MAUDE: inferring expression changes in sorting-based CRISPR screens

摘要

MAUDE approach to scoring expression-based screens. Method overview. – MAUDE approach to estimating the optimal mean expression level per guide. Estimation of cell density distribution for each guide. Cell density distribution ( axis) of target gene expression ( axis) for each guide is modeled as a normal distribution, with mean . is optimized by calculating, for each bin (bins A and B, top), the fraction of cells expected to be in the bin under the null model (no overall change in expression; ( , 0)) and the fraction of cells expected in the bin given the current value of ( ( ,  )). Estimation of optimal . We find the optimal by calculating the log likelihood ( axis) of the observed bin read abundances given each value of ( axis). Expected number of guide reads per bin. The fraction of reads ( axis) of guide ( ( |  )) expected in each bin ( axis), for different values of (colors)