A fair experimental comparison of neural network architectures for latent representations of multi-omics for drug response prediction

Table 2 Mean AUPRC on the test sets from cross-validation

Drug	Omics stacking	MOLI	Super.FELT	Early integration	OmiEmbed	MOMA	PCA
Gemcitabine TCGA	0.161 ± 0.053	0.155 ± 0.065	0.108 ± 0.024	0.117 ± 0.025	0.132 ± 0.072	0.138 ± 0.041	0.166 ± 0.084
Gemcitabine PDX	0.151 ± 0.085	0.154 ± 0.066	0.130 ± 0.050	0.138 ± 0.046	0.082 ± 0.024	0.111 ± 0.027	0.141 ± 0.054
Cisplatin	0.293 ± 0.089	0.316 ± 0.084	0.282 ± 0.052	0.222 ± 0.065	0.204 ± 0.086	0.262 ± 0.075	0.710 ± 0.072
Docetaxel	0.316 ± 0.083	0.345 ± 0.093	0.373 ± 0.027	0.312 ± 0.117	0.251 ± 0.071	0.281 ± 0.090	0.278 ± 0.040
Erlotinib	0.479 ± 0.153	0.446 ± 0.108	0.499 ± 0.162	0.346 ± 0.120	0.468 ± 0.162	0.476 ± 0.151	0.347 ± 0.066
Cetuximab	0.376 ± 0.104	0.357 ± 0.075	0.400 ± 0.095	0.290 ± 0.057	0.329 ± 0.091	0.347 ± 0.070	0.259 ± 0.064
Paclitaxel	0.220 ± 0.108	0.163 ± 0.087	0.245 ± 0.116	0.160 ± 0.045	0.270 ± 0.079	0.213 ± 0.074	0.118 ± 0.049
Mean Rank	2.29	2.86	2.86	5.57	5.14	4.29	5.00

Best results are shown in bold and second best are italic. The values represent the means and standard deviations over five iterations

ISSN: 1471-2105