Improved anticancer drug response prediction in cell lines using matrix factorization with similarity regularization

Table 2 The comparison results of different methods obtained under the 10-fold cross validation on CCLE dataset

Methods	Drug-averaged PCC_S/R	Drug-averaged RMSE_S/R	Drug-averaged PCC	Drug-averaged RMSE
SRMF (drug response + gene expression)	0.78 (±0.07)	0.74 (±0.23)	0.71 (±0.09)	0.57 (±0.18)
SRMF (drug response)	0.76 (±0.08)	0.75 (±0.23)	0.69 (±0.09)	0.60 (±0.23)
KBMF	0.65 (±0.10)	0.81 (±0.20)	0.71 (±0.10)	0.64 (±0.17)
DLN	0.71 (±0.06)	0.99 (±0.43)	0.64 (±0.06)	0.86 (±0.42)
RF	0.69 (±0.10)	0.79 (±0.26)	0.62 (±0.11)	0.61 (±0.20)

PCC_S/R—Drug-averaged Pearson correlation for responses from sensitive and resistant cell lines; RMSE_S/R—Drug-averaged root-mean-square error for responses from sensitive and resistant cell lines; PCC—Drug-averaged Pearson correlation for responses across all cell lines; RMSE—Drug-averaged root-mean-square error for responses across all cell lines. The value shown in the bracket represents standard deviation

ISSN: 1471-2407