Applications of Bayesian network models in predicting types of hematological malignancies

Bayesian networks in modeling leucocyte interplay following brain irradiation: A comprehensive framework

BACKGROUND AND OBJECTIVE

Understanding the intricate interactions among leucocyte subpopulations following radiotherapy is crucial for advancing cancer research and immunology. Recently, interest in recent radiotherapy modalities, such as protons, has increased. Herein, we present a framework utilizing Bayesian networks to uncover these complex relationships via an illustrative example of brain irradiation in rodents.

METHODS

We utilized data from 96 healthy C57BL/6 adult mice subjected to either X-ray or proton brain irradiation. Leucocyte subpopulations in the blood collected 12 h after the final irradiated fraction were quantified. We employed Bayesian networks to detect causal interplay between physiological parameters, radiation variables and circulating leucocytes. The causal structure was learned via the use of the Bayesian information criterion as a scored criterion. Parameter estimation was performed to quantify the strength of the identified causal relationships. Cross-validation was used to validate our Bayesian network model's performance.

RESULTS

In the X-ray model, we discovered previously undisclosed interactions between NK-cells and neutrophils, and between monocytes and T-CD4+ cells. The proton model revealed an interplay involving T-CD4+ cells and neutrophils. Both X-rays and protons led to heightened interactions between T-CD8+ cells and B cells, indicating their significant role in orchestrating immune responses. Additionally, the proton model displayed strengthened interactions between T-CD4+ and T-CD8+ cells, emphasizing a dynamic and coordinated immune response post-irradiation. Cross-validation results demonstrated the robustness of the Bayesian network model in explaining data uncertainty.

CONCLUSION

The use of Bayesian networks as tools for causal structure discovery has revealed novel insights into the dynamics of immune responses to radiation exposure.

Systems genetics analysis of human body fat distribution genes identifies adipocyte processes

Excess abdominal fat is a sexually dimorphic risk factor for cardio-metabolic disease and is approximated by the waist-to-hip ratio adjusted for body mass index (WHRadjBMI). Whereas this trait is highly heritable, few causal genes are known. We aimed to identify novel drivers of WHRadjBMI using systems genetics. We used two independent cohorts of adipose tissue gene expression and constructed sex- and depot-specific Bayesian networks to model gene-gene interactions from 8,492 genes. Using key driver analysis, we identified genes that, in silico and putatively in vitro, regulate many others. 51-119 key drivers in each network were replicated in both cohorts. In other cell types, 23 of these genes are found in crucial adipocyte pathways: Wnt signaling or mitochondrial function. We overexpressed or down-regulated seven key driver genes in human subcutaneous pre-adipocytes. Key driver genes ANAPC2 and RSPO1 inhibited adipogenesis, whereas PSME3 increased adipogenesis. RSPO1 increased Wnt signaling activity. In differentiated adipocytes, MIGA1 and UBR1 down-regulation led to mitochondrial dysfunction. These five genes regulate adipocyte function, and we hypothesize that they regulate fat distribution.

CausNet-partial : 'Partial Generational Orderings' based search for optimal sparse Bayesian networks via dynamic programming with parent set constraints

Background

In our recent work, we developed a novel dynamic programming algorithm to find optimal Bayesian networks (BNs) with parent set constraints. This 'generational orderings' based dynamic programming search algorithm - CausNet - efficiently searches the space of possible BNs given the possible parent sets. The algorithm supports both continuous and categorical data, as well as continuous, binary and survival outcomes. In the present work, we develop a variant of CausNet - CausNet-partial - which searches the space of 'partial generational orderings', which further reduces the search space and is suited for finding smaller sparse optimal Bayesian networks; and can be applied to 1000s of variables.

Results

We test this method both on synthetic and real data. Our algorithm performs better than three state-of-art algorithms that are currently used extensively to find optimal BNs. We apply it to simulated continuous data and also to a benchmark discrete Bayesian network ALARM, a Bayesian network designed to provide an alarm message system for patient monitoring. We first apply the original CausNet and then CausNet-partial varying the partial order from 5 to 2. CausNet-partial discovers small sparse networks with drastically reduced runtime as expected from theory.

Conclusions

Our partial generational orderings based search for small optimal networks, is both an efficient and highly scalable approach for finding optimal sparse and small Bayesian Networks and can be applied to 1000s of variables. Using specifiable parameters - correlation, FDR cutoffs, in-degree, and partial order - one can increase or decrease the number of nodes and density of the networks. Availability of two scoring option - BIC and Bge - and implementation for survival outcomes and mixed data types makes our algorithm very suitable for many types of high dimensional data in a variety of fields.

BIC-LP: A Hybrid Higher-Order Dynamic Bayesian Network Score Function for Gene Regulatory Network Reconstruction

Reconstructing gene regulatory networks(GRNs) is an increasingly hot topic in bioinformatics. Dynamic Bayesian network(DBN) is a stochastic graph model commonly used as a vital model for GRN reconstruction. But probabilistic characteristics of biological networks and the existence of data noise bring great challenges to GRN reconstruction and always lead to many false positive/negative edges. ScoreLasso is a hybrid DBN score function combining DBN and linear regression with good performance. Its performance is, however, limited by first-order assumption and ignorance of the initial network of DBN. In this article, an integrated model based on higher-order DBN model, higher-order Lasso linear regression model and Pearson correlation model is proposed. Based on this, a hybrid higher-order DBN score function for GRN reconstruction is proposed, namely BIC-LP. BIC-LP score function is constructed by adding terms based on Lasso linear regression coefficients and Pearson correlation coefficients on classical BIC score function. Therefore, it could capture more information from dataset and curb information loss, compared with both many existing Bayesian family score functions and many state-of-the-art methods for GRN reconstruction. Experimental results show that BIC-LP can reasonably eliminate some false positive edges while retaining most true positive edges, so as to achieve better GRN reconstruction performance.

Integrating DNA methylation and gene expression data in a single gene network using the iNETgrate package

Analyzing different omics data types independently is often too restrictive to allow for detection of subtle, but consistent, variations that are coherently supported based upon different assays. Integrating multi-omics data in one model can increase statistical power. However, designing such a model is challenging because different omics are measured at different levels. We developed the iNETgrate package ( https://bioconductor.org/packages/iNETgrate/ ) that efficiently integrates transcriptome and DNA methylation data in a single gene network. Applying iNETgrate on five independent datasets improved prognostication compared to common clinical gold standards and a patient similarity network approach.

Evaluation of a decision support system using Bayesian network modeling in an applied Multi-INT surveillance environment

Sensemaking and decision-making are fundamental components of applied Intelligence, Surveillance, and Reconnaissance (ISR). Analysts acquire information from multiple sources over a period of hours, days, or even over the scale of months or years, that must be interpreted and integrated to predict future adversarial events. Sensemaking is essential for developing an appropriate mental model that will lead to accurate predictions sooner. Decision Support Systems (DSS) are one proposed solution to improve analyst decision-making outcomes by leveraging computers to conduct calculations that may be difficult for human operators and provide recommendations. In this study, we tested two simulated DSS that were informed by a Bayesian Network Model as a potential prediction-assistive tool. Participants completed a simulated multi-day, multi-source intelligence task and were asked to make predictions regarding five potential outcomes on each day. Participants in both DSS conditions were able to converge on the correct solution significantly faster than the control group, and between 36-44% more of the sample was able to reach the correct conclusion. Furthermore, we found that a DSS representing projected outcome probabilities as numerical, rather than using verbal ordinal labels, were better able to differentiate which outcomes were extremely unlikely than the control group or verbal-probability DSS.

Systems genetics analysis of human body fat distribution genes identifies Wnt signaling and mitochondrial activity in adipocytes

BACKGROUND

Excess fat in the abdomen is a sexually dimorphic risk factor for cardio-metabolic disease. The relative storage between abdominal and lower-body subcutaneous adipose tissue depots is approximated by the waist-to-hip ratio adjusted for body mass index (WHRadjBMI). Genome-wide association studies (GWAS) identified 346 loci near 495 genes associated with WHRadjBMI. Most of these genes have unknown roles in fat distribution, but many are expressed and putatively act in adipose tissue. We aimed to identify novel sex- and depot-specific drivers of WHRadjBMI using a systems genetics approach.

METHODS

We used two independent cohorts of adipose tissue gene expression with 362 - 444 males and 147 - 219 females, primarily of European ancestry. We constructed sex- and depot- specific Bayesian networks to model the gene-gene interactions from 8,492 adipose tissue genes. Key driver analysis identified genes that, in silico and putatively in vitro, regulate many others, including the 495 WHRadjBMI GWAS genes. Key driver gene function was determined by perturbing their expression in human subcutaneous pre-adipocytes using lenti-virus or siRNA.

RESULTS

51 - 119 key drivers in each network were replicated in both cohorts. We used single-cell expression data to select replicated key drivers expressed in adipocyte precursors and mature adipocytes, prioritized genes which have not been previously studied in adipose tissue, and used public human and mouse data to nominate 53 novel key driver genes (10 - 21 from each network) that may regulate fat distribution by altering adipocyte function. In other cell types, 23 of these genes are found in crucial adipocyte pathways: Wnt signaling or mitochondrial function. We selected seven genes whose expression is highly correlated with WHRadjBMI to further study their effects on adipogenesis/Wnt signaling (ANAPC2, PSME3, RSPO1, TYRO3) or mitochondrial function (C1QTNF3, MIGA1, PSME3, UBR1).Adipogenesis was inhibited in cells overexpressing ANAPC2 and RSPO1 compared to controls. RSPO1 results are consistent with a positive correlation between gene expression in the subcutaneous depot and WHRadjBMI, therefore lower relative storage in the subcutaneous depot. RSPO1 inhibited adipogenesis by increasing β-catenin activation and Wnt-related transcription, thus repressing PPARG and CEBPA. PSME3 overexpression led to more adipogenesis than controls. In differentiated adipocytes, MIGA1 and UBR1 downregulation led to mitochondrial dysfunction, with lower oxygen consumption than controls; MIGA1 knockdown also lowered UCP1 expression.

SUMMARY

ANAPC2, MIGA1, PSME3, RSPO1, and UBR1 affect adipocyte function and may drive body fat distribution.

"iNETgrate": integrating DNA methylation and gene expression data in a single gene network

Integrating multi-omics data in one model can increase statistical power. However, designing such a model is challenging because different omics are measured at different levels. We developed the iNETgrate package (https://bioconductor.org/packages/iNETgrate/) that efficiently integrates transcriptome and DNA methylation data in a single gene network. Applying iNETgrate on five independent datasets improved prognostication compared to common clinical gold standards and a patient similarity network approach.

Nicotinamide mononucleotide (NMN) alleviates the poly(I:C)-induced inflammatory response in human primary cell cultures

NMN is the direct precursor of nicotinamide adenine dinucleotide (NAD+) and is considered as a key factor for increasing NAD+ levels and mitochondrial activity in cells. In this study, based on transcriptome analysis, we showed that NMN alleviates the poly(I:C)-induced inflammatory response in cultures of two types of human primary cells, human pulmonary microvascular endothelial cells (HPMECs) and human coronary artery endothelial cells (HCAECs). Major inflammatory mediators, including IL6 and PARP family members, were grouped into coexpressed gene modules and significantly downregulated under NMN exposure in poly(I:C)-activated conditions in both cell types. The Bayesian network analysis of module hub genes predicted common genes, including eukaryotic translation initiation factor 4B (EIF4B), and distinct genes, such as platelet-derived growth factor binding molecules, in HCAECs, which potentially regulate the identified inflammation modules. These results suggest a robust regulatory mechanism by which NMN alleviates inflammatory pathway activation, which may open up the possibility of a new role for NMN replenishment in the treatment of chronic or acute inflammation.

Bayesian network enables interpretable and state-of-the-art prediction of immunotherapy responses in cancer patients

Immune checkpoint inhibitors, especially PD-1/PD-L1 blockade, have revolutionized cancer treatment and brought tremendous benefits to patients who otherwise would have had a limited prognosis. Nonetheless, only a small fraction of patients respond to immunotherapy, and the costs and side effects of immune checkpoint inhibitors cannot be ignored. With the advent of machine and deep learning, clinical and genetic data have been used to stratify patient responses to immunotherapy. Unfortunately, these approaches have typically been "black-box" methods that are unable to explain their predictions, thereby hindering their responsible clinical application. Herein, we developed a "white-box" Bayesian network model that achieves accurate and interpretable predictions of immunotherapy responses against nonsmall cell lung cancer (NSCLC). This tree-augmented naïve Bayes (TAN) model accurately predicted durable clinical benefits and distinguished two clinically significant subgroups with distinct prognoses. Furthermore, our state-of-the-art white-box TAN approach achieved greater accuracy than previous methods. We hope that our model will guide clinicians in selecting NSCLC patients who truly require immunotherapy and expect our approach to be easily applied to other types of cancer.

CausNet: generational orderings based search for optimal Bayesian networks via dynamic programming with parent set constraints

BACKGROUND

Finding a globally optimal Bayesian Network using exhaustive search is a problem with super-exponential complexity, which severely restricts the number of variables that can feasibly be included. We implement a dynamic programming based algorithm with built-in dimensionality reduction and parent set identification. This reduces the search space substantially and can be applied to large-dimensional data. We use what we call 'generational orderings' based search for optimal networks, which is a novel way to efficiently search the space of possible networks given the possible parent sets. The algorithm supports both continuous and categorical data, as well as continuous, binary and survival outcomes.

RESULTS

We demonstrate the efficacy of our algorithm on both synthetic and real data. In simulations, our algorithm performs better than three state-of-art algorithms that are currently used extensively. We then apply it to an Ovarian Cancer gene expression dataset with 513 genes and a survival outcome. Our algorithm is able to find an optimal network describing the disease pathway consisting of 6 genes leading to the outcome node in just 3.4 min on a personal computer with a 2.3 GHz Intel Core i9 processor with 16 GB RAM.

CONCLUSIONS

Our generational orderings based search for optimal networks is both an efficient and highly scalable approach for finding optimal Bayesian Networks and can be applied to 1000 s of variables. Using specifiable parameters-correlation, FDR cutoffs, and in-degree-one can increase or decrease the number of nodes and density of the networks. Availability of two scoring option-BIC and Bge-and implementation for survival outcomes and mixed data types makes our algorithm very suitable for many types of high dimensional data in a variety of fields.

Fuzzy weighted Bayesian belief network: a medical knowledge-driven Bayesian model using fuzzy weighted rules

In this current work, Weighted Bayesian Association rules using the Fuzzy set theory are proposed with the new concept of Fuzzy Weighted Bayesian Association Rules to design and develop a Clinical Decision Support System on the Bayesian Belief Network, which is an appropriate area to work in Clinical Domain as it has a higher degree of unpredictability and causality. Weighted Bayesian Association rules to construct a Bayesian network are already proposed. A "Sharp boundary" issue related to quantitative attribute domains may cause erroneous predictions in medicine and treatment in the medical environment. So to eradicate sharp boundary problems in the medical field, the fuzzy theory is applied in attributes to deal with real-life situations. A new algorithm is designed and implemented in this paper to set up a new Bayesian belief network using the concept of Fuzzy Weighted Association rule mining under the Predictive Modeling paradigm named Fuzzy weighted Bayesian belief network using numerous clinical datasets with outshone results.

Risk factors and a Bayesian network model to predict ischemic stroke in patients with dilated cardiomyopathy

OBJECTIVE

This study aimed to identify risk factors and create a predictive model for ischemic stroke (IS) in patients with dilated cardiomyopathy (DCM) using the Bayesian network (BN) approach.

MATERIALS AND METHODS

We collected clinical data of 634 patients with DCM treated at three referral management centers in Beijing between 2016 and 2021, including 127 with and 507 without IS. The patients were randomly divided into training (441 cases) and test (193 cases) sets at a ratio of 7:3. A BN model was established using the Tabu search algorithm with the training set data and verified with the test set data. The BN and logistic regression models were compared using the area under the receiver operating characteristic curve (AUC).

RESULTS

Multivariate logistic regression analysis showed that hypertension, hyperlipidemia, atrial fibrillation/flutter, estimated glomerular filtration rate (eGFR), and intracardiac thrombosis were associated with IS. The BN model found that hyperlipidemia, atrial fibrillation (AF) or atrial flutter, eGFR, and intracardiac thrombosis were closely associated with IS. Compared to the logistic regression model, the BN model for IS performed better or equally well in the training and test sets, with respective accuracies of 83.7 and 85.5%, AUC of 0.763 [95% confidence interval (CI), 0.708-0.818] and 0.822 (95% CI, 0.748-0.896), sensitivities of 20.2 and 44.2%, and specificities of 98.3 and 97.3%.

CONCLUSION

Hypertension, hyperlipidemia, AF or atrial flutter, low eGFR, and intracardiac thrombosis were good predictors of IS in patients with DCM. The BN model was superior to the traditional logistic regression model in predicting IS in patients with DCM and is, therefore, more suitable for early IS detection and diagnosis, and could help prevent the occurrence and recurrence of IS in this patient cohort.

Machine learning with in silico analysis markedly improves survival prediction modeling in colon cancer patients

BACKGROUND

Predicting the survival of cancer patients provides prognostic information and therapeutic guidance. However, improved prediction models are needed for use in diagnosis and treatment.

OBJECTIVE

This study aimed to identify genomic prognostic biomarkers related to colon cancer (CC) based on computational data and to develop survival prediction models.

METHODS

We performed machine-learning (ML) analysis to screen pathogenic survival-related driver genes related to patient prognosis by integrating copy number variation and gene expression data. Moreover, in silico system analysis was performed to clinically assess data from ML analysis, and we identified RABGAP1L, MYH9, and DRD4 as candidate genes. These three genes and tumor stages were used to generate survival prediction models. Moreover, the genes were validated by experimental and clinical analyses, and the theranostic application of the survival prediction models was assessed.

RESULTS

RABGAP1L, MYH9, and DRD4 were identified as survival-related candidate genes by ML and in silico system analysis. The survival prediction model using the expression of the three genes showed higher predictive performance when applied to predict the prognosis of CC patients. A series of functional analyses revealed that each knockdown of three genes reduced the protumor activity of CC cells. In particular, validation with an independent cohort of CC patients confirmed that the coexpression of MYH9 and DRD4 gene expression reflected poorer clinical outcomes in terms of overall survival and disease-free survival.

CONCLUSIONS

Our survival prediction approach will contribute to providing information on patients and developing a therapeutic strategy for CC patients.

A comprehensive survey on computational learning methods for analysis of gene expression data

Computational analysis methods including machine learning have a significant impact in the fields of genomics and medicine. High-throughput gene expression analysis methods such as microarray technology and RNA sequencing produce enormous amounts of data. Traditionally, statistical methods are used for comparative analysis of gene expression data. However, more complex analysis for classification of sample observations, or discovery of feature genes requires sophisticated computational approaches. In this review, we compile various statistical and computational tools used in analysis of expression microarray data. Even though the methods are discussed in the context of expression microarrays, they can also be applied for the analysis of RNA sequencing and quantitative proteomics datasets. We discuss the types of missing values, and the methods and approaches usually employed in their imputation. We also discuss methods of data normalization, feature selection, and feature extraction. Lastly, methods of classification and class discovery along with their evaluation parameters are described in detail. We believe that this detailed review will help the users to select appropriate methods for preprocessing and analysis of their data based on the expected outcome.

eQTLs as causal instruments for the reconstruction of hormone linked gene networks

Hormones act within in highly dynamic systems and much of the phenotypic response to variation in hormone levels is mediated by changes in gene expression. The increase in the number and power of large genetic association studies has led to the identification of hormone linked genetic variants. However, the biological mechanisms underpinning the majority of these loci are poorly understood. The advent of affordable, high throughput next generation sequencing and readily available transcriptomic databases has shown that many of these genetic variants also associate with variation in gene expression levels as expression Quantitative Trait Loci (eQTLs). In addition to further dissecting complex genetic variation, eQTLs have been applied as tools for causal inference. Many hormone networks are driven by transcription factors, and many of these genes can be linked to eQTLs. In this mini-review, we demonstrate how causal inference and gene networks can be used to describe the impact of hormone linked genetic variation upon the transcriptome within an endocrinology context.

Weighted Bayesian Belief Network: A Computational Intelligence Approach for Predictive Modeling in Clinical Datasets

There are growing concerns about the mortality due to Breast cancer many of which often result from delayed detection and treatment. So an effective computational approach is needed to develop a predictive model which will help patients and physicians to manage the situation timely. This study presented a Weighted Bayesian Belief Network (WBBN) modeling for breast cancer prediction using the UCI breast cancer dataset. New automated ranking method was used to assign proper weights to attribute value pair based on their impact on causing the disease. Association between attributes was generated using weighted association rule mining between two attributes, multiattributes, and with class labels to generate rules. Weighted Bayesian confidence and weighted Bayesian lift measures were used to produce strong rules to build the model. To build WBBN, the Open Markov tool was used for structure and parametric learning using generated strong rules. The model was trained using 70% records and tested on 30% records with a threshold value of minimum support = 36% and confidence = 70% which produced results with an accuracy of 97.18%. Experimental results show that WBBN achieved better results in most cases compared to other predictive models. The study would contribute to the fight against breast cancer and the quality of treatment.

Practical application of a Bayesian network approach to poultry epigenetics and stress

Background

Relationships among genetic or epigenetic features can be explored by learning probabilistic networks and unravelling the dependencies among a set of given genetic/epigenetic features. Bayesian networks (BNs) consist of nodes that represent the variables and arcs that represent the probabilistic relationships between the variables. However, practical guidance on how to make choices among the wide array of possibilities in Bayesian network analysis is limited. Our study aimed to apply a BN approach, while clearly laying out our analysis choices as an example for future researchers, in order to provide further insights into the relationships among epigenetic features and a stressful condition in chickens (Gallus gallus).

Results

Chickens raised under control conditions (n = 22) and chickens exposed to a social isolation protocol (n = 24) were used to identify differentially methylated regions (DMRs). A total of 60 DMRs were selected by a threshold, after bioinformatic pre-processing and analysis. The treatment was included as a binary variable (control = 0; stress = 1). Thereafter, a BN approach was applied: initially, a pre-filtering test was used for identifying pairs of features that must not be included in the process of learning the structure of the network; then, the average probability values for each arc of being part of the network were calculated; and finally, the arcs that were part of the consensus network were selected. The structure of the BN consisted of 47 out of 61 features (60 DMRs and the stressful condition), displaying 43 functional relationships. The stress condition was connected to two DMRs, one of them playing a role in tight and adhesive intracellular junctions in organs such as ovary, intestine, and brain.

Conclusions

We clearly explain our steps in making each analysis choice, from discrete BN models to final generation of a consensus network from multiple model averaging searches. The epigenetic BN unravelled functional relationships among the DMRs, as well as epigenetic features in close association with the stressful condition the chickens were exposed to. The DMRs interacting with the stress condition could be further explored in future studies as possible biomarkers of stress in poultry species.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04800-0.

Comparing Bayesian-Based Reconstruction Strategies in Topology-Based Pathway Enrichment Analysis

The development of high-throughput omics technologies has enabled the quantification of vast amounts of genes and gene products in the whole genome. Pathway enrichment analysis (PEA) provides an intuitive solution for extracting biological insights from massive amounts of data. Topology-based pathway analysis (TPA) represents the latest generation of PEA methods, which exploit pathway topology in addition to lists of differentially expressed genes and their expression profiles. A subset of these TPA methods, such as BPA, BNrich, and PROPS, reconstruct pathway structures by training Bayesian networks (BNs) from canonical biological pathways, providing superior representations that explain causal relationships between genes. However, these methods have never been compared for their differences in the PEA and their different topology reconstruction strategies. In this study, we aim to compare the BN reconstruction strategies of the BPA, BNrich, PROPS, Clipper, and Ensemble methods and their PEA and performance on tumor and non-tumor classification based on gene expression data. Our results indicate that they performed equally well in distinguishing tumor and non-tumor samples (AUC > 0.95) yet with a varying ranking of pathways, which can be attributed to the different BN structures resulting from the different cyclic structure removal strategies. This can be clearly seen from the reconstructed JAK-STAT networks by different strategies. In a nutshell, BNrich, which relies on expert intervention to remove loops and cyclic structures, produces BNs that best fit the biological facts. The plausibility of the Clipper strategy can also be partially explained by intuitive biological rules and theorems. Our results may offer an informed reference for the proper method for a given data analysis task.

Molecular characterization of depression trait and state

Major depressive disorder (MDD) is a brain disorder often characterized by recurrent episode and remission phases. The molecular correlates of MDD have been investigated in case-control comparisons, but the biological alterations associated with illness trait (regardless of clinical phase) or current state (symptomatic and remitted phases) remain largely unknown, limiting targeted drug discovery. To characterize MDD trait- and state-dependent changes, in single or recurrent depressive episode or remission, we generated transcriptomic profiles of subgenual anterior cingulate cortex of postmortem subjects in first MDD episode (n = 20), in remission after a single episode (n = 15), in recurrent episode (n = 20), in remission after recurring episodes (n = 15) and control subject (n = 20). We analyzed the data at the gene, biological pathway, and cell-specific molecular levels, investigated putative causal events and therapeutic leads. MDD-trait was associated with genes involved in inflammation, immune activation, and reduced bioenergetics (q < 0.05) whereas MDD-states were associated with altered neuronal structure and reduced neurotransmission (q < 0.05). Cell-level deconvolution of transcriptomic data showed significant change in density of GABAergic interneurons positive for corticotropin-releasing hormone, somatostatin, or vasoactive-intestinal peptide (p < 3 × 10^-3). A probabilistic Bayesian-network approach showed causal roles of immune-system-activation (q < 8.67 × 10^-3), cytokine-response (q < 4.79 × 10^-27) and oxidative-stress (q < 2.05 × 10^-3) across MDD-phases. Gene-sets associated with these putative causal changes show inverse associations with the transcriptomic effects of dopaminergic and monoaminergic ligands. The study provides first insights into distinct cellular and molecular pathologies associated with trait- and state-MDD, on plasticity mechanisms linking the two pathologies, and on a method of drug discovery focused on putative disease-causing pathways.

Profiling senescent cells in human brains reveals neurons with CDKN2D/p19 and tau neuropathology

Senescent cells contribute to pathology and dysfunction in animal models1. Their sparse distribution and heterogenous phenotype have presented challenges for detecting them in human tissues. We developed a senescence eigengene approach to identify these rare cells within large, diverse populations of postmortem human brain cells. Eigengenes are useful when no single gene reliably captures a phenotype, like senescence; they also help to reduce noise, which is important in large transcriptomic datasets where subtle signals from low-expressing genes can be lost. Each of our eigengenes detected ~2% senescent cells from a population of ~140,000 single nuclei derived from 76 postmortem human brains with various levels of Alzheimer's disease (AD) pathology. More than 97% of the senescent cells were excitatory neurons and overlapped with tau-containing neurofibrillary tangles (NFTs). Cyclin dependent kinase inhibitor 2D (CDKN2D/p19) was predicted as the most significant contributor to the primary senescence eigengene. RNAscope and immunofluorescence confirmed its elevated expression in AD brain tissue whereby p19-expressing neurons had 1.8-fold larger nuclei and significantly more cells with lipofuscin than p19-negative neurons. These hallmark senescence phenotypes were further elevated in the presence of NFTs. Collectively, CDKN2D/p19-expressing neurons with NFTs represent a unique cellular population in human AD with a senescence phenotype. The eigengenes developed may be useful in future senescence profiling studies as they accurately identified senescent cells in snRNASeq datasets and predicted biomarkers for histological investigation.

Addressing noise in co-expression network construction

Gene co-expression networks (GCNs) provide multiple benefits to molecular research including hypothesis generation and biomarker discovery. Transcriptome profiles serve as input for GCN construction and are derived from increasingly larger studies with samples across multiple experimental conditions, treatments, time points, genotypes, etc. Such experiments with larger numbers of variables confound discovery of true network edges, exclude edges and inhibit discovery of context (or condition) specific network edges. To demonstrate this problem, a 475-sample dataset is used to show that up to 97% of GCN edges can be misleading because correlations are false or incorrect. False and incorrect correlations can occur when tests are applied without ensuring assumptions are met, and pairwise gene expression may not meet test assumptions if the expression of at least one gene in the pairwise comparison is a function of multiple confounding variables. The ‘one-size-fits-all’ approach to GCN construction is therefore problematic for large, multivariable datasets. Recently, the Knowledge Independent Network Construction toolkit has been used in multiple studies to provide a dynamic approach to GCN construction that ensures statistical tests meet assumptions and confounding variables are addressed. Additionally, it can associate experimental context for each edge of the network resulting in context-specific GCNs (csGCNs). To help researchers recognize such challenges in GCN construction, and the creation of csGCNs, we provide a review of the workflow.

Cellular, molecular, and therapeutic characterization of pilocarpine-induced temporal lobe epilepsy

Animal models have expanded our understanding of temporal lobe epilepsy (TLE). However, translating these to cell-specific druggable hypotheses is not explored. Herein, we conducted an integrative insilico-analysis of an available transcriptomics dataset obtained from animals with pilocarpine-induced-TLE. A set of 119 genes with subtle-to-moderate impact predicted most forms of epilepsy with ~ 97% accuracy and characteristically mapped to upregulated homeostatic and downregulated synaptic pathways. The deconvolution of cellular proportions revealed opposing changes in diverse cell types. The proportion of nonneuronal cells increased whereas that of interneurons, except for those expressing vasoactive intestinal peptide (Vip), decreased, and pyramidal neurons of the cornu-ammonis (CA) subfields showed the highest variation in proportion. A probabilistic Bayesian-network demonstrated an aberrant and oscillating physiological interaction between nonneuronal cells involved in the blood–brain-barrier and Vip interneurons in driving seizures, and their role was evaluated insilico using transcriptomic changes induced by valproic-acid, which showed opposing effects in the two cell-types. Additionally, we revealed novel epileptic and antiepileptic mechanisms and predicted drugs using causal inference, outperforming the present drug repurposing approaches. These well-powered findings not only expand the understanding of TLE and seizure oscillation, but also provide predictive biomarkers of epilepsy, cellular and causal micro-circuitry changes associated with it, and a drug-discovery method focusing on these events.

Principles and Practice of Explainable Machine Learning

Artificial intelligence (AI) provides many opportunities to improve private and public life. Discovering patterns and structures in large troves of data in an automated manner is a core component of data science, and currently drives applications in diverse areas such as computational biology, law and finance. However, such a highly positive impact is coupled with a significant challenge: how do we understand the decisions suggested by these systems in order that we can trust them? In this report, we focus specifically on data-driven methods-machine learning (ML) and pattern recognition models in particular-so as to survey and distill the results and observations from the literature. The purpose of this report can be especially appreciated by noting that ML models are increasingly deployed in a wide range of businesses. However, with the increasing prevalence and complexity of methods, business stakeholders in the very least have a growing number of concerns about the drawbacks of models, data-specific biases, and so on. Analogously, data science practitioners are often not aware about approaches emerging from the academic literature or may struggle to appreciate the differences between different methods, so end up using industry standards such as SHAP. Here, we have undertaken a survey to help industry practitioners (but also data scientists more broadly) understand the field of explainable machine learning better and apply the right tools. Our latter sections build a narrative around a putative data scientist, and discuss how she might go about explaining her models by asking the right questions. From an organization viewpoint, after motivating the area broadly, we discuss the main developments, including the principles that allow us to study transparent models vs. opaque models, as well as model-specific or model-agnostic post-hoc explainability approaches. We also briefly reflect on deep learning models, and conclude with a discussion about future research directions.

The role of clinical signs and spirometry in the diagnosis of obstructive airway diseases: a systematic analysis adapted to general practice settings

Background

In general practice (GP), the diagnosis of obstructive airway diseases much relies on diagnostic questions, in view of the limited availability of lung function. We systematically assessed the relative importance of such questions for diagnosing asthma and chronic obstructive pulmonary disease (COPD), either without or with information from spirometry.

Methods

We used data obtained in a pulmonary practice to ensure the validity of diagnoses and assessments. Subjects with a diagnosis of COPD (n=260), or asthma (n=433), or other respiratory diseases (n=230), and subjects without respiratory diseases (n=364, controls) were included. The diagnostic questions comprised eight items, covering smoking history, self-attributed allergic rhinitis, dyspnea, cough, phlegm and wheeze. Optionally standard parameters of the flow-volume-curve were included. Decision trees for the diagnosis of COPD and asthma were constructed, moreover a probabilistic diagnostic network based on the results of path analyses describing the relationship between variables.

Results

In the decision trees, age, sex, current smoking, wheezing, dyspnea upon mild exertion, self-attributed allergic rhinitis, phlegm, forced expiratory volume in one second (FEV₁), and expiratory flow rates were relevant, depending on the diagnostic comparison, while cough, dyspnea upon strong exertion and ex-smoker status were not relevant. In contrast, the probabilistic network for the diagnosis of COPD and asthma versus controls incorporated all diagnostic questions, i.e., dyspnea upon mild or strong exertion, current smoking, ex-smoking, wheezing, cough and phlegm but from spirometry only FEV₁. Depending on the individual pattern, the probability for COPD could raise from 25% to 81%, while the diagnostic gain for asthma was lower.

Conclusions

The study developed simple diagnostic algorithms for asthma and COPD that take into account the relative importance of clinical signs and history, as well as spirometric data if available. The diagnostic accuracy was especially high for COPD. These algorithms may be helpful as a starting point in the standardisation of diagnostic strategies in GP practices.

Trial registration

The study is registered under DRKS00013935 at German Clinical Trials Register (DRKS, Date of registration 01/03/2018).

Making Sense of Genetic Information: The Promising Evolution of Clinical Stratification and Precision Oncology Using Machine Learning

Precision medicine is a medical approach to administer patients with a tailored dose of treatment by taking into consideration a person's variability in genes, environment, and lifestyles. The accumulation of omics big sequence data led to the development of various genetic databases on which clinical stratification of high-risk populations may be conducted. In addition, because cancers are generally caused by tumor-specific mutations, large-scale systematic identification of single nucleotide polymorphisms (SNPs) in various tumors has propelled significant progress of tailored treatments of tumors (i.e., precision oncology). Machine learning (ML), a subfield of artificial intelligence in which computers learn through experience, has a great potential to be used in precision oncology chiefly to help physicians make diagnostic decisions based on tumor images. A promising venue of ML in precision oncology is the integration of all available data from images to multi-omics big data for the holistic care of patients and high-risk healthy subjects. In this review, we provide a focused overview of precision oncology and ML with attention to breast cancer and glioma as well as the Bayesian networks that have the flexibility and the ability to work with incomplete information. We also introduce some state-of-the-art attempts to use and incorporate ML and genetic information in precision oncology.

An integrative Bayesian network approach to highlight key drivers in systemic lupus erythematosus

BACKGROUND

A comprehensive intuition of the systemic lupus erythematosus (SLE), as a complex and multifactorial disease, is a biological challenge. Dealing with this challenge needs employing sophisticated bioinformatics algorithms to discover the unknown aspects. This study aimed to underscore key molecular characteristics of SLE pathogenesis, which may serve as effective targets for therapeutic intervention.

METHODS

In the present study, the human peripheral blood mononuclear cell (PBMC) microarray datasets (n = 6), generated by three platforms, which included SLE patients (n = 220) and healthy control samples (n = 135) were collected. Across each platform, we integrated the datasets by cross-platform normalization (CPN). Subsequently, through BNrich method, the structures of Bayesian networks (BNs) were extracted from KEGG-indexed SLE, TCR, and BCR signaling pathways; the values of the node (gene) and edge (intergenic relationships) parameters were estimated within each integrated datasets. Parameters with the FDR < 0.05 were considered significant. Finally, a mixture model was performed to decipher the signaling pathway alterations in the SLE patients compared to healthy controls.

RESULTS

In the SLE signaling pathway, we identified the dysregulation of several nodes involved in the (1) clearance mechanism (SSB, MACROH2A2, TRIM21, H2AX, and C1Q gene family), (2) autoantigen presentation by MHCII (HLA gene family, CD80, IL10, TNF, and CD86), and (3) end-organ damage (FCGR1A, ELANE, and FCGR2A). As a remarkable finding, we demonstrated significant perturbation in CD80 and CD86 to CD28, CD40LG to CD40, C1QA and C1R to C2, and C1S to C4A edges. Moreover, we not only replicated previous studies regarding alterations of subnetworks involved in TCR and BCR signaling pathways (PI3K/AKT, MAPK, VAV gene family, AP-1 transcription factor) but also distinguished several significant edges between genes (PPP3 to NFATC gene families). Our findings unprecedentedly showed that different parameter values assign to the same node based on the pathway topology (the PIK3CB parameter values were 1.7 in TCR vs - 0.5 in BCR signaling pathway).

CONCLUSIONS

Applying the BNrich as a hybridized network construction method, we highlight under-appreciated systemic alterations of SLE, TCR, and BCR signaling pathways in SLE. Consequently, having such a systems biology approach opens new insights into the context of multifactorial disorders.

Wireless Body Area Network (WBAN)-Based Telemedicine for Emergency Care

This paper is a collection of telemedicine techniques used by wireless body area networks (WBANs) for emergency conditions. Furthermore, Bayes’ theorem is proposed for predicting emergency conditions. With prior knowledge, the posterior probability can be found along with the observed evidence. The probability of sending emergency messages can be determined using Bayes’ theorem with the likelihood evidence. It can be viewed as medical decision-making, since diagnosis conditions such as emergency monitoring, delay-sensitive monitoring, and general monitoring are analyzed with its network characteristics, including data rate, cost, packet loss rate, latency, and jitter. This paper explains the network model with 16 variables, with one describing immediate consultation, as well as another three describing emergency monitoring, delay-sensitive monitoring, and general monitoring. The remaining 12 variables are observations related to latency, cost, packet loss rate, data rate, and jitter.

Co-expression Networks Identify DHX15 RNA Helicase as a B Cell Regulatory Factor

Genome-wide co-expression analysis is often used for annotating novel gene functions from high-dimensional data. Here, we developed an R package with a Shiny visualization app that creates immuno-networks from RNAseq data using a combination of Weighted Gene Co-expression Network Analysis (WGCNA), xCell immune cell signatures, and Bayesian Network Learning. Using a large publicly available RNAseq dataset we generated a Gene Module-Immune Cell (GMIC) network that predicted causal relationships between DEAH-box RNA helicase (DHX)15 and genes associated with humoral immunity, suggesting that DHX15 may regulate B cell fate. Deletion of DHX15 in mouse B cells led to impaired lymphocyte development, reduced peripheral B cell numbers, and dysregulated expression of genes linked to antibody-mediated immune responses similar to the genes predicted by the GMIC network. Moreover, antigen immunization of mice demonstrated that optimal primary IgG1 responses required DHX15. Intrinsic expression of DHX15 was necessary for proliferation and survival of activated of B cells. Altogether, these results support the use of co-expression networks to elucidate fundamental biological processes.

Identification of Factors Influencing Out-of-county Hospitalizations in the New Cooperative Medical Scheme

Throughout the duration of the New Cooperative Medical Scheme (NCMS), it was found that an increasing number of rural patients were seeking out-of-county medical treatment, which posed a great burden on the NCMS fund. Our study was conducted to examine the prevalence of out-of-county hospitalizations and its related factors, and to provide a scientific basis for follow-up health insurance policies. A total of 215 counties in central and western China from 2008 to 2016 were selected. The total out-of-county hospitalization rate in nine years was 16.95%, which increased from 12.37% in 2008 to 19.21% in 2016 with an average annual growth rate of 5.66%. Its related expenses and compensations were shown to increase each year, with those in the central region being higher than those in the western region. Stepwise logistic regression reveals that the increase in out-of-county hospitalization rate was associated with region (X1), rural population (X2), per capita per year net income (X3), per capita gross domestic product (GDP) (X4), per capita funding amount of NCMS (X5), compensation ratio of out-of-county hospitalization cost (X6), per time average in-county (X7) and out-of-county hospitalization cost (X8). According to Bayesian network (BN), the marginal probability of high out-of-county hospitalization rate was as high as 81.7%. Out-of-county hospitalizations were directly related to X8, X3, X4 and X6. The probability of high out-of-county hospitalization obtained based on hospitalization expenses factors, economy factors, regional characteristics and NCMS policy factors was 95.7%, 91.1%, 93.0% and 88.8%, respectively. And how these factors affect out-of-county hospitalization and their interrelationships were found out. Our findings suggest that more attention should be paid to the influence mechanism of these factors on out-of-county hospitalizations, and the increase of hospitalizations outside the county should be reasonably supervised and controlled and our results will be used to help guide the formulation of proper intervention policies.

Application of tabu search-based Bayesian networks in exploring related factors of liver cirrhosis complicated with hepatic encephalopathy and disease identification

This study aimed to explore the related factors and strengths of hepatic cirrhosis complicated with hepatic encephalopathy (HE) by multivariate logistic regression analysis and tabu search-based Bayesian networks (BNs), and to deduce the probability of HE in patients with cirrhosis under different conditions through BN reasoning. Multivariate logistic regression analysis indicated that electrolyte disorders, infections, poor spirits, hepatorenal syndrome, hepatic diabetes, prothrombin time, and total bilirubin are associated with HE. Inferences by BNs found that infection, electrolyte disorder and hepatorenal syndrome are closely related to HE. Those three variables are also related to each other, indicating that the occurrence of any of those three complications may induce the other two complications. When those three complications occur simultaneously, the probability of HE may reach 0.90 or more. The BN constructed by the tabu search algorithm can analyze not only how the correlative factors affect HE but also their interrelationships. Reasoning using BNs can describe how HE is induced on the basis of the order in which doctors acquire patient information, which is consistent with the sequential process of clinical diagnosis and treatment.