Identification of RNA Binding Proteins and RNA Binding Residues Using Effective Machine Learning Techniques

Identification and annotation of RNA Binding Proteins (RBPs) and RNA Binding residues from sequence information alone is one of the most challenging problems in computational biology. RBPs play crucial roles in several fundamental biological functions including transcriptional regulation of RNAs and RNA metabolism splicing. Existing experimental techniques are time-consuming and costly. Thus, efficient computational identification of RBPs directly from the sequence can be useful to annotate RBP and assist the experimental design. Here, we introduce AIRBP, a computational sequence-based method, which utilizes features extracted from evolutionary information, physiochemical properties, and disordered properties to train a machine learning method designed using stacking, an advanced machine learning technique, for effective prediction of RBPs. Furthermore, it makes use of efficient machine learning algorithms like Support Vector Machine, Logistic Regression, K-Nearest Neighbor and XGBoost (Extreme Gradient Boosting Algorithm). In this research work, we also propose another predictor for efficient annotation of RBP residues. This RBP residue predictor also uses stacking and evolutionary algorithms for efficient annotation of RBPs and RNA Binding residue. The RNA-binding residue predictor also utilizes various evolutionary, physicochemical and disordered properties to train a robust model. This thesis presents a possible solution to the RBP and RNA binding residue prediction problem through two independent predictors, both of which outperform existing state-of-the-art approaches