Secondary Structure Prediction - Encyclopedia of Bioinformatics

The best modern methods of secondary structure prediction in proteins reach about 80% accuracy; this high accuracy allows the use of the predictions in fold recognition and ab initio protein structure prediction, classification of structural motifs, and refinement of sequence alignments. The accuracy of current protein secondary structure prediction methods is assessed in weekly benchmarks such as LiveBench and EVA.

The problems of predicting RNA secondary structure are broadly related but dependent mainly on base pairing and base stacking interactions; many RNA molecules have several possible three-dimensional structures, so predicting these structures remains out of reach unless obvious sequence and functional similarity to a known class of RNA molecules, such as transfer RNA or microRNA, is observed. Many RNA secondary structure prediction methods rely on variations of dynamic programming and therefore are unable to efficiently identify pseudoknots.

Protein structure

Early methods of secondary structure prediction, introduced in the 1960s and early 1970s,^[2] focused on identifying likely alpha helices and were based mainly on helix-coil transition models.^[3] Significantly more accurate predictions that included beta sheets were introduced in the 1970s and relied on statistical assessments based on probability parameters derived from known solved structures. These methods, applied to a single sequence, are typically at most about 60-65% accurate, and often underpredict beta sheets.^[1] The evolutionary conservation of secondary structures can be exploited by simultaneously assessing many homologous sequences in a multiple sequence alignment, by calculating the net secondary structure propensity of an aligned column of amino acids. In concert with larger databases of known protein structures and modern machine learning methods such as neural nets and support vector machines, these methods can achieve up 80% overall accuracy in globular proteins.^[4] The theoretical upper limit of accuracy is around 90%,^[4] partly due to idiosyncrasies in DSSP assignment near the ends of secondary structures, where local conformations vary under native conditions but may be forced to assume a single conformation in crystals due to packing constraints. Limitations are also imposed by secondary structure prediction's inability to account for tertiary structure; for example, a sequence predicted as a likely helix may still be able to adopt a beta-strand conformation if it is located within a beta-sheet region of the protein and its side chains pack well with their neighbors. Dramatic conformational changes related to the protein's function or environment can also alter local secondary structure.

Chou-Fasman method

The Chou-Fasman method was among the first secondary structure prediction algorithms developed and relies predominantly on probability parameters determined from relative frequencies of each amino acid's appearance in each type of secondary structure.^[5] The original Chou-Fasman parameters, determined from the small sample of structures solved in the mid-1970s, produce poor results compared to modern methods, though the parameterization has been updated since it was first published. The Chou-Fasman method is roughtly 50-60% accurate in predicting secondary structures.^[1]

GOR method

The GOR method, named for the three scientists who developed it - Garnier, Osguthorpe, and Robson - is an information theory-based method developed not long after Chou-Fasman that uses more powerful probabilistic techniques of Bayesian inference.^[6] The GOR method takes into account not only the probability of each amino acid having a particular secondary structure, but also the conditional probability of the amino acid assuming each structure given that its neighbors assume the same structure. This method is both more sensitive and more accurate because amino acid structural propensities are only strong for a small number of amino acids such as proline and glycine. The original GOR method is roughly 65% accurate and is dramatically more successful in predicting alpha helices than beta sheets, which it frequently mispredicts as loops or disorganized regions.^[1]

Machine learning

Neural network methods use training sets of solved structures to identify common sequence motifs associated with particular arrangements of secondary structures. These methods are over 70% accurate in their predictions, although beta strands are still often underpredicted due to the lack of three-dimensional structural information that would allow assessment of hydrogen bonding patterns that can promote formation of the extended conformation required for the presence of a complete beta sheet.^[1]

Support vector machines have proven particularly useful for predicting the locations of turns, which are difficult to identify with statistical methods.^[7] The requirement of relatively small training sets has also been cited as an advantage to avoid overfitting to existing structural data.^[8]

Extensions of machine learning techniques attempt to predict more fine-grained local properties of proteins, such as backbone dihedral angles in unassigned regions. Both SVMs^[9] and neural networks^[10] have been applied to this problem.

RNA structure

Dynamic programming algorithms are commonly used to detect base pairing patterns that are "well-nested", that is, form hydrogen bonds only to bases that do not overlap one another in sequence position. Secondary structures that fall into this category include double helices, stem-loops, and variants of the "cloverleaf" pattern found in transfer RNA molecules. These methods rely on precalculated parameters estimating the free energy associated with particular types of base-pairing interactions, including Watson-Crick and Hoogsteen base pairs. Depending on the complexity of the method, single base pairs may be considered, or short two- or three-base segments to incorporate the effects of base stacking. This method cannot identify pseudoknots, which are not well nested, without substantial algorithmic modifications that are extremely computationally expensive.^[11]

Sequence covariation methods rely on the existence of a data set composed of multiple homologous RNA sequences with related but dissimilar sequences. These methods analyze the covariation of individual base sites in evolution; maintenance at two widely separated sites of a pair of base-pairing nucleotides indicates the presence of a structurally required hydrogen bond between those positions. The general problem of pseudoknot prediction has been shown to be NP-complete.^[12]

References