Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: Unsupervised reduction of random noise in complex data by a row-specific, sorted principal component-guided method

Figure 1

A statistical test to determine significant PC-coordinates. The squared PC-coordinates of four example probe sets in the PC-coordinated system are shown. The square-root scale is used for the vertical axis of each plot. (A) The squared PC-coordinates were plotted along the PCs in descending order of their associated variances. The dashed curve shows the variance associated with each PC. (B) The squared PC-coordinates were plotted in descending order of their values after scaling. The dashed curve shows the 50 percentile values of the ordered noise distributions, and the dotted curve shows a p-value of 0.01 (upper tail) for the ordered noise distributions. (C) A close-up view of the left part of (B). The coordinates whose squared values were above the p-value threshold were designated significant. Although the FDR-corrected p-value was used for this statistical test in the actual core RSPR-NR procedure, the uncorrected p-value is used for the sake of explaining the concepts in this figure.

Back to article page