Fig 1.
RNA-Seq classification workflow.
Table 1.
Description of real RNA-Seq datasets used in this study.
Fig 2.
Genewise dispersion estimations for real datasets.
Fig 3.
Simulation results for k = 2, dkj = 10%, transformation: rlog.
Figure shows the performance results of classifiers with changing parameters of sample size (n), number of genes (p) and type of dispersion (φ = 0.01: very slight, φ = 0.1: substantial, φ = 1: very high).
Fig 4.
Simulation results for k = 3, dkj = 10%, transformation: rlog.
Figure shows the performance results of classifiers with changing parameters of sample size (n), number of genes (p) and type of dispersion (φ = 0.01: very slight, φ = 0.1: substantial, φ = 1: very high).
Fig 5.
Results obtained from real datasets.
Figure shows the performance results of classifiers for datasets with changing number of most significant number of genes. Note that PLDA and NBLDA methods are not performed on the transformed data. However, the results for both transformed and non-transformed data are given in the same figure for the comparison purpose.