Avocado: a multi-scale deep tensor factorization method learns a latent representation of the human epigenome

Table 1 Evaluation of ChromImpute, PREDICTD, and Avocado

MSE-	Global	1obs	1imp	Prom	Gene	Enh
ChromImpute	0.113	0.941	1.09	0.325	0.149	0.316
PREDICTD	0.100	1.76	0.897	0.258	0.129	0.267
Avocado	0.100	1.66	0.845	0.249	0.130	0.260

Six performance measures are reported, reflecting MSE of different subsets of genomic positions. The best result for each metric is in boldface and corresponds to an adjusted two-sided paired t test p value <0.01 when compared to both other approaches. For MSE-global and MSE-Gene, both PREDICTD and Avocado are bolded because the difference between the two is not statistically significant, i.e., has a p value > 0.01

ISSN: 1474-760X