Comparison of bioinformatics algorithms and tools related to RNA-Protein interactions

Proteins and nucleic acids play an important role in the biological diversity of living organisms, and each has its structural and functional characteristics. Therefore, RPI (RNA-Protein Interaction) prediction is of significant importance. Protein-RNA interactions (RPI) play a key role in most cellular processes such as transcription, reverse transcription, post-transcription processing, replication, RNA transmission, translation, and regulation of RNA values in the cell. Various high-power laboratory methods are used to identify RPIs that generate valuable data, but these tests are costly, and time-consuming. Therefore, various computational methods have been developed to study, analyze, and build RPI networks. In this regard eight important algorithms (PRINTR, PLPIHS, XGBPRH, LPI-IBNRA, Xpred RBR, Struct-NB, SRC PRED, and RPiRLS) were investigated. Assessment of accuracy indices using LooCV and ۱۰-fold-cross validation showed that LPI-IBNRA algorithm, with accuracy (۸۸%) and percision (۸۷%) in the ۴۷۹۶ data set known protein-lncRNA interaction from NPInter V۲.۰ database is more powerful than other comparable algorithms.Finally, it should be noted that some important and unsolved biological problems in this field are related to the specific binding of a protein to its target RNA. Therefore, creating algorithms for the role of RPI monitoring in diseases,identifying RPI in viruses, introducing more suitable features for use in RPI prediction algorithms, determining RPI patterns in mitochondria, and chloroplasts of plants and using combination methods and deep machine learning, and has been proposed to produce more powerful bioinformatics tools to predict RPI.

Keywords:

Algorithms , accuracy indicators , software , computational methods , machine learning