Similarity and Prioritization of Disease Proteins using Path Length Measure

Semantic similarity measures have been used successfully and extensively in the biomedical research with various applications. As the biomedical ontologies, which form the main ground for most of the similarity measures, are growing and progressing towards more completeness and higher accuracy, the results and outcomes of these semantic similarity measures become more acceptable and more reliable in the field. In this paper, we investigate a path length based measure for prioritization of disease proteins and for computing the similarity between diseases and proteins. Our measure is based on the GO annotation terms of the proteins and uses a simple exponential transfer function to convert the path length to similarity score. The evaluation results prove that this similarity measure is fairly effective in assessing the closeness of proteins and diseases in the disease protein ranking and protein prioritization experiments.