Skip to main content

Table 4 The comparison of prediction performances on the 16th QTL-MAS Workshop dataset

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

Traits/accuracy

Non-genotyped (n = 100)

Genotyped (n = 1000)

PBLUP

SSBLUP

SSKAML

PBLUP

LMM

SSBLUP

KAML

SSKAML

T1

0.7823

0.8648

0.8819

0.4549

0.7317

0.7322

0.8011

0.7969

T2

0.7954

0.8726

0.8901

0.5331

0.7708

0.7707

0.8433

0.8378

T3

0.7401

0.8238

0.8418

0.4704

0.7581

0.7579

0.8324

0.8262

  1. PBLUP pedigree information-based BLUP method, LMM genomic information-based BLUP method, SSBLUP single-step BLUP method. For SSBLUP, the weight of pedigree-based relationship matrix for genotyped individuals was 0.05. The diagonal and off-diagonal of both pedigree and genomic relationship matrices were adjusted to the same scale
Back to article page

Genome Biology

ISSN: 1474-760X

Contact us