Skip to main content

Table 1 AIC and BIC values, computed using gamlss, for a number of non-inflated and inflated distribution models of the null probability density function of the T statistic, along with their (rounded) effective degrees of freedom (DF)

From: Ranking genomic features using an information-theoretic measure of epigenetic discordance

 

Non-inflated Model

Inflated Model

Model

AIC

BIC

DF

AIC

BIC

DF

Promoter

 BE

-486,026

-485,855

17

-486,229

-486,041

19

 GB1

-488,175

-487,961

22

-488,408

-488,181

23

 logitNO

-487,985

-487,895

9

-492,544

-492,384

16

 logitTF

-493,963

-493,792

18

-493,944

-493,762

19

 logitSST

-494,260

-494,019

25

-494,242

-493,992

26

Gene body

 BE

-482,633

-482,474

17

-482,608

-482,440

18

 GB1

-484,822

-484,615

27

-484,798

-484,581

23

 logitNO

-490,890

-490,730

17

-490,872

-490,702

18

 logitTF

-495,777

-495,594

19

-495,751

-495,560

20

 logitSST

-498,362

-498,133

24

-498,337

-498,099

25

  1. The results were obtained by fitting each model to T statistic values within promoter regions and gene bodies obtained from all six reference/reference comparisons. Entries in bold highligth the proposed model, which is shown here to produce the best AIC and BIC scores
  2. BE: beta distribution; GB1: generalized beta type 1 distribution; logitNO: logit normal distribution; logitTF: logit t-family distribution; logitSST: logit skewed Student’s t distribution