10x genomics single cell transcriptomics  (10X Genomics)

Figure 1 adapted from Tyson (1991), Gustafsson et al. (2023), Park et al. (2016), Moss et al. (2021). Created using biorender.com . " title="... in annotated collections and compendiums such as GEO (transcriptomics), ArrayExpress (transcriptomics), MetaboLights (metabolomics), and PRIDE (proteomics). This ..." property="contentUrl" width="100%" height="100%"/>
Figure Legend Snippet: Data sources for advancing computational modeling “Not So Big” and “Big” data sources can be utilized collectively to build computational models of varying complexity. “Not So Big” data are usually disparate, require extensive data collation, and must be obtained individually from various literature sources in the form of data tables present in the supplementary material. The “Not So Big” data are derived from targeted and focused experiments and provides tissue-level detail for mechanistic models such as the first cell cycle gene regulatory network from Tyson (1991). Bulk and single-cell “Big” data are derived from targeted and unbiased assays, and are usually stored in annotated collections and compendiums such as GEO (transcriptomics), ArrayExpress (transcriptomics), MetaboLights (metabolomics), and PRIDE (proteomics). This “Big” data provides genome-scale detail for informing correlation networks and genome scale metabolic models. Single-cell “Big” data from reference atlases, including The Cancer Genome Atlas, the Human Cell Atlas, HuBMAP, and Tabula Sapiens , provide untargeted and unbiased assays at the whole-body physiological scale. These data can be utilized to inform future virtual human models at various scales, including the molecular level (i.e., Wnt/B-Catenin signaling pathway), single-cell level (i.e., gene correlation networks), and physiological level (i.e., multi-organ interactions). Figure 1 adapted from Tyson (1991), Gustafsson et al. (2023), Park et al. (2016), Moss et al. (2021). Created using biorender.com .

Techniques Used: Derivative Assay

Figure Legend Snippet: Molecularly targeted methods for single-cell and spatial transcriptomics

Techniques Used:

Figure Legend Snippet: Examples of transcriptome-proteome multiomics technologies

Techniques Used: Single-cell Isolation, RNA Detection, Reverse Transcription, Mass Cytometry, Staining

Figure Legend Snippet: Computational models informed by experimental data The components, interactions, correlations, and patterns extracted from “Big” data (multi-omics data including transcriptomics, proteomics, metabolomics, and spatial omics) and the components, interactions and mechanisms extracted from “Not So Big” data (i.e., western blots, immuno-staining, and qPCR) can be utilized to generate and inform molecular signaling networks, putative cellular networks, and gene regulatory networks. For instance, while the MAP kinase pathway was discovered using “Not So Big” data sources (solid line) many “Big” data sources (dashed line) have confirmed and further explained and complemented these initial findings. Similarly, while gene regulatory networks have been mainly developed using “Big” data (solid line), “Not So Big” data (dashed line) can also be informative when generating such networks. For example, Park et al., (2016) modeled neurons during the circadian cycle. First, five neuronal groups were identified according to their unique transcriptional landscapes with marker genes shown for each of the groups. A gene regulatory network was then developed based on the major molecular interactions between key neuropeptides (VIP, AVP, PROK2, and PACAP) and the neuronal groups. “Big” and “Not So Big” data (solid lines) have be analyzed in combination to identify putative cellular networks. Cell types can be identified within the “Big” data by using information from “Not So Big” data. Then, cell states within each cell type community can be determined by molecular markers. The cell types and states can then be used to infer cell state transitions, trajectories, and interactions. A greater influence of “Big” and “Not So Big” data on developing the various networks is shown with solid lines with lesser influence shown by dashed lines. GF: growth factor, GFR: growth factor receptor, VIP: Vasoactive Intestinal Peptide, AVP: Arginine Vasopressin, PROK2: Prokineticin 2, PACAP: Pituitary Adenylate Cyclase-Activating Polypeptide. Fig. adapted from Park et al., (2016). Created using biorender.com .

Techniques Used: Biomarker Discovery, Western Blot, Immunostaining, Marker

Figure 4 A adapted from Nazari et al., (2018). Figure 4 B adapted from Cook et al., (2018). Created using biorender.com . " title="... informed by single-cell omics (A) Single-cell omics, including transcriptomics, proteomics, and metabolomics can be used for modeling ..." property="contentUrl" width="100%" height="100%"/>
Figure Legend Snippet: Computational models informed by single-cell omics (A) Single-cell omics, including transcriptomics, proteomics, and metabolomics can be used for modeling tumor cell differentiation dynamics. The specific cell types of interest that were identified within the tumor tissue include stem, progenitor, and differentiated cell types. State transitions, trajectories, and interactions between these cell types can then be inferred such that a network model can be generated. The tumor cell differentiation model can then be simulated to determine how the individual cell populations within the tumor change over time. (B) Single-cell omics experiments can be performed on the liver following resection to elucidate liver-specific cell types including Kupffer cells, Stellate cells and hepatocytes. For simplicity, we only show the hepatocyte cell states (replicating, quiescent, and primed), which are informed by molecular markers from the single-cell data. State transitions, trajectories and interactions can then be inferred from the cell states. A systems network model of liver regeneration can then be developed using the features extracted from the single-cell data and the model can be simulated for liver mass recovery and cellular dynamics. The total mass recovery as well as the populations of primed and replicating hepatocytes populations during regeneration are shown. Additionally, the populations of pro- and anti-regenerative stellate cell populations during regeneration are shown. Figure 4 A adapted from Nazari et al., (2018). Figure 4 B adapted from Cook et al., (2018). Created using biorender.com .

Techniques Used: Cell Differentiation, Generated

Figure Legend Snippet: Highlighted algorithms for computational modeling informed by single-cell and spatial omics data

Techniques Used: Expressing, Gene Expression, Spatial Proteomics

Figure 7 B adapted from Manchel et al., (2022). Figure 7 C adapted from Verma, Manchel et al., (2021). Created using biorender.com . " title="Patient-specific models informed by omics data (A) Metabolomics, transcriptomics, and proteomics data can be collected from a ..." property="contentUrl" width="100%" height="100%"/>
Figure Legend Snippet: Patient-specific models informed by omics data (A) Metabolomics, transcriptomics, and proteomics data can be collected from a patient’s liver sample. A patient-specific genome scale metabolic model (GEM) of the liver can then be generated by integrating the transcriptomics and proteomics data with a generic GEM (i.e., Human1 or Recon2 ). Metabolic fluxes are constrained using the metabolomics data and predicted by flux balance analysis. (B) Bulk and single-cell RNA-seq data can be utilized to generate context-specific metabolic models in health and disease (i.e., liver disease). Metabolic fluxes can be predicted by flux balance analysis and significantly perturbed metabolic pathways/subsystems can be identified in health vs. disease. For example, our analysis of liver transcriptomics data from alcoholic liver disease identified significant metabolic dysregulation in the glutathione (GSH) metabolic pathway. Specifically, the metabolic flux activity of specific solute transporters (LAT1, BAT1, OATP1A2) within the GSH pathway decreased with liver disease, while healthy livers showed an increase in flux along the pathway. (C) Zone-specific hepatocyte populations can be elucidated from single-cell omics data sources. The metabolic expression for genes in the B-oxidation and gluconeogenesis pathways decreases from zone 3 to zone 1, while it increases from zone 3 to zone 1 for genes in the glycolysis and lipogenesis pathways. Marker expression for each of the zonated hepatocyte populations within the “Big” data can be utilized in conjunction with “Not So Big” experimental data (i.e., neural tracings, calcium imaging, and glycogenolytic distribution analyses) to parameterize and structure a computational model of liver innervation, calcium signaling, and glycogenolysis. Additionally, the extent of innervation to the liver can be tuned in the model to the species of interest based on physiological evidence from the literature. Figure 7 B adapted from Manchel et al., (2022). Figure 7 C adapted from Verma, Manchel et al., (2021). Created using biorender.com .

Techniques Used: Generated, RNA Sequencing, Activity Assay, Expressing, Marker, Imaging