Plant-mSubP: a machine learning tool to predict single- and multi-target protein subcellular localisation

Plant cells are three-dimensional spaces composed of several compartments, each with its own physicochemical environment and function. The subcellular localisation of proteins, the cell’s functional machinery, is very important for characterizing their functions in a cell. Improper localisation can result in disease and cell death. Thus, predicting the subcellular localisation of proteins is a hot topic in proteomics, yet doing so through biochemical experiments can be laborious, expensive, and time-consuming. Computational prediction can be an effective alternative allowing us to decipher protein function and speed up genome annotation.

In a new article published in AoBP, Sahu et al. present their new software, Plant-mSubP, which is a publicly available web platform that uses machine learning to predict the subcellular localisation of proteins based on amino acid sequence features. Plant-mSubP was developed using training data sets of known subcellular localisation, resulting in over 16000 unique protein sequences annotated to 14 different subcellular locations. Using an independent data set for each localisation class, the authors then compared their method with other available sequence-based prediction tools and found it to outperform all existing methods. They highlight that until now very limited work has been conducted on the subcellular localisation of plant proteins but hope that their new web accessible tool will begin to address this.

Plant-mSubP can be accessed online at http://bioinfo.usu.edu/Plant-mSubP/

Researcher highlight

Rakesh Kaundal grew up in India and in 2007 moved to USA to conduct a Postdoctoral Research Fellowship in bioinformatics & computational biology at the Noble Research Institute, Ardmore, Oklahoma. Rakesh currently holds an Assistant Professor position in the Department of Plants, Soils, and Climate at Utah State University (USU), and as the Director of Bioinformatics Facility in the Center for Integrated BioSystems, USU. He also has an adjunct professor appointment in the Department of Computer science.

Rakesh has developed an independent and collaborative research program in bioinformatics, primarily focusing on computational mining of large multi-dimensional -omics datasets, computational modeling using supervised (Machine Learning) and unsupervised (Bayesian-based) learning, and is actively developing novel tools and software to apply the gained knowledge towards organismal improvement.