Skip to main content

Table 4 Imputation performance of GNNImpute and GCN architecture model (PBMC dataset)

From: An efficient scRNA-seq dropout imputation method using graph attention network

 

MSE

MAE

PCC

CS

GCN (without attention)

3.3047

0.8022

0.9436

0.9510

GNNImpute (1 attention head)

2.8800

0.7736

0.9478

0.9547

GNNImpute (3 attention heads)

2.7767

0.7684

0.9494

0.9561

GNNImpute (5 attention heads)

2.8748

0.7730

0.9480

0.9550

GNNImpute (8 attention heads)

2.8494

0.7699

0.9482

0.9551

  1. The bold values indicate the best or better scores that can be obtained through different methods under different indicators
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us