DAGBagM: learning directed acyclic graphs of mixed variables with an application to identify protein biomarkers for treatment response in ovarian cancer

DAGSLAM: causal Bayesian network structure learning of mixed type data and its application in identifying disease risk factors

BACKGROUND

Identifying and understanding disease risk factors is crucial in epidemiology, particularly for chronic and noncommunicable diseases that often have complex interrelationships. Traditional statistical methods struggle to capture these complexities, necessitating more sophisticated analytical frameworks. Bayesian networks and directed acyclic graphs (DAGs) provide powerful tools for exploring the complex relationships between variables. However, existing DAG structure learning algorithms still have limitations in handling mixed-type data (including continuous and discrete variables), which restricts their practical utility. Therefore, developing DAG structure learning methods that can effectively handle mixed data is highly important for obtaining an in-depth understanding of disease risk factors and pathogenic mechanisms.

METHODS

This study proposes an extension of the NOTEARS algorithm, termed DAGSLAM, which is designed for Bayesian network structure learning with mixed-type data. The algorithm integrates continuous and categorical variables through a tailored loss function, enhancing its applicability to real-world epidemiological datasets.

RESULTS

Extensive simulations were conducted across eight distinct scenarios, specifically, variations in the number of nodes, changes in the proportion of categorical nodes, different sample sizes, levels of categorical nodes, variations in edge sparsity, adjustments to the weight scale, different graph types, and diverse noise distributions. These scenarios demonstrate that DAGSLAM consistently outperforms existing methods such as HC, TABU, mDAG, and DAGBagM across key metrics, including precision, recall, F1 score, and structural Hamming distance (SHD). Furthermore, the robustness of DAGSLAM is validated through its application to the National Health and Nutrition Examination Survey (NHANES) dataset, revealing critical causal relationships among risk factors for CHD and diabetes.

CONCLUSIONS

DAGSLAM provides a powerful and scalable tool for uncovering causal relationships in complex disease networks, with significant implications for risk factor identification and public health research.

Learning directed acyclic graphs for ligands and receptors based on spatially resolved transcriptomic data of ovarian cancer

To unravel the mechanism of immune activation and suppression within tumors, a critical step is to identify transcriptional signals governing cell-cell communication between tumor and immune/stromal cells in the tumor microenvironment. Central to this communication are interactions between secreted ligands and cell-surface receptors, creating a highly connected signaling network among cells. Recent advancements in in situ-omics profiling, particularly spatial transcriptomic (ST) technology, provide unique opportunities to directly characterize ligand-receptor signaling networks that power cell-cell communication. In this paper, we propose a novel statistical method, LRnetST, to characterize the ligand-receptor interaction networks between adjacent tumor and immune/stroma cells based on ST data. LRnetST utilizes a directed acyclic graph model with a novel approach to handle the zero-inflated distributions of ST data. It also leverages existing ligand-receptor regulation databases as prior information, and employs a bootstrap aggregation strategy to achieve robust network estimation. Application of LRnetST to ST data of high-grade serous ovarian tumor samples revealed both common and distinct ligand-receptor regulations across different tumors. Some of these interactions were validated through both a MERFISH dataset and a CosMx SMI dataset of independent ovarian tumor samples. These results cast light on biological processes relating to the communication between tumor and immune/stromal cells in ovarian tumors. An open-source R package of LRnetST is available on GitHub at https://github.com/jie108/LRnetST.

Causal diagramming for assessing human system risk in spaceflight

For over a decade, the National Aeronautics and Space Administration (NASA) has tracked and configuration-managed approximately 30 risks that affect astronaut health and performance before, during and after spaceflight. The Human System Risk Board (HSRB) at NASA Johnson Space Center is responsible for setting the official risk posture for each of the human system risks and determining-based on evaluation of the available evidence-when that risk posture changes. The ultimate purpose of tracking and researching these risks is to find ways to reduce spaceflight-induced risk to astronauts. The adverse effects of spaceflight begin at launch and continue throughout the duration of the mission, and in some cases, across the lifetime of the astronaut. Historically, research has been conducted in individual risk "silos" to characterize risk, however, astronauts are exposed to all risks simultaneously. In January of 2020, the HSRB at NASA began assessing the potential value of causal diagramming as a tool to facilitate understanding of the complex causes and effects that contribute to spaceflight-induced human system risk. Causal diagrams in the form of directed acyclic graphs (DAGs) are used to provide HSRB stakeholders with a shared mental model of the causal flow of risk. While primarily improving communication among those stakeholders, DAGs also allow a composite risk network to be created that can be tracked and configuration managed. This paper outlines the HSRB's pilot process for this effort, the lessons learned, and future goals for data-driven risk management approaches.

Proteogenomic analysis of chemo-refractory high-grade serous ovarian cancer

To improve the understanding of chemo-refractory high-grade serous ovarian cancers (HGSOCs), we characterized the proteogenomic landscape of 242 (refractory and sensitive) HGSOCs, representing one discovery and two validation cohorts across two biospecimen types (formalin-fixed paraffin-embedded and frozen). We identified a 64-protein signature that predicts with high specificity a subset of HGSOCs refractory to initial platinum-based therapy and is validated in two independent patient cohorts. We detected significant association between lack of Ch17 loss of heterozygosity (LOH) and chemo-refractoriness. Based on pathway protein expression, we identified 5 clusters of HGSOC, which validated across two independent patient cohorts and patient-derived xenograft (PDX) models. These clusters may represent different mechanisms of refractoriness and implicate putative therapeutic vulnerabilities.