Motivation: The gut microbiome shapes cancer therapy response through its influence on host metabolism. While prior studies examine pairwise associations between individual genera and metabolites, there is limited methodology for identifying microbial genera that systematically regulate the overall metabolome. Scalable statistical tools are needed to uncover such system-level 'master predictors' in high-dimensional microbiome-metabolome data. Results: We introduce B-MASTER, a scalable Bayesian multivariate regression framework combining L1 sparsity and L2 group shrinkage to identify essential cross-metabolite regulators. A Gibbs sampler enables near-linear computational scaling, supporting models with millions of parameters. The method is supported by theoretical guarantees, including posterior contraction and selection consistency. Analysis of colorectal cancer microbiome-metabolome data reveals key microbial genera that govern global and cancer-associated metabolite patterns, highlighting system-level regulatory structure. Availability: The B-MASTER code, including demonstration scripts, is available at https://github.com/priyamdas2/B-MASTER. An archived snapshot of the code corresponding to this manuscript is available on Zenodo with DOI: 10.5281/zenodo.20484958.

翻译：动机：肠道微生物群通过影响宿主代谢来塑造癌症治疗反应。尽管既往研究探讨了单个属与代谢物之间的成对关联，但用于识别系统性调控整体代谢组的微生物属的方法学仍十分有限。因此，亟需可扩展的统计工具来揭示此类高维微生物组-代谢组数据中的系统级"主预测因子"。结果：我们提出了B-MASTER，这是一个结合L1稀疏性与L2分组收缩的可扩展贝叶斯多元回归框架，用于识别关键的跨代谢物调控因子。吉布斯采样器实现了近乎线性的计算扩展性，可支持包含数百万个参数的模型。该方法具有理论保障，包括后验收缩性和选择一致性。通过对结直肠癌微生物组-代谢组数据的分析，揭示了支配整体及癌症相关代谢物模式的关键微生物属，突出了系统级调控结构。可用性：B-MASTER代码（含演示脚本）可在https://github.com/priyamdas2/B-MASTER获取。与本文对应的代码存档快照已通过DOI：10.5281/zenodo.20484958发布在Zenodo平台。