CMB_2024v14n2

Computational Molecular Biology 2024, Vol.14, No.2, 84-94 http://bioscipublisher.com/index.php/cmb 91 on a large scale. Causal inference methods such as Mendelian randomization, Bayesian networks, and Granger causality are frequently applied to prioritize genes and pathways involved in disease processes. For example, causal inference algorithms like Tumour-specific Causal Inference (TCI) have been used to identify somatic genome alterations that affect gene expression in cancer. TCI was applied to data from The Cancer Genome Atlas (TCGA) to identify causal gene modules in breast cancer and glioblastoma, revealing subgroups of patients with distinct pathway aberrations and survival outcomes (Xue et al., 2019). Other methods, such as deep learning models, have been used to infer gene relationships and causality from single-cell RNA sequencing data, helping to identify disease-related genes more accurately than traditional methods (Yuan & Bar-Joseph, 2018). By leveraging these computational approaches, researchers can identify genes that drive disease progression, such as transcription factors and signaling molecules, providing valuable targets for therapeutic interventions. 6.2 Understanding disease mechanisms Causal inference methods have significantly advanced our understanding of disease mechanisms by revealing how genetic and environmental factors interact to cause disease. In complex diseases like cancer, Alzheimer's disease, and diabetes, multiple genes and pathways contribute to disease onset and progression. Network-based approaches, such as Bayesian networks and multi-layered network models, have been employed to construct gene regulatory networks and infer the causal relationships between genes and phenotypes. For example, a causal network inference algorithm was applied to gene transcriptional data from A549 cells exposed to glucocorticoids, identifying key regulatory genes and their effects on gene expression patterns related to cellular stress responses (Lu et al., 2019). Another study focused on the molecular mechanisms of Alzheimer’s disease, constructing a differential gene network by integrating omics data, which revealed the role of epigenetic regulation and ribosomal processes in disease progression (Park et al., 2017). 7 Advances in Computational Tools and Software 7.1 Recent developments in software for causal inference Recent advances in computational tools for causal inference have led to the development of highly efficient, specialized software designed to handle large-scale biological datasets. Tools like PREMER (Parallel Reverse Engineering with Mutual Information & Entropy Reduction) use information theory to infer biological network structures, enabling users to distinguish between direct and indirect interactions within networks and to determine causal links. PREMER, developed with OpenMP for parallel execution, alleviates computational bottlenecks, especially in large-scale network inference problems, and supports multiple operating systems and programming interfaces such as Python and MATLAB (Villaverde et al., 2018). Another notable development is SIGNET, a software package designed to infer gene regulatory networks using large-scale transcriptomic and genotypic data. SIGNET incorporates genotypic variants as instrumental variables to infer causal relationships across the entire transcriptome, making it particularly suitable for high-dimensional genomic data. By leveraging parallel computing environments, SIGNET is optimized for handling computationally intensive tasks and provides an interactive interface for parameter tuning and network visualization (Zhang et al., 2023). In addition, the netZoo platform, developed for the inference and analysis of multi-omics biological networks, integrates multiple omics data sources and provides tools to infer gene regulatory networks and conduct differential analyses. This platform is optimized for multi-tiered cancer data analysis, such as from the Cancer Cell Line Encyclopedia (CCLE), and helps identify novel regulatory elements involved in cancer development (Guebila et al., 2022). 7.2 User-friendly tools for biologists As more biologists seek to conduct causal inference without deep expertise in computational biology, the demand for user-friendly tools has risen. Tools like CausalR have been developed to bridge this gap. CausalR is a causal network analysis platform implemented in R that allows for easy integration with popular software such as Cytoscape. This platform provides a user-friendly interface for biologists, enabling them to perform causal inference on genome-scale data without needing advanced programming skills (Bradley & Barrett, 2017).

RkJQdWJsaXNoZXIy MjQ4ODYzNA==