Skip to main content

Table 9 Table summarizes the accuracy of ANN, GRU, BILSTM, LSTM, and 1DCNN model for 1560 SNPs

From: Eye-color and Type-2 diabetes phenotype prediction from genotype data using deep learning methods

Model, SNPs = 1560 Activation Dropout Optimizer Batchsize Epochs Validation Accuracy
ANN Sigmoid 0.2 Adam 1 10 0.3 0.94
Sigmoid 0.2 Adam 1 100 0.2 0.93
Sigmoid 0.2 SGD 10 100 0.2 0.945
Relu 0.2 SGD 15 100 0.2 0.94
Sigmoid 0.2 Adam 15 10 0.3 0.94
BILSTM Sigmoid 0.2 SGD 1 100 0.1 0.93
Sigmoid 0.2 SGD 1 30 0.2 0.94
GRU Sigmoid 0.2 SGD 1 50 0.2 0.94
Sigmoid 0.2 Adam 1 20 0.3 0.94
LSTM Sigmoid 0.2 Adam 10 10 0.1 0.945
Sigmoid 0.2 SGD 1 30 0.3 0.93
1DCNN Relu 0.2 RMSprop 15 20 0.2 0.91
Relu 0.2 RMSprop 20 50 0.3 0.9
Relu 0.3 RMSprop 20 50 0.2 0.91
Relu 0.5 RMSprop 20 50 0.1 0.91
  1. ANN and LSTM perform well with an accuracy of 0.945%