SGTCDA: Prediction of circRNA-drug sensitivity associations with interpretable graph transformers and effective assessment.

ABSTRACT: CircRNAs are a type of circular non-coding RNA whose associations with drug sensitivities have been demonstrated in recent studies. Due to the high cost of biomedical experiments for detecting the associations between circRNAs and drug sensitivities, several computational methods have been developed. However, these methods were evaluated mainly based on 5- or tenfold cross-validation, which are often over-optimistic. Furthermore, there are technique issues with these models, such as over-smoothing and over-squashing. To address these issues, we propose a strategy to evaluate models based on independent test sets for association prediction-related studies. In the light of this effective assessment, we constructed a model, SGTCDA, by integrating structural deep network embedding (SDNE) and a

SUBMITTER: Xia H 

PROVIDER: S-EPMC11577602 | biostudies-literature | 2024 Nov

REPOSITORIES: biostudies-literature