The large amounts of available protein structures emerges the need for computational methods for protein function prediction. Predicting protein function is mainly based on finding similarities between proteins with unknown function with already annotated proteins. This may be achieved using different protein characteristics: sequences, interactions, localization, structure and or psychochemical. A lot of review papers mainly focus on sequence and psychochemical featuresbased methods. This is because sequence and psychochemical data are easy to deal with and to interpret the results, and much available compared to protein structures. However, structure-based computational methods provide additional accuracy and reliability of protein function prediction. Therefore, unlike many review papers, this paper presents an up-to-date review on the structure-based protein function prediction. The aim was to provide a recent and comprehensive review of protein structure related topics: function aspects, structural classification, databases, tools and methods.

Protein function prediction is an active research area in bioinformatics. Protein functions are highly related to their structures. Therefore, effective structure based protein representations are required. Pires et al. [BMC Genomics, 12, S12 (2011)] proposed a cutoff scanning matrix (CSM) method for protein representation that utilizes distance patterns between protein residues and a maximum cutoff. This paper proposes a modified cutoff scanning matrix (MCSM) representation for enhancing protein function prediction. The proposed representation considers the whole protein instead of using cutoff. A comparative analysis was done to evaluate the proposed MCSM method and the original CSM method. Two different classification algorithms, Random Forest and K-nearest neighbor (KNN), were used in the analysis. The aspect of protein function considered is based on enzyme activity. The results show that the proposed MCSM representation outperforms the CSM representation with a prediction accuracy of 90.12% and 80.27% for superfamily and family level, respectively, with accuracy improvement of about 5% on average.

No abstract