Table 1

Definition of the seven scoring functions used in PIA v. 2.0.

Metric

Description

Formulaa


1

%Correct

(NTP + NTN)/(NTP + NFN + NFP + NTN)

2

Sensitivity + Specificity

[NTP/(NTP+NFN)] + [NTN/(NFP+NTN)]

3

Positive Predictive Value (PPV)+ Negative Predictive Value (NPV)

[NTP/(NTP+NFP)] + [NTN/(NFN+NTN)]

4

Risk Ratio

[(NTP)(NFP+NTN)]/[(NFP)(NTP+NFN)]

5

Odds Ratio

[(NTP)(NTN)]/[(NFP)(NFN)]

6

Gini Indexb

GINIparent - GINIsplit

GINI(k) = 1.0 - ∑j = 1, J [p(j|k)]

GINIsplit = ∑k = 1, K [(nk/n) GINI(k)]

7

Absolute Probability Differencec

Σk = 1, K |P1(k) - P2(k)|


a NTP, Number of true positives; NTN, number of true negatives; NFN, number of false negatives; NFP, number of false positives.

b Gini Index is used in CART decision trees [25]. The scoring for Gini index is described under "Algorithm."

c Scoring is the probability of finding a case (P1) in cell, k, minus the probability of observing a control (P2) in cell, k, summed over all the K cells in the genotype-phenotype table.

Mechanic et al. BMC Bioinformatics 2008 9:146   doi:10.1186/1471-2105-9-146

Open Data