Twilight Zone of Protein Sequence Alignments：（《暮光之城》的蛋白质序列比对）.pdfVIP

Protein Engineering vol.12 no.2 pp.85–94, 1999 Twilight zone of protein sequence alignments Burkhard Rost1,2,3 pairs appear to have less than 12% pairwise sequence identity 1 2 (Rost, 1997). Furthermore, the average sequence identity EMBL, 69 012 Heidelberg, LION Bioscience AG, Im Neuenheimer Feld 517, 69 120 Heidelberg, Germany and 3Columbia University, Department of between all pairs of similar structures is supposedly 8–10%, Biochemistry and Molecular Biophysics, 650 West 168 Street, New York, and the observed distribution (Gaussian peaking around 8% NY 10032, USA identity) marks another region, the midnight zone (Rost, 1997). Sequence alignments unambiguously distinguish between The midnight zone is populated by protein structure pairs that protein pairs of similar and non-similar structure when may have become similar by convergent or divergent evolution the pairwise sequence identity is high (40% for long (Doolittle, 1994; Rost, 1997). Threading algorithms ultimately alignments). The signal gets blurred in the twilight zone of aim at revealing homologous pairs from the midnight zone 20–35% sequence identity. Here, more than a million (Wodak and Rooman, 1993; Bryant and Altschul, 1995; Sippl, sequence alignments were analysed between protein pairs 1995; Rost and Sander, 1996; Sippl and Floeckner, 1996; of known structures to re-define a line distinguishing Fischer et al., 1996; Rost and O’Donoghue, 1997). Conven- between true and false positives for low levels of similarity. tional sequence alignment methods become