Comprehensive analysis of transmembrane protein sequences in 39 microbial genomes

摘要

The genome-wide analysis of transmembrane (TM) proteins (the TM protein proportion in proteome, the distribution of TM topology, comprehensive functional identification of TM proteins, etc.) has been tried recently by using TM topology prediction methods. The performance of these prediction methods is, however, not so high enough, as confirmed by our re-assessment of the prediction performance by using a dataset of experimentally-characterized TM topologies (202 entries): HMMTOP 2.0 is the highestwith only 57.3 percent accuracy in predicting TM topology for prokaryotic sequences. In these analyses, the straight treatment of signal peptides (SPs) is usually avoided: e.g., the sequences with probable SP are excluded from the analysis. This preventsus from estimating accurately the TM protein proportions in proteomes. And also, soluble protein sequences with SP axe easily predicted as single-spanning TM protein. In this study, we carried out genome-wide analysis of prokaryotic TM proteins (39 genomes) by using the "consensus TM topology prediction method" (67.7 percent accuracy with TM topology prediction) and by treating the SPs properly.