Skip to main content

Table 6 Learned hyperparameters of Auto-HMM-LMF method based on GDSC dataset

From: Auto-HMM-LMF: feature selection based method for prediction of drug response via autoencoder and hidden Markov model

Hyperparameter

Description

Value

k

Number of nearest neighbors (Eq. 9)

20

α

Effectiveness of cell line similarity (Eq. 11)

0.5

β

Effectiveness of drug similarity (Eq. 11)

0.1

λc

Variance parameter of cell lines (Eq. 11)

0.5

λd

Variance parameter of drugs (Eq. 11)

0.5

λ

Importance of SimEXP (Eq. 8)

2

γ

Importance of SimCNV (Eq. 8)

2

ϕ

Importance of SimMUT (Eq. 8)

2

ψ

Importance of SimIC50 (Eq. 8)

5

ρ

Importance of SimTISSUE (Eq. 8)

2

threshold

Threshold parameter

0.4

  1. The parameter k were selected from 1 to 50. The impact factors of nearest neighbors α and β in equations were selected from {2–5, 2–4, …, 22}. The variance parameters, λc and λd, were chosen from {2–5, 2–4, …, 21}. The five parameters γ, λ, ϕ, ψ, and ρ were selected from 1 to 5