Efficient Sampling and Structure Learning of Bayesian Networks

Using Causal Bayesian Networks to Assess the Role of Patient-Centered Care and Psychosocial Factors on Durable HIV Viral Suppression

We assessed the role of patient-centered care on durable viral suppression (i.e., all viral load test results < 200 copies per ml during 2019) by conducting a retrospective cohort study of clients medically case managed by the Miami-Dade County Ryan White Program (RWP). Summary measures of patient-centered care practices of RWP-affiliated providers were obtained from a survey of 1352 clients. Bayesian network models analyzed the complex relationship between psychosocial and patient-centered care factors. Of 5037 clients, 4135 (82.1%) had durable viral suppression. Household income was the factor most strongly associated with durable viral suppression. Further, mean healthcare relationship score and mean "provider knows patient as a person" score were both associated with durable viral suppression. Healthcare relationship score moderated the association between low household income and lack of durable viral suppression. Although patient-centered care supports patient HIV care success, wrap around support is also needed for people with unmet psychosocial needs.

Sexual abuse and psychotic phenomena: a directed acyclic graph analysis of affective symptoms using English national psychiatric survey data

BACKGROUND

Sexual abuse and bullying are associated with poor mental health in adulthood. We previously established a clear relationship between bullying and symptoms of psychosis. Similarly, we would expect sexual abuse to be linked to the emergence of psychotic symptoms, through effects on negative affect.

METHOD

We analysed English data from the Adult Psychiatric Morbidity Surveys, carried out in 2007 (N = 5954) and 2014 (N = 5946), based on representative national samples living in private households. We used probabilistic graphical models represented by directed acyclic graphs (DAGs). We obtained measures of persecutory ideation and auditory hallucinosis from the Psychosis Screening Questionnaire, and identified affective symptoms using the Clinical Interview Schedule. We included cannabis consumption and sex as they may determine the relationship between symptoms. We constrained incoming edges to sexual abuse and bullying to respect temporality.

RESULTS

In the DAG analyses, contrary to our expectations, paranoia appeared early in the cascade of relationships, close to the abuse variables, and generally lying upstream of affective symptoms. Paranoia was consistently directly antecedent to hallucinations, but also indirectly so, via non-psychotic symptoms. Hallucinosis was also the endpoint of pathways involving non-psychotic symptoms.

CONCLUSIONS

Via worry, sexual abuse and bullying appear to drive a range of affective symptoms, and in some people, these may encourage the emergence of hallucinations. The link between adverse experiences and paranoia is much more direct. These findings have implications for managing distressing outcomes. In particular, worry may be a salient target for intervention in psychosis.

A survey of Bayesian Network structure learning

Bayesian Networks (BNs) have become increasingly popular over the last few decades as a tool for reasoning under uncertainty in fields as diverse as medicine, biology, epidemiology, economics and the social sciences. This is especially true in real-world areas where we seek to answer complex questions based on hypothetical evidence to determine actions for intervention. However, determining the graphical structure of a BN remains a major challenge, especially when modelling a problem under causal assumptions. Solutions to this problem include the automated discovery of BN graphs from data, constructing them based on expert knowledge, or a combination of the two. This paper provides a comprehensive review of combinatoric algorithms proposed for learning BN structure from data, describing 74 algorithms including prototypical, well-established and state-of-the-art approaches. The basic approach of each algorithm is described in consistent terms, and the similarities and differences between them highlighted. Methods of evaluating algorithms and their comparative performance are discussed including the consistency of claims made in the literature. Approaches for dealing with data noise in real-world datasets and incorporating expert knowledge into the learning process are also covered.

Multi-omics subtyping of hepatocellular carcinoma patients using a Bayesian network mixture model

Comprehensive molecular characterization of cancer subtypes is essential for predicting clinical outcomes and searching for personalized treatments. We present bnClustOmics, a statistical model and computational tool for multi-omics unsupervised clustering, which serves a dual purpose: Clustering patient samples based on a Bayesian network mixture model and learning the networks of omics variables representing these clusters. The discovered networks encode interactions among all omics variables and provide a molecular characterization of each patient subgroup. We conducted simulation studies that demonstrated the advantages of our approach compared to other clustering methods in the case where the generative model is a mixture of Bayesian networks. We applied bnClustOmics to a hepatocellular carcinoma (HCC) dataset comprising genome (mutation and copy number), transcriptome, proteome, and phosphoproteome data. We identified three main HCC subtypes together with molecular characteristics, some of which are associated with survival even when adjusting for the clinical stage. Cluster-specific networks shed light on the links between genotypes and molecular phenotypes of samples within their respective clusters and suggest targets for personalized treatments.

Gene expression data analysis using Hellinger correlation in weighted gene co-expression networks (WGCNA)

Weighted gene co-expression network analysis (WGCNA) is used to detect clusters with highly correlated genes. Measurements of correlation most typically rely on linear relationships. However, a linear relationship does not always model pairwise functional-related dependence between genes. In this paper, we first compared 6 different correlation methods in their ability to capture complex dependence between genes in three different tissues. Next, we compared their gene-pairwise coefficient results and corresponding WGCNA results. Finally, we applied a recently proposed correlation method, Hellinger correlation, as a more sensitive correlation measurement in WGCNA. To test this method, we constructed gene networks containing co-expression gene modules from RNA-seq data of human frontal cortex from Alzheimer's disease patients. To test the generality, we also used a microarray data set from human frontal cortex, single cell RNA-seq data from human prefrontal cortex, RNA-seq data from human temporal cortex, and GTEx data from heart. The Hellinger correlation method captures essentially similar results as other linear correlations in WGCNA, but provides additional new functional relationships as exemplified by uncovering a link between inflammation and mitochondria function. We validated the network constructed with the microarray and single cell sequencing data sets and a RNA-seq dataset of temporal cortex. We observed that this new correlation method enables the detection of non-linear biologically meaningful relationships among genes robustly and provides a complementary new approach to WGCNA. Thus, the application of Hellinger correlation to WGCNA provides a more flexible correlation approach to modelling networks in gene expression analysis that uncovers novel network relationships.

The FEDHC Bayesian Network Learning Algorithm

The paper proposes a new hybrid Bayesian network learning algorithm, termed Forward Early Dropping Hill Climbing (FEDHC), devised to work with either continuous or categorical variables. Further, the paper manifests that the only implementation of MMHC in the statistical software R is prohibitively expensive, and a new implementation is offered. Further, specifically for the case of continuous data, a robust to outliers version of FEDHC, which can be adopted by other BN learning algorithms, is proposed. The FEDHC is tested via Monte Carlo simulations that distinctly show that it is computationally efficient, and that it produces Bayesian networks of similar to, or of higher accuracy than MMHC and PCHC. Finally, an application of FEDHC, PCHC and MMHC algorithms to real data, from the field of economics, is demonstrated using the statistical software R.

Discovering gene regulatory networks of multiple phenotypic groups using dynamic Bayesian networks

Dynamic Bayesian networks (DBNs) can be used for the discovery of gene regulatory networks (GRNs) from time series gene expression data. Here, we suggest a strategy for learning DBNs from gene expression data by employing a Bayesian approach that is scalable to large networks and is targeted at learning models with high predictive accuracy. Our framework can be used to learn DBNs for multiple groups of samples and highlight differences and similarities in their GRNs. We learn these DBN models based on different structural and parametric assumptions and select the optimal model based on the cross-validated predictive accuracy. We show in simulation studies that our approach is better equipped to prevent overfitting than techniques used in previous studies. We applied the proposed DBN-based approach to two time series transcriptomic datasets from the Gene Expression Omnibus database, each comprising data from distinct phenotypic groups of the same tissue type. In the first case, we used DBNs to characterize responders and non-responders to anti-cancer therapy. In the second case, we compared normal to tumor cells of colorectal tissue. The classification accuracy reached by the DBN-based classifier for both datasets was higher than reported previously. For the colorectal cancer dataset, our analysis suggested that GRNs for cancer and normal tissues have a lot of differences, which are most pronounced in the neighborhoods of oncogenes and known cancer tissue markers. The identified differences in gene networks of cancer and normal cells may be used for the discovery of targeted therapies.

Estimation of cancer cell fractions and clone trees from multi-region sequencing of tumors

MOTIVATION

Multi-region sequencing of solid tumors can improve our understanding of intratumor subclonal diversity and the evolutionary history of mutational events. Due to uncertainty in clonal composition and the multitude of possible ancestral relationships between clones, elucidating the most probable relationships from bulk tumor sequencing poses statistical and computational challenges.

RESULTS

We developed a Bayesian hierarchical model called PICTograph to model uncertainty in assigning mutations to subclones, to enable posterior distributions of cancer cell fractions, and to visualize the most probable ancestral relationships between subclones. Compared to available methods, PICTograph provided more consistent and accurate estimates of cancer cell fractions and improved tree inference over a range of simulated clonal diversity. Application of PICTograph to multi-region whole exome sequencing of tumors from individuals with pancreatic cancer precursor lesions confirmed known early-occurring mutations and indicated substantial molecular diversity, including 6-12 distinct subclones and intra-sample mixing of subclones. Using ensemble-based visualizations, we highlight highly probable evolutionary relationships recovered in multiple models. PICTograph provides a useful approximation to evolutionary inference from cross-sectional multi-region sequencing, particularly for complex cases.

AVAILABILITY

https://github.com/KarchinLab/pictograph.

A Better Mechanistic Understanding of Big Data through an Order Search Using Causal Bayesian Networks

Every year, biomedical data is increasing at an alarming rate and is being collected from many different sources, such as hospitals (clinical Big Data), laboratories (genomic and proteomic Big Data), and the internet (online Big Data). This article presents and evaluates a practical causal discovery algorithm that uses modern statistical, machine learning, and informatics approaches that have been used in the learning of causal relationships from biomedical Big Data, which in turn integrates clinical, omics (genomic and proteomic), and environmental aspects. The learning of causal relationships from data using graphical models does not address the hidden (unknown or not measured) mechanisms that are inherent to most measurements and analyses. Also, many algorithms lack a practical usage since they do not incorporate current mechanistic knowledge. This paper proposes a practical causal discovery algorithm using causal Bayesian networks to gain a better understanding of the underlying mechanistic process that generated the data. The algorithm utilizes model averaging techniques such as searching through a relative order (e.g., if gene A is regulating gene B, then we can say that gene A is of a higher order than gene B) and incorporates relevant prior mechanistic knowledge to guide the Markov chain Monte Carlo search through the order. The algorithm was evaluated by testing its performance on datasets generated from the ALARM causal Bayesian network. Out of the 37 variables in the ALARM causal Bayesian network, two sets of nine were chosen and the observations for those variables were provided to the algorithm. The performance of the algorithm was evaluated by comparing its prediction with the generating causal mechanism. The 28 variables that were not in use are referred to as hidden variables and they allowed for the evaluation of the algorithm’s ability to predict hidden confounded causal relationships. The algorithm’s predicted performance was also compared with other causal discovery algorithms. The results show that incorporating order information provides a better mechanistic understanding even when hidden confounded causes are present. The prior mechanistic knowledge incorporated in the Markov chain Monte Carlo search led to the better discovery of causal relationships when hidden variables were involved in generating the simulated data.