Improved computational prediction of prokaryotic protein subcellular localization

摘要

@@BackgroundComputational prediction of protein subcellular localization for bacteria has reached a level of precision comparable to certain highthroughput experimental determination techniques. The Brinkman lab has previously developed PSORTb, capable of assigning a query bacteria protein to one of four Gram-positive or five Gram-negative localization sites with a measured overall precision of 97%; however,due to its lower recall, it only makes predictions for approximately 50% of proteins encoded in typical Gram-negative bacterial genomes and 75% of those in Gram-positive bacterial genomes.