Protein Structure and Function: Application of Bioinformatics Methods - John Rigden 2014

Prediction of Protein Function Based on Theoretical Models
Protein Models as a Publicly Available Resource

As mentioned earlier in Chapter 3, one of the goals of structural Genomics and, in particular, the Cell/13.html">Protein Structure Initiative (PSI), is to experimentally determine protein structures that provide the widest possible coverage of protein fold space. Over the past 7 years, research centers participating in this project have solved approximately 3,000 structures, accounting for about 40% of all new structures with previously unknown folds deposited in the PDB database (Service 2008a). At the same time, substantial efforts in the field of Protein Structure Prediction have focused on improving algorithms and software to bring theoretical models as close as possible to experimentally determined structures. Successful rounds of the CASP (Critical Assessment of techniques for protein Structure Prediction) experiment, which serves as a benchmark for progress in this field (Kryshtafovych et al. 2005), have demonstrated a continuous increase in prediction accuracy. While poorly constructed models would offer little practical utility, in reality, for an average of 80% of the target Proteins featured in CASP, the generated models are sufficiently close and contain more useful information than the template itself (Kryshtafovych et al. 2007). (The added value of models relative to the sequence and template is discussed in more detail below.) The growing number of newly described fold types, driven by structural genomics and advances in computational accuracy, enables the generation of an increasing number of protein models. According to recent estimates, new structures solved within the PSI project allow for the generation of about 40,000 Homology models that otherwise could not have been obtained (Service 2008b). However, to fully leverage this vast array of models, they—along with information regarding their reliability—must be freely accessible to biologists.