Abstract:
Predicting Protein Function is one of the important tasks of bioinformatics in post genomic era. Genome sequencing projects are scientific attempts that ultimately aim to determine the complete genome sequence of an organism. Although these sequences provide us with a lot of information, the functions of many of these are yet to be characterized. Computational biology methods provide powerful tools for this to minimize this sequence-function gap.
Though a large number of methods have been proposed and implemented for predicting protein function, a complete framework which considers all aspects for functional relatedness is missing. A great amount of research is carried out in finding the association based on similarity measures, constructing the phylogenetic tree and comparing them for• phylogeny and assigning weights while finding the association, but still there are insufficient methods that have all the things.
In this - Dissertation entitled "PREDICTING PROTEIN FUNCTION USING PHYLOGENETIC PROFILES ", a solution is proposed which considers the co-evolution of the target genome which gives the basic similarity measure, the background phylogeny of reference genomes for profiles generation and assigning weights to the reference genomes. The ordering of genomes is used to show phylogeny which is computationally feasible.
The proposed strategy can be extended to increasing number of reference genomes. The accuracy of the predictions has been compared with existing approaches and the predictions are validated using the standard dataset. The possibility of using Functional Catalogue database for predicting protein function using Support Vector Machine classifier with radial basis as kernel function is also explored.
