Homologous proteins prediction

CA Schiffer, JW Caldwell, PA Kollman, RM Stroud. Prediction of homologous protein structures based on conformational searches and energetics. Pi otems 8 30-43, 1990. 

Prediction methods for secondary structure benefit from multiple alignment of homologous proteins... 

Several programs are now available that use multiple alignment of homologous proteins for prediction of secondary structure. One such program, called PHD, which was developed by Chris Sander and coworkers, EMBL, Heidelberg, has reached a mean accuracy of prediction of 72% for new structures. 

Homologous proteins have similar three-dimensional structures. They contain a core region, a scaffold of secondary structure elements, where the folds of the polypeptide chains are very similar. Loop regions that connect the building blocks of the scaffolds can vary considerably both in length and in structure. From a database of known immunoglobulin structures it has, nevertheless, been possible to predict successfully the conformation of hyper-variable loop regions of antibodies of known amino acid sequence. 

Methods for the prediction of the secondary structure of a set of homologous proteins can reach an accuracy of about 75%, most of the errors occur at the ends of a helices or p strands. The central regions of these secondary structure elements are often correctly predicted but the methods do not always correctly distinguish between a helices and p strands. 

From the human genome project it is known, that roughly 30,000 proteins exist in humans. Currently only the 3D-structures of few thousand human pr oteins or protein domains are known. Structures of membrane-bound proteins are several magnitudes rarer. Beside efforts to solve further structures like structural genomics, there is a challenge for computational approaches to predict structures and function for homologous proteins. 

From active site labelling experiments and from sequence comparison of NRs with homologous proteins, functionally important residues can be predicted and localised on the NR sequences. 

Tramontano A, Morea V. Assessment of homology-based predictions in CASP5. Proteins 2004 55 782. 

Methods based on prediction algorithms that use known structures of homologous proteins to assign secondary structures (Gamier et al., 1978 Zvelebil et al., 1987). 

Submit/Run Prediction button. The predicted results for the secondary structure (PHDsec), solvent accessibility (PHDacc), and helical transmembrane regions (PHDhtm) are returned by e-mail (Figure 12.9). The prediction reports the multiple alignment (with statistical summary) of homologous proteins, and it reports the query sequence (AA) with... 

Although the overall accuracy of secondary structure prediction is significantly improved with the PHDsec design, not all proteins can be equally well predicted. Worst predicted proteins are those with unusual features and those with bad sequence alignments. When information about homologous proteins is not available (i.e., no multiple sequence alignment), the predictive accuracy may be reduced by about 10%. 

The basic information of protein tertiary structural class can help improve the accuracy of secondary structure prediction (Kneller et al., 1990). Chandonia and Karplus (1995) showed that information obtained from a secondary structure prediction algorithm can be used to improve the accuracy for structural class prediction. The input layer had 26 units coded for the amino acid composition of the protein (20 units), the sequence length (1 unit), and characteristics of the protein (5 units) predicted by a separate secondary structure neural network. The secondary structure characteristics include the predicted percent helix and sheet, the percentage of strong helix and sheet predictions, and the predicted number of alterations between helix and sheet. The output layer had four units, one for each of the tertiary super classes (all-a, all-p, a/p, and other). The inclusion of the single-sequence secondary structure predictions improved the class prediction for non-homologous proteins significantly by more than 11%, from a predictive accuracy of 62.3% to 73.9%. 

---