Skip to main content
Fig. 4 | Biology Direct

Fig. 4

From: Finite-size effects in transcript sequencing count distribution: its power-law correction necessarily precedes downstream normalization and comparative analysis

Fig. 4

Post power-law correction, rank-frequency plots of NUGC3 dilution and spike-in background datasets. a, b, c, d and e show the rank-frequency plots for the 1.5p pair, 3p pair, 6p pair, single 12p replicate and the 11 UHR replicates against the best estimate of the original signals (marked in black) after the power-law correction. The observed alias noise (marked in blue) and the theoretical alias noise S o (f s  − f)α (marked in magenta), are also shown. In each subplot, the sampling frequency f s and the mean square error (MSE is defined as the residual error between the observed and theoretical alias noise) are given as well. The overall low MSE values of between 6.00e-4 to 1.87e-3 indicates a good fit between the theoretical model and the observed alias. Generally speaking, the corrected datasets shows a general absence of undersampling. For all plots, the observed alias noise (marked in blue), as well as the theoretical alias noise S o (f s  − f)α (marked in magenta), shows very slight aliasing in all cases given their new sampling frequencies of 1720, 1311, 1783, 3315 and 1920 respectively

Back to article page