- The solution was delivered by our Bioinformatics team. We used our expertise in Bioinformatics, genome-wide screening, R, Python, and machine learning. We evaluated existing approaches and advised on and evaluated methodology using publicly available and proprietary multi-omics data to ensure the results were scientifically sound.
- We developed an R package that handles user input and output while leveraging automated conda environment calls to seamlessly handle operations in Python modules. Calls to Python were implemented using reticulate and conda environments were managed using basilisk.
- We built a machine learning model based on previously published scientific literature to predict sgRNA efficiency and efficacy. The trained model utilizes genetics and epigenetics information obtained from multi-omics datasets to make these predictions.
- We utilized ground truth data to develop, train, and test the machine learning model for predictive scoring of sgRNA candidates.
- We delivered an R package that follows Bioconductor guidelines and best practices.
- The complete package, including trained model and Python components, were delivered in a portable, easy-to-use R package.
- Due to the novelty of the tool, we are working with the client to prepare the tool for scientific publication and Bioconductor submission to make it available to the wider community.