3D structure prediction

The JME Editor is a Java program which allows one to draw, edit, and display molecules and reactions directly within a web page and may also be used as an application in a stand-alone mode. The editor was originally developed for use in an in-house web-based chemoinformatics system but because of many requests it was released to the public. The JME currently is probably the most popular molecule entry system written in Java. Internet sites that use the JME applet include several structure databases, property prediction services, various chemoinformatics tools (such as for generation of 3D structures or molecular orbital visualization), and interactive sites focused on chemistry education [209]. 

For each fold one searches for the best alignment of the target sequence that would be compatible with the fold the core should comprise hydrophobic residues and polar residues should be on the outside, predicted helical and strand regions should be aligned to corresponding secondary structure elements in the fold, and so on. In order to match a sequence alignment to a fold, Eisenberg developed a rapid method called the 3D profile method. The environment of each residue position in the known 3D structure is characterized on the basis of three properties (1) the area of the side chain that is buried by other protein atoms, (2) the fraction of side chain area that is covered by polar atoms, and (3) the secondary stmcture, which is classified in three states helix, sheet, and coil. The residue positions are rather arbitrarily divided into six classes by properties 1 and 2, which in combination with property 3 yields 18 environmental classes. This classification of environments enables a protein structure to be coded by a sequence in an 18-letter alphabet, in which each letter represents the environmental class of a residue position. 

This branch of bioinformatics is concerned with computational approaches to predict and analyse the spatial structure of proteins and nucleic acids. Whereas in many cases the primary sequence uniquely specifies the 3D structure, the specific rules are not well understood, and the protein folding problem remains largely unsolved. Some aspects of protein structure can already be predicted from amino acid content. Secondary structure can be deduced from the primary sequence with statistics or neural networks. When using a multiple sequence alignment, secondary structure can be predicted with an accuracy above 70%. 

Protein-protein interactions predicted on the sequence level can be studied in more detail on the structure level. Single Nucleotide Polymorphisms can be mapped on 3D structures of proteins in order to elucidate specific structural causes of disease. 

The aim of the second dimension depth is to consider protein 3D-stmctures to uncover structure-function relationships. Starting from the protein sequences, the steps in the depth dimension are structure prediction, homology modeling of protein structures, and the simulation of protein-protein interactions and ligand-complexes. 

Approaches of de novo predictions, which try to calculate how the structural elements are folded into the 3D-stmcture (tertiary structure) of complete proteins are nowadays far away from reliable large-scale applications. On the other, hand this topic is under strong development indicated by recent successful results at the contest for structural prediction methods CASP4. With the fast growing number of experimentally solved 3D-stmctures of protein and new promising approaches like threading tools combined with experimental structural constraints, one can expect more reliable de novo predictions for 3D-protein structures in the future. 

The strongest verification for a 3D-protein model comes from the experimental 3D-structure. This is the objective of the Critical Assessment of Techniques for Protein Structure Prediction, CASP ( http //predic tioncenter.org), where the structural models are made in advance of the experimental structure of a particular protein. 

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. 

The protein structure prediction problem refers to the combinatorial problem to calculate the 3D structure of a protein from its sequence alone. It is one of the biggest challenges in structural bioinformatics. 

Threading techniques try to match a target sequence on a library of known 3D structures by threading the target sequence over the known coordinates. In this manner, threading tries to predict the 3D structure starting from a given protein sequence. It is sometimes successful when comparisons based on sequences or sequence profiles alone fail due to a too low sequence similarity. 

Here, an attempt to classify different strategies to generate 3D molecular models is undertaken with the aim to specify the remit of methods which will be covered under the term automatic 3D structure generators . The focus will be on methods designed for small, dmg-like molecules. The prediction of the geometry of polymers, in parhcular of biopolymers, is a task of its own and not even attempted by the approaches discussed here. 

The importance of methods to predict log P from chemical structure was described in Chapter 14, which is focused on fragment- and atom-based approaches. In this chapter property-based approaches are reviewed, which comprise two main categories (i) methods that use three-dimensional (3D) structure representation and (ii) methods that are based on topological descriptors. 

---