Skip to main content

Table 2 Comparison of prediction accuracy performances of LMM, BSLMM, BayesR, and KAML by using seven case/control diseases in the WTCCC1 dataset

From: KAML: improving genomic prediction accuracy of complex traits using machine learning determined parameters

Traits/AUROC

Methods

LMM

BSLMM

BayesR

KAML

CAD

0.586 (0.0043)

0.599 (0.0041)

0.602 (0.0039)

0.600 (0.0041)

HT

0.597 (0.0038)

0.596 (0.0038)

0.596 (0.0041)

0.594 (0.0038)

T2D

0.600 (0.0032)

0.618 (0.0032)

0.620 (0.0034)

0.618 (0.0031)

BD

0.641 (0.0033)

0.641 (0.0034)

0.647 (0.0033)

0.638 (0.0035)

CD

0.628 (0.0039)

0.669 (0.0041)

0.668 (0.0046)

0.669 (0.0039)

RA

0.614 (0.0034)

0.704 (0.0032)

0.708 (0.0033)

0.717 (0.0037)

T1D

0.646 (0.0045)

0.858 (0.0019)

0.861 (0.0018)

0.862 (0.0019)

Average

0.616

0.669

0.672

0.671

  1. Prediction accuracy performance was measured by the area under the ROC curve (AUROC). For prediction assessment, total samples were split to two subsets: 80% of the samples were used as the reference dataset and 20% were used as the validation dataset; the procedure was repeated 20 times; and the mean AUROC values and standard deviations of each trait are shown in the table
Back to article page

Genome Biology

ISSN: 1474-760X

Contact us