SOPM: a self-optimized method for protein secondary structure prediction

• DOI: 10.1093/protein/7.2.157
• Publication Date: 2007-01-05T04:15:27Z
• Authors: C. Geourjon; G. Deléage
• Keywords: Sequence (biology); Protein Data Bank

A new method called the self-optimized prediction (SOPM) has been developed to improve success rate in of secondary structure proteins. This checked against an updated release Kabsch and Sander database, 'DATABASE.DSSP', comprising 239 protein chains. The first step SOPM is build sub-databases sequences their known structures drawn from 'DATABASE.DSSP' by (i) making binary comparisons all (ii) taking into account structural classes second submit each sub-database a using predictive algorithm based on sequence similarity. third iteratively determine parameters that optimize quality whole sub-database. last apply final query sequence. correctly predicts 69% amino acids for three-state description (a helix, (3 sheet coil) database (46 Oil acids). correlation coefficients are Ca=0.54, Q=0.50 Cc=0.48. Root mean square deviations 10% content obtained. Implications users so as derive accuracy at acid level provide user with guide prediction. available anonymous ftp ibcp.fr.