Skip to main content
Fig. 2 | BMC Genomics

Fig. 2

From: Sort-seq under the hood: implications of design choices on large-scale characterization of sequence-function relations

Fig. 2

Bias and efficiency of mean fluorescence estimates. a Histograms of input fluorescence distributions for three variants with different cell-to-cell variability. Shaded regions indicate the 6 log-spaced sort gates (w=0.8). Dashed lines bound the two outer gates. b Relative bias and c efficiency of the simple mean estimator from sort-seq simulations, plotted against the input mean for variants with different levels of σ. d Gate configuration for MLEs, which feature semibound gates that capture all cells with fluorescence above or below thresholds on the right and left, respectively. e Relative bias and f efficiency plotted for the MLE mean as in bc. Simulations used N=100 sort-seq reads per repeat, averaging over 1000 repeats per set of parameter values. (G) A sort-seq dataset [8] is used to infer mean and CV using MLEs for 5255 yeast promoters. Inferred CV is used to define variants with high CV (σ>0.35, blue) and low CV (σ<0.20, red). Shaded regions indicate original sort gates. h Sort-seq gates were re-grouped by combining reads corresponding to adjacent gates, resulting in larger gate width w. ij Relative change between the estimates of the mean using the re-grouped data and the full data. i 16 gates (w≈0.25) and j 8 gates, (w≈0.65). Lower panels indicate median absolute change. Shades indicate re-grouped gates

Back to article page
\