Enhancement and Cloud Deployment of CaDrA, a software tool for Candidate Driver Analysis of Multiomics Data

我们的目标是进一步开发、优化、文档化、容器化并交付一个R包CaDrA（候选驱动程序）

We aim to further develop, optimize, document, containerize and deliver as an R package CaDrA (Candidate Driver Analysis), an open-source, user-friendly computational tool for the analyses of cancer multi-omics datasets. A prototype of the package already exists, and was applied in studies published in high-impact journals. The under- lying methodology was evaluated with simulated and real data, and it was shown to have high sensitivity and specificity. The methodology was developed in part under the auspices of a NIDCR-sponsored F31 (awardee, Vinay Kartha), with the associated publications receiving considerable attention, and we received several requests to share the code and clarify its use. We plan to utilize this analytical approach in the context of our parent grant (NIDCR MPI 1R01DE030350-01A1), aimed at investigating, through experimental and in-silico approaches, the β- catenin/CBP axis in head and neck cancer. An expanded and optimized tool will support the analysis of bulk and single cell RNAseq data we are generating as part of that grant, as well as the querying of public datasets, such as TCGA, CPTAC, CCLE and others, based on transcriptional signatures derived from our own generated data. While some analysis methods that can integrate and interpret multiple experimental and molecular data types exist, they tend to either provide for highly interactive interfaces but relatively basic analytic functionalities, or for sophisticated analysis methods not adequately supported by well-documented and user-friendly tools. Our proposed tool occupies the “sweet spot” in between and will provide for advanced statistical techniques, a user- friendly interface, and a well-documented, open-source R package, which will be available to the research community as a stand-alone tool, or “integratable” into other tools and analytical workflows, and will facilitate reproducibility and transparency of the analyses performed. To facilitate the tool’s wider adoption, applicability, and scalability, its robustness, portability, and user-friendliness need to be enhanced. This proposal would allow us to pursue it extension with new functionalities, its enhancement and optimization, documentation, and containerization into an open-source package made available through GitHub, and installable through CRAN or Bioconductor. Our work plan includes extension of the statistical functionalities supported, code cleaning, profiling, testing and optimization to improve run-time efficiency, adoption of data structures and constructs compliant with Bioconductor guidelines, packaging to ensure easy code installation and testing, containerization to support cloud deployment, development of a graphical interface to increase the tool’s user-friendliness, and design of a web- based lending page to be hosted on our university server. Relevance and Impact. Availability of the tool will be instrumental to the pursuit of our parent grant aims. In parti- cular, it will support the analysis of our own generated bulk and single cell RNAseq data from mouse models and human samples. R packaging and containerization will also facilitate adoption by the research community.