Network based subcellular localization prediction for multi-label proteins

Abstract

Many proteins are sorted to multiple subcellular localizations within the cell. However, computational prediction of multi-location proteins remains a challenging task. Here we applied a logistic regression and diffusion kernel based algorithm NetLoc for predicting multiplex proteins and explored its capability and limitations. Experiment shows that the overall and true success rates for physical protein-protein interaction network are 65% and 41% respectively, and for mixed PPI network these values are 88% and 75% respectively. Our study also showed that the performance of NetLoc in predicting protein localization is limited by the network characteristics such as ratio of the number of co-localized protein-protein interactions (coPPI) to the number of non-co-localized PPI (ncPPI) and the density of annotated coPPI in the network. For a given network with a specific number of proteins, NetLoc performance increases with higher coPPI/ncPPI ratio and higher density of annotated coPPI.

Scored Protein-Protein Interaction to Predict Subcellular Localizations for Yeast Using Diffusion Kernel

“…In their previous work [7,[10][11][12] Mondal and Hu used NetLoc model to predict subcellular localization using PPI network without score. In the present work, we used NetLoc to explore it's capability of similar prediction but using scored PPI from STRING database [13].…”

“…Where is the sum of weights of interactions with protein i and represents the matrix exponential of the matrix . It is noticeable that , 1 represents the diffusion kernel for non-score based PPI network or PPI with score of unity, which is used in [7,[10][11][12]. Kernel function ( , ) K i j represents the similarity distance between protein i and protein j in the network.…”

“…For classification, we applied the diffusion kernel-based logistic regression (KLR) model [14] as used in [7,[10][11][12] to predict protein subcellular localization. Given a protein-protein interaction network with proteins , … , with of them , … , with unknown subcellular locations, the objective is to find subcellular locations of unknown proteins using the locations of known proteins and proteinprotein interaction network.…”

“…In their previous work [7,[10][11][12], Mondal and Hu explored protein localization prediction using non-scored PPI networks, namely, COEXP, GPPI, and PPPI. In order to compare we need two networks composed of same number of proteins and same PPIs, one with PPI score and the other without PPI score.…”

“…One of the limitations of these methods that these are not capable of utilizing the inherent network information that naturally appears among proteins [7,10]. In their recent work, Mondal and Hu [7,[10][11][12] exploit the PPI network information in predicting subcellular protein localization using diffusion kernel. But these studies do not utilize the PPI score meaning they used score of unity for each PPI.…”