GOstruct 2.0: Automated Protein Function Prediction for Annotated Proteins

摘要: Automated Protein Function Prediction is the task of automatically predicting functional annotations for a protein based on gold-standard derived from experimental assays. These experiment-based accumulate over time: proteins without get annotated, and new functions already annotated are discovered. Therefore, function prediction can be considered combination two sub-tasks: making predictions previously unannotated proteins. In previous work, we analyzed performance several methods in these scenarios. Our results showed that GOstruct, which structured output framework, had lower accuracy with existing annotations, while its un-annotated was similar to cross-validation. this present GOstruct 2.0 includes improvements allow model make use information protein's current better handle novel This highly important organisms where most have some level annotations. Experimental human data show outperforms original task, demonstrating effectiveness proposed improvements. first study focuses adapting framework applications labels incomplete by nature.