The Challenge of Protein Domain Annotation with Supervised Learn-ing Approach: A Systematic Review

The protein domain is still considered one of the most critical evolutionary unit in the cellular, molecular mechanism. Hence, providing the quantitative measurements of the domain contents of proteins would be essential to give insights on the dynamics of the cell’s biochemical machinery. Bioinformatics, as a quantitative science, has successfully provided several approaches to comprehend the domain contents and their dynamics. It was done mainly with supervised learning approach. In eukaryote domain, there is the tendency of transcription factor domain avoidance in the higher organism, and co-occurrence in a single cell has shed light on the complexity of domain functionality. It is widely assumed that protein domain tends to avoid each other as the organism gained more sophisticated molecular features. However, more sample organisms should be provided to obtain better insight on the domain co-occurrence in the cells. This systematic review was conducted by searching literature in Google Scholars and PubMed. To this end, automatic pipelines should be provided by software packages such as DOMOSAIC with the help of the R-based scripting to uphold the statistical significance. The latest updates on this topic are covering annotation on the orphan domain in Drosophila, and architecture plasticity in the eukaryote.