Project Summary
Estimating biological variability first requires a quantitative model of technical variability so these two sources can be decomposed in RNA-seq data. Current methods to estimate technical variability leverage the use of spike-ins, UMIs or inbuilt controls which can add cost to already expensive experiments. The holy grail of single cell bioinformatics would be to derive a way to estimate or infer technical variability in single cell RNA-seq data without appealing to these design elements. The use of Bayesian hierarchical models suggest that these kinds of probabilistic models contain value in this arena and are under-utilized in the current space of bioinformatics approaches. Development of these methods would have widespread impact for quality control of single cell data arising from stem cells, tissue engineering, and disease models.
