Cell2Cell: Explorative Cell Interaction Analysis in Multi-Volumetric Tissue Data.

We present Cell2Cell, a novel visual analytics approach for quantifying and visualizing networks of cell-cell interactions in three-dimensional (3D) multi-channel cancerous tissue data. By analyzing cellular interactions, biomedical experts can gain a more accurate understanding of the intricate relationships between cancer and immune cells. Recent methods have focused on inferring interaction based on the proximity of cells in low-resolution 2D multi-channel imaging data. By contrast, we analyze cell interactions by quantifying the presence and levels of specific proteins within a tissue sample (protein expressions) extracted from high-resolution 3D multi-channel volume data. Such analyses have a strong exploratory nature and require a tight integration of domain experts in the analysis loop to leverage their deep knowledge. We propose two complementary semi-automated approaches to cope with the increasing size and complexity of the data interactively: On the one hand, we interpret cell-to-cell interactions as edges in a cell graph and analyze the image signal (protein expressions) along those edges, using spatial as well as abstract visualizations. Complementary, we propose a cell-centered approach, enabling scientists to visually analyze polarized distributions of proteins in three dimensions, which also captures neighboring cells with biochemical and cell biological consequences. We evaluate our application in three case studies, where biologists and medical experts use Cell2Cell to investigate tumor micro-environments to identify and quantify T-cell activation in human tissue data. We confirmed that our tool can fully solve the use cases and enables a streamlined and detailed analysis of cell-cell interactions.

A General Framework for Comparing Embedding Visualizations Across Class-Label Hierarchies.

Projecting high-dimensional vectors into two dimensions for visualization, known as embedding visualization, facilitates perceptual reasoning and interpretation. Comparing multiple embedding visualizations drives decision-making in many domains, but traditional comparison methods are limited by a reliance on direct point correspondences. This requirement precludes comparisons without point correspondences, such as two different datasets of annotated images, and fails to capture meaningful higher-level relationships among point groups. To address these shortcomings, we propose a general framework for comparing embedding visualizations based on shared class labels rather than individual points. Our approach partitions points into regions corresponding to three key class concepts-confusion, neighborhood, and relative size-to characterize intra- and inter-class relationships. Informed by a preliminary user study, we implemented our framework using perceptual neighborhood graphs to defne these regions and introduced metrics to quantify each concept. We demonstrate the generality of our framework with usage scenarios from machine learning and single-cell biology, highlighting our metrics' ability to draw insightful comparisons across label hierarchies. To assess the effectiveness of our approach, we conducted an evaluation study with fve machine learning researchers and six single-cell biologists using an interactive and scalable prototype built with Python, JavaScript, and Rust. Our metrics enable more structured comparisons through visual guidance and increased participants' confdence in their fndings.

Vitessce: integrative visualization of multimodal and spatially resolved single-cell data.

Multiomics technologies with single-cell and spatial resolution make it possible to measure thousands of features across millions of cells. However, visual analysis of high-dimensional transcriptomic, proteomic, genome-mapped and imaging data types simultaneously remains a challenge. Here we describe Vitessce, an interactive web-based visualization framework for exploration of multimodal and spatially resolved single-cell data. We demonstrate integrative visualization of millions of data points, including cell-type annotations, gene expression quantities, spatially resolved transcripts and cell segmentations, across multiple coordinated views. The open-source software is available at http://vitessce.io .

A foundation model for clinician-centered drug repurposing.

Drug repurposing-identifying new therapeutic uses for approved drugs-is often a serendipitous and opportunistic endeavour to expand the use of drugs for new diseases. The clinical utility of drug-repurposing artificial intelligence (AI) models remains limited because these models focus narrowly on diseases for which some drugs already exist. Here we introduce TxGNN, a graph foundation model for zero-shot drug repurposing, identifying therapeutic candidates even for diseases with limited treatment options or no existing drugs. Trained on a medical knowledge graph, TxGNN uses a graph neural network and metric learning module to rank drugs as potential indications and contraindications for 17,080 diseases. When benchmarked against 8 methods, TxGNN improves prediction accuracy for indications by 49.2% and contraindications by 35.1% under stringent zero-shot evaluation. To facilitate model interpretation, TxGNN's Explainer module offers transparent insights into multi-hop medical knowledge paths that form TxGNN's predictive rationales. Human evaluation of TxGNN's Explainer showed that TxGNN's predictions and explanations perform encouragingly on multiple axes of performance beyond accuracy. Many of TxGNN's new predictions align well with off-label prescriptions that clinicians previously made in a large healthcare system. TxGNN's drug-repurposing predictions are accurate, consistent with off-label drug use, and can be investigated by human experts through multi-hop interpretable rationales.

Understanding Visualization Authoring Techniques for Genomics Data in the Context of Personas and Tasks.

Genomics experts rely on visualization to extract and share insights from complex and large-scale datasets. Beyond off-the-shelf tools for data exploration, there is an increasing need for platforms that aid experts in authoring customized visualizations for both exploration and communication of insights. A variety of interactive techniques have been proposed for authoring data visualizations, such as template editing, shelf configuration, natural language input, and code editors. However, it remains unclear how genomics experts create visualizations and which techniques best support their visualization tasks and needs. To address this gap, we conducted two user studies with genomics researchers: (1) semi-structured interviews (n = 20) to identify the tasks, user contexts, and current visualization authoring techniques and (2) an exploratory study (n = 13) using visual probes to elicit users' intents and desired techniques when creating visualizations. Our contributions include (1) a characterization of how visualization authoring is currently utilized in genomics visualization, identifying limitations and benefits in light of common criteria for authoring tools, and (2) generalizable design implications for genomics visualization authoring tools based on our findings on task- and user-specific usefulness of authoring techniques. All supplemental materials are available at https://osf.io/bdj4v/.

Learnable and Expressive Visualization Authoring through Blended Interfaces.

A wide range of visualization authoring interfaces enable the creation of highly customized visualizations. However, prioritizing expressiveness often impedes the learnability of the authoring interface. The diversity of users, such as varying computational skills and prior experiences in user interfaces, makes it even more challenging for a single authoring interface to satisfy the needs of a broad audience. In this paper, we introduce a framework to balance learnability and expressivity in a visualization authoring system. Adopting insights from learnability studies, such as multimodal interaction and visualization literacy, we explore the design space of blending multiple visualization authoring interfaces for supporting authoring tasks in a complementary and fexible manner. To evaluate the effectiveness of blending interfaces, we implemented a proof-of-concept system, Blace, that combines four common visualization authoring interfaces-template-based, shelf confguration, natural language, and code editor-that are tightly linked to one another to help users easily relate unfamiliar interfaces to more familiar ones. Using the system, we conducted a user study with 12 domain experts who regularly visualize genomics data as part of their analysis workfow. Participants with varied visualization and programming backgrounds were able to successfully reproduce unfamiliar visualization examples without a guided tutorial in the study. Feedback from a post-study qualitative questionnaire further suggests that blending interfaces enabled participants to learn the system easily and assisted them in confdently editing unfamiliar visualization grammar in the code editor, enabling expressive customization. Refecting on our study results and the design of our system, we discuss the different interaction patterns that we identifed and design implications for blending visualization authoring interfaces.

A foundation model for clinician-centered drug repurposing.

Drug repurposing - identifying new therapeutic uses for approved drugs - is often serendipitous and opportunistic, expanding the use of drugs for new diseases. The clinical utility of drug repurposing AI models remains limited because the models focus narrowly on diseases for which some drugs already exist. Here, we introduce TXGNN, a graph foundation model for zero-shot drug repurposing, identifying therapeutic candidates even for diseases with limited treatment options or no existing drugs. Trained on a medical knowledge graph, TXGNN utilizes a graph neural network and metric-learning module to rank drugs as potential indications and contraindications across 17,080 diseases. When benchmarked against eight methods, TXGNN improves prediction accuracy for indications by 49.2% and contraindications by 35.1% under stringent zero-shot evaluation. To facilitate model interpretation, TXGNN's Explainer module offers transparent insights into multi-hop medical knowledge paths that form TXGNN's predictive rationales. Human evaluation of TXGNN's Explainer showed that TXGNN's predictions and explanations perform encouragingly on multiple axes of performance beyond accuracy. Many of TxGNN's novel predictions align with off-label prescriptions clinicians make in a large healthcare system. TXGNN's drug repurposing predictions are accurate, consistent with off-label drug use, and can be investigated by human experts through multi-hop interpretable rationales.

MAMS: matrix and analysis metadata standards to facilitate harmonization and reproducibility of single-cell data.

Many datasets are being produced by consortia that seek to characterize healthy and disease tissues at single-cell resolution. While biospecimen and experimental information is often captured, detailed metadata standards related to data matrices and analysis workflows are currently lacking. To address this, we develop the matrix and analysis metadata standards (MAMS) to serve as a resource for data centers, repositories, and tool developers. We define metadata fields for matrices and parameters commonly utilized in analytical workflows and developed the rmams package to extract MAMS from single-cell objects. Overall, MAMS promotes the harmonization, integration, and reproducibility of single-cell data across platforms.

Data-driven prioritization of genetic disorders for global genomic newborn screening programs.

Genomic sequencing is poised to expand newborn screening for treatable childhood-onset disorders. Over 30 international research studies and companies are exploring its use, collectively aiming to screen more than 500,000 infants. A key challenge is determining which genes to include in screening. Among 27 newborn sequencing programs, the number of genes analyzed ranged from 134 to 4,299, with only 74 genes included by over 80% of programs. To understand this variability, we assembled a dataset with 25 characteristics of 4,389 genes included in any program and used a multivariate regression analysis to identify characteristics associated with inclusion across programs. These characteristics included presence on the US Recommended Uniform Screening panel, evidence regarding the natural history of disease, and efficacy of treatment. We then used a machine learning model to generate a ranked list of genes, offering a data-driven approach to the future prioritization of disorders for public health newborn screening efforts.

Neurological diagnoses in hospitalized COVID-19 patients associated with adverse outcomes: A multinational cohort study.

Few studies examining the patient outcomes of concurrent neurological manifestations during acute COVID-19 leveraged multinational cohorts of adults and children or distinguished between central and peripheral nervous system (CNS vs. PNS) involvement. Using a federated multinational network in which local clinicians and informatics experts curated the electronic health records data, we evaluated the risk of prolonged hospitalization and mortality in hospitalized COVID-19 patients from 21 healthcare systems across 7 countries. For adults, we used a federated learning approach whereby we ran Cox proportional hazard models locally at each healthcare system and performed a meta-analysis on the aggregated results to estimate the overall risk of adverse outcomes across our geographically diverse populations. For children, we reported descriptive statistics separately due to their low frequency of neurological involvement and poor outcomes. Among the 106,229 hospitalized COVID-19 patients (104,031 patients ≥18 years; 2,198 patients <18 years, January 2020-October 2021), 15,101 (14%) had at least one CNS diagnosis, while 2,788 (3%) had at least one PNS diagnosis. After controlling for demographics and pre-existing conditions, adults with CNS involvement had longer hospital stay (11 versus 6 days) and greater risk of (Hazard Ratio = 1.78) and faster time to death (12 versus 24 days) than patients with no neurological condition (NNC) during acute COVID-19 hospitalization. Adults with PNS involvement also had longer hospital stay but lower risk of mortality than the NNC group. Although children had a low frequency of neurological involvement during COVID-19 hospitalization, a substantially higher proportion of children with CNS involvement died compared to those with NNC (6% vs 1%). Overall, patients with concurrent CNS manifestation during acute COVID-19 hospitalization faced greater risks for adverse clinical outcomes than patients without any neurological diagnosis. Our global informatics framework using a federated approach (versus a centralized data collection approach) has utility for clinical discovery beyond COVID-19.

The chromatin landscape of healthy and injured cell types in the human kidney.

There is a need to define regions of gene activation or repression that control human kidney cells in states of health, injury, and repair to understand the molecular pathogenesis of kidney disease and design therapeutic strategies. Comprehensive integration of gene expression with epigenetic features that define regulatory elements remains a significant challenge. We measure dual single nucleus RNA expression and chromatin accessibility, DNA methylation, and H3K27ac, H3K4me1, H3K4me3, and H3K27me3 histone modifications to decipher the chromatin landscape and gene regulation of the kidney in reference and adaptive injury states. We establish a spatially-anchored epigenomic atlas to define the kidney's active, silent, and regulatory accessible chromatin regions across the genome. Using this atlas, we note distinct control of adaptive injury in different epithelial cell types. A proximal tubule cell transcription factor network of ELF3, KLF6, and KLF10 regulates the transition between health and injury, while in thick ascending limb cells this transition is regulated by NR2F1. Further, combined perturbation of ELF3, KLF6, and KLF10 distinguishes two adaptive proximal tubular cell subtypes, one of which manifested a repair trajectory after knockout. This atlas will serve as a foundation to facilitate targeted cell-specific therapeutics by reprogramming gene regulatory networks.

Cistrome Data Browser: integrated search, analysis and visualization of chromatin data.

The Cistrome Data Browser is a resource of ChIP-seq, ATAC-seq and DNase-seq data from humans and mice. It provides maps of the genome-wide locations of transcription factors, cofactors, chromatin remodelers, histone post-translational modifications and regions of chromatin accessible to endonuclease activity. Cistrome DB v3.0 contains approximately 45 000 human and 44 000 mouse samples with about 32 000 newly collected datasets compared to the previous release. The Cistrome DB v3.0 user interface is implemented as a single page application that unifies menu driven and data driven search functions and provides an embedded genome browser, which allows users to find and visualize data more effectively. Users can find informative chromatin profiles through keyword, menu, and data-driven search tools. Browser search functions can predict the regulators of query genes as well as the cell type and factor dependent functionality of potential cis-regulatory elements. Cistrome DB v3.0 expands the display of quality control statistics, incorporates sequence logos into motif enrichment displays and includes more expansive sample metadata. Cistrome DB v3.0 is available at http://db3.cistrome.org/browser.

DRAVA: Aligning Human Concepts with Machine Learning Latent Dimensions for the Visual Exploration of Small Multiples.

Latent vectors extracted by machine learning (ML) are widely used in data exploration (e.g., t-SNE) but suffer from a lack of interpretability. While previous studies employed disentangled representation learning (DRL) to enable more interpretable exploration, they often overlooked the potential mismatches between the concepts of humans and the semantic dimensions learned by DRL. To address this issue, we propose Drava, a visual analytics system that supports users in 1) relating the concepts of humans with the semantic dimensions of DRL and identifying mismatches, 2) providing feedback to minimize the mismatches, and 3) obtaining data insights from concept-driven exploration. Drava provides a set of visualizations and interactions based on visual piles to help users understand and refine concepts and conduct concept-driven exploration. Meanwhile, Drava employs a concept adaptor model to fine-tune the semantic dimensions of DRL based on user refinement. The usefulness of Drava is demonstrated through application scenarios and experimental validation.

Alerts and Collections for Automating Patients' Sensemaking and Organizing of Their Electronic Health Record Data for Reflection, Planning, and Clinical Visits: Qualitative Research-Through-Design Study.

BACKGROUND: Electronic health record (EHR) data from multiple providers often exhibit important but convoluted and complex patterns that patients find hard and time-consuming to identify and interpret. However, existing patient-facing applications lack the capability to incorporate automatic pattern detection robustly and toward supporting making sense of the patient's EHR data. In addition, there is no means to organize EHR data in an efficient way that suits the patient's needs and makes them more actionable in real-life settings. These shortcomings often result in a skewed and incomplete picture of the patient's health status, which may lead to suboptimal decision-making and actions that put the patient at risk. OBJECTIVE: Our main goal was to investigate patients' attitudes, needs, and use scenarios with respect to automatic support for surfacing important patterns in their EHR data and providing means for organizing them that best suit patients' needs. METHODS: We conducted an inquisitive research-through-design study with 14 participants. Presented in the context of a cutting-edge application with strong emphasis on independent EHR data sensemaking, called Discovery, we used high-level mock-ups for the new features that were supposed to support automatic identification of important data patterns and offer recommendations-Alerts-and means for organizing the medical records based on patients' needs, much like photos in albums-Collections. The combined audio recording transcripts and in-study notes were analyzed using the reflexive thematic analysis approach. RESULTS: The Alerts and Collections can be used for raising awareness, reflection, planning, and especially evidence-based patient-provider communication. Moreover, patients desired carefully designed automatic pattern detection with safe and actionable recommendations, which produced a well-tailored and scoped landscape of alerts for both potential threats and positive progress. Furthermore, patients wanted to contribute their own data (eg, progress notes) and log feelings, daily observations, and measurements to enrich the meaning and enable easier sensemaking of the alerts and collections. On the basis of the findings, we renamed Alerts to Reports for a more neutral tone and offered design implications for contextualizing the reports more deeply for increased actionability; automatically generating the collections for more expedited and exhaustive organization of the EHR data; enabling patient-generated data input in various formats to support coarser organization, richer pattern detection, and learning from experience; and using the reports and collections for efficient, reliable, and common-ground patient-provider communication. CONCLUSIONS: Patients need to have a flexible and rich way to organize and annotate their EHR data; be introduced to insights from these data-both positive and negative; and share these artifacts with their physicians in clinical visits or via messaging for establishing shared mental models for clear goals, agreed-upon priorities, and feasible actions.

How surgeons use risk calculators and non-clinical factors for informed consent and shared decision making: A qualitative study.

BACKGROUND: The discussion of risks, benefits, and alternatives to surgery with patients is a defining component of informed consent. As shared-decision making has become central to surgeon-patient communication, risk calculators have emerged as a tool to aid communication and decision-making. To optimize informed consent, it is necessary to understand how surgeons assess and communicate risk, and the role of risk calculators in this process. METHODS: We conducted interviews with 13 surgeons from two institutions to understand how surgeons assess risk, the role of risk calculators in decision-making, and how surgeons approach risk communication during informed consent. We performed a qualitative analysis of interviews based on SRQR guidelines. RESULTS: Our analysis yielded insights regarding (a) the landscape and approach to obtaining surgical consent; (b) detailed perceptions regarding the value and design of assessing and communicating risk; and (c) practical considerations regarding the future of personalized risk communication in decision-making. Above all, we found that non-clinical factors such as health and risk literacy are changing how surgeons assess and communicate risk, which diverges from traditional risk calculators. CONCLUSION: Principally, we found that surgeons incorporate a range of clinical and non-clinical factors to risk stratify patients and determine how to optimally frame and discuss risk with individual patients. We observed that surgeons' perception of risk communication, and the importance of eliciting patient preferences to direct shared-decision making, did not consistently align with patient priorities. This study underscored criticisms of risk calculators and novel decision-aids - which must be addressed prior to greater adoption.

Organ Mapping Antibody Panels: a community resource for standardized multiplexed tissue imaging.

Multiplexed antibody-based imaging enables the detailed characterization of molecular and cellular organization in tissues. Advances in the field now allow high-parameter data collection (>60 targets); however, considerable expertise and capital are needed to construct the antibody panels employed by these methods. Organ mapping antibody panels are community-validated resources that save time and money, increase reproducibility, accelerate discovery and support the construction of a Human Reference Atlas.

Advances and prospects for the Human BioMolecular Atlas Program (HuBMAP).

The Human BioMolecular Atlas Program (HuBMAP) aims to create a multi-scale spatial atlas of the healthy human body at single-cell resolution by applying advanced technologies and disseminating resources to the community. As the HuBMAP moves past its first phase, creating ontologies, protocols and pipelines, this Perspective introduces the production phase: the generation of reference spatial maps of functional tissue units across many organs from diverse populations and the creation of mapping tools and infrastructure to advance biomedical research.

The chromatin landscape of healthy and injured cell types in the human kidney.

There is a need to define regions of gene activation or repression that control human kidney cells in states of health, injury, and repair to understand the molecular pathogenesis of kidney disease and design therapeutic strategies. However, comprehensive integration of gene expression with epigenetic features that define regulatory elements remains a significant challenge. We measured dual single nucleus RNA expression and chromatin accessibility, DNA methylation, and H3K27ac, H3K4me1, H3K4me3, and H3K27me3 histone modifications to decipher the chromatin landscape and gene regulation of the kidney in reference and adaptive injury states. We established a comprehensive and spatially-anchored epigenomic atlas to define the kidney's active, silent, and regulatory accessible chromatin regions across the genome. Using this atlas, we noted distinct control of adaptive injury in different epithelial cell types. A proximal tubule cell transcription factor network of ELF3 , KLF6 , and KLF10 regulated the transition between health and injury, while in thick ascending limb cells this transition was regulated by NR2F1 . Further, combined perturbation of ELF3 , KLF6 , and KLF10 distinguished two adaptive proximal tubular cell subtypes, one of which manifested a repair trajectory after knockout. This atlas will serve as a foundation to facilitate targeted cell-specific therapeutics by reprogramming gene regulatory networks.