cGLMM

"cGLMM" ( collaborative Generalized Linear Mixed Model) is an R package to address privacy concerns in collaborative genome-wide association studies (GWAS) involving large cohorts of patients across multiple institutions. GWAS aims to identify genetic variants significantly associated with phenotypes, such as diseases. Still, collaborative efforts in this area often face challenges due to the sensitive nature of sharing personal genomic and health data.

Key Features: - Privacy-Preserving Approach: cGLMM introduces a privacy-preserving Expectation-Maximization (EM) algorithm that allows for the construction of a GLMM without requiring the transfer of individual data sets to a central server. This approach ensures that the privacy of patient data is maintained while still allowing for collaborative analysis. - Data Partitioning: The tool is designed for scenarios where data are horizontally partitioned among participating parties, meaning each party holds a subset of records. Despite the distribution, all records are considered under the same set of known fixed and random effects.

- Mathematical Equivalence and Efficiency: The collaborative EM algorithm implemented in cGLMM is mathematically equivalent to the traditional EM algorithm used in GLMM construction, ensuring the reliability of the analysis. It has also been tested for efficiency on simulated and real human genomic data, proving its practical utility for GWAS analysis.

- Implementation and Communication: cGLMM is implemented in R, with data communication facilitated by the rsocket package, ensuring an efficient and secure exchange of information between parties involved in the collaborative analysis.