Automation of Gene Expression Profile Analysis in Single Cell Data

We designed and implemented an automated system, named Pierse for pattern recognition of single cell transcriptomics (SCT) data. The Pierse system takes sparse matrices and corresponding metadata as input to generate SCDC profiles (SCT gene expression profiles characteristic of types or subtypes of cells). These profiles can be used for profile comparison, feature extraction, and differential gene expression analysis. Hierarchical clustering is used for similarity analysis between SCDC profiles and resulting heatmaps are produced. We performed a demonstration study to test functional modules in the Pierse system. To improve efficiency, we deployed parallel programming scripts and implemented efficient matrix analysis functions in the demonstration study.