Skip to main content

Table 1 Empirical type-I error rate and power for bipolar methylation detection

From: Nonparametric Bayesian clustering to detect bipolar methylated genomic loci

Ï„ =0

Type-I error ∗

Power

 

m =10

m =20

m =100

m =10

m =20

m =100

w=10%

.079

.075

.087

.279

.580

.983

w=20%

.090

.077

.082

.590

.875

.997

w=30%

.082

.080

.094

.875

.976

.998

w=40%

.084

.083

.088

.887

.986

.997

w=50%

.085

.088

.088

.931

.995

.998

Ï„ =0 . 32

Type-I error ∗

Power

 

m =10

m =20

m =100

m =10

m =20

m =100

w=10%

.032

.015

.006

.275

.556

.871

w=20%

.034

.015

.004

.528

.771

.984

w=30%

.031

.017

.008

.770

.937

.996

w=40%

.027

.020

.008

.778

.946

.996

w=50%

.025

.016

.006

.782

.951

.997

  1. The empirical type-I error rate and power are calculated from 5,000 simulations under significance level 0.05, for different number of reads m and for different cell-type proportion w. In all simulations, we set the threshold parameter δ= 0.35. The type-I error rates∗ for the same m but different w are not the same because we set different methylation probability vectors for different w when generating data under H 0, although these probabilities are all sampled from beta(8, 8).
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us