Additional file 2 of FastProject: a tool for low-dimensional analysis of single-cell RNA-Seq data DeTomasoDavid YosefNir 2016 Simulation of False-Negative Weight Estimation Procedure. A) A simulation to test FastProject’s ability to distinguish genes that are biologically inactive (True Negatives) from genes dropped out due to technical artifacts (False Negatives). B) The Kolmogorov-Smirnov (KS) statistic is used to distinguish between distributions of true and false negatives. Here it can be seen that the true negatives tend to be assigned higher weights. This is contrasted with the weighting scheme used in a previous study (Gaublomme et al. 2015 [7]) in which the true and false negatives are not as differentiated. C-E) The KS statistic’s for either scheme, as in (B), as simulation parameters are varied. C) Tests varying the parameter which controls the mean of the exponential distribution from which μ is drawn. D) False-negative curves are estimated using a mixture of housekeeping genes and non-housekeeping genes. It can be seen that the choice of good housekeeping genes is beneficial, but not critical. E) The steepness of the generated false-negative curves, α, is varied. (PNG 258 kb)