Brooke McElreath, M. | Search bioRxiv

bioRxiv2026

The Bayesian Inference library for Python R and Julia

O_LIBayesian modeling is a cornerstone of modern ecological and evolutionary research, offering the flexibility to account for hierarchical structures, imperfect detection, and spatial dependencies. However, as ecological datasets grow in scale and complexity--from high-resolution telemetry to phylogenomics--researchers increasingly face a "computational ceiling" where traditional CPUbound inference becomes prohibitively slow. C_LIO_LIThe current software landscape is fragmented. Researchers must often choose between high-level interfaces (e.g., brms in R) that are intuitive but sometimes rigid or slow for massive datasets, and low-level probabilistic programming languages (e.g., Stan, PyMC, JAX) that offer high performance but require specialized programming expertise. C_LIO_LIThis fragmentation is compounded by an "interoperability tax," where code developed in one language (e.g., R) cannot easily leverage the hardware-accelerated backends (GPUs/TPUs) typically found in Python-centric machine learning frameworks. C_LIO_LITo address these issues, we introduce BayesForge(BF), a cross-platform software ecosystem available in Python, R, and Julia. BF provides a unified, intuitive syntax that bridges the gap between ease-of-use and high-performance computation. By leveraging JAX-based backends (NumPyro and TensorFlow Probability), BF enables seamless hardware acceleration under-the-hood. C_LIO_LIWe demonstrate BFs utility through three ecological case studies: social network analysis (Social Relations Model), macroevolutionary uncertainty propagation across posterior tree sets, and latent-variable estimation for vocal repertoires. Benchmarks reveal that BF can achieve up to a 270-fold speedup over Stan implementations for large-scale networks, transforming weeks of computation into minutes and enabling more robust, uncertainty-aware ecological inference. C_LI

bioinformatics↗

bioRxiv2024

Robust Bayesian analysis of animal networks subject to biases in sampling intensity and censoring

Data collection biases are a persistent issue for studies of social networks. This issue has been particularly important in Animal Social Network Analysis (ASNA), where data are unevenly sampled and such biases may potentially lead to incorrect inferences about animal social behavior. Here, we address the issue by developing a Bayesian generative model, which not only estimates network structure, but also explicitly accounts for sampling and censoring biases. Using a set of simulation experiments designed to reflect various sampling and observational biases encountered in real-world scenarios, we systematically validate our model and evaluate its performance relative to other common ASNA methodologies. By accounting for differences in node-level censoring (i.e., the probability of missing an individual interaction.), our model permits the recovery of true latent social connections, even under a wide range of conditions where some key individuals are intermittently unobserved. Our model outperformed all other existing approaches and accurately captured network structure, as well as individual-level and dyad-level effects. Antithetically, permutation-based and simple linear regression aprroaches performed the worst across many conditions. These results highlight the advantages of generative network models for ASNA, as they offer greater flexibility, robustness, and adaptability to real-world data complexities. Our findings underscore the importance of generative models that jointly estimate network structure and adjust for measurement biases typical in empirical studies of animal social behaviour.

animal behavior and cognition↗