Low molecular weight proteins: a challenge for post-genomic research.

The EcoGene project involves the examination of Escherichia coli K-12 DNA sequences and accompanying annotation in the public databases in order to refine the representation and prediction of the entire set of E. coli K-12 chromosomally encoded protein sequences. The results of this ongoing effort have been deposited in the SWISSPROT protein sequence database as sequencing of the E. coli genome has progressed to completion in recent years. Through this continuing research, we have discovered that the prediction of low molecular weight (small) proteins, arbitrarily defined as protein sequences < or = 150 amino acids (aa) in length, is problematic and requires special attention. We describe the small protein subset of EcoGene and the approach used to derive this subset from the complete E. coli genome sequence and database annotations. These E. coli proteins have helped to identify new small genes in other organisms and to identify conserved residues (motifs) using database searches and multiple alignments. Two thirds of the E. coli small proteins have not been characterized experimentally. The careful application of computer and laboratory methods to the analysis of small proteins is needed for accurate prediction, verification and characterization. The problem of accurate protein sequence identification is not limited to small proteins or to E. coli; these problems are encountered to varying degrees throughout all sequence databases.