An Automated Framework For Evaluation Of Deep Learning Models For Splice Site Predictions

This article presents a novel framework for the automated evaluation of various deep learning-based splice site detectors. The framework eliminates time-consuming development and experimenting activities for different codebases, architectures, and configurations to obtain the best models for a given RNA splice site dataset. As a case study, the article compares CNN and BLSTM models’ learning capabilities as building blocks for RNA splice site prediction in two different datasets. The results show that CNN performed better with accuracy, F1 score, and AUC-PR in human splice site prediction.