Computational Method to Predict Mitochondrially Imported Proteins and their Targeting Sequences

TLDR

Mitochondrial proteins are imported across the organelle’s double membrane, guided by sequence and structural cues that lack direct experimental evidence. The authors applied discriminant analysis using 47 parameters to a large SwissProt-derived set of mitochondrial proteins. Their computational method predicts 75–97 % of mitochondrial proteins as imported and identifies targeting sequences in 76–94 % of precursors, revealing many uncharacterized yeast ORFs likely encode mitochondrial proteins.

Abstract

Most of the proteins that are used in mitochondria are imported through the double membrane of the organelle. The information that guides the protein to mitochondria is contained in its sequence and structure, although no direct evidence can be obtained. In this article, discriminant analysis has been performed with 47 parameters and a large set of mitochondrial proteins extracted from the SwissProt database. A computational method that facilitates the analysis and objective prediction of mitochondrially imported proteins has been developed. If only the amino acid sequence is considered, 75–97% of the mitochondrial proteins studied have been predicted to be imported into mitochondria. Moreover, the existence of mitochondrial‐targeting sequences is predicted in 76–94% of the analyzed mitochondrial precursor proteins. As a practical application, the number of unknown yeast open reading frames that might be mitochondrial proteins has been predicted, which revealed that many of them are clustered.