correlated with the combinatorial 'dose' of the transcription factors IRF8 and PU.1, which was transmitted to most progeny cells and was reinforced by upregulation of IRF8 expression driven by the hematopoietic cytokine FLT3L during cell division. We propose a model in which specification to the DC lineage is driven by parallel and inheritable transcriptional programs in HSCs and is reinforced over cell division by recursive interactions between transcriptional programs and extrinsic signals.