A Hidden Markov Model applied to the protein 3D structure analysis (original) (raw)
Related papers
A hidden Markov model applied to the analysis of protein 3D-structures
Understanding and predicting protein structures depends on the complexity and the accuracy of the models used to represent them. We have setup a Hidden Markov Model to optimally compress three dimensional (3D) conformation of protein into a structural alphabet, i.e. a library of exhaustive and representative states (describing short fragments) learnt simultaneously with connection logic. The discretization of protein backbone local conformation as a series of states results in a simplification of protein 3D coordinates into a unique unidimensional (1D) representation. We present some evidence that such approach can constitute a very relevant way to the analysis of protein architecture in particular for protein structure comparison or prediction.
A Hidden Markov Model Derived Structural Alphabet for Proteins
Journal of Molecular Biology, 2004
Understanding and predicting protein structures depends on the complexity and the accuracy of the models used to represent them. We have set up a hidden Markov model that discretizes protein backbone conformation as series of overlapping fragments (states) of four residues length. This approach learns simultaneously the geometry of the states and their connections. We obtain, using a statistical criterion, an optimal systematic decomposition of the conformational variability of the protein peptidic chain in 27 states with strong connection logic. This result is stable over different protein sets. Our model fits well the previous knowledge related to protein architecture organisation and seems able to grab some subtle details of protein organisation, such as helix sub-level organisation schemes. Taking into account the dependence between the states results in a description of local protein structure of low complexity. On an average, the model makes use of only 8.3 states among 27 to describe each position of a protein structure. Although we use short fragments, the learning process on entire protein conformations captures the logic of the assembly on a larger scale. Using such a model, the structure of proteins can be reconstructed with an average accuracy close to 1.1 Å root-mean-square deviation and for a complexity of only 3. Finally, we also observe that sequence specificity increases with the number of states of the structural alphabet. Such models can constitute a very relevant approach to the analysis of protein architecture in particular for protein structure prediction.
Biochimica et Biophysica Acta (BBA) - General Subjects, 2005
Understanding and predicting protein structures depend on the complexity and the accuracy of the models used to represent them. We have recently set up a Hidden Markov Model to optimally compress protein three-dimensional conformations into a one-dimensional series of letters of a structural alphabet. Such a model learns simultaneously the shape of representative structural letters describing the local conformation and the logic of their connections, i.e. the transition matrix between the letters. Here, we move one step further and report some evidence that such a model of protein local architecture also captures some accurate amino acid features. All the letters have specific and distinct amino acid distributions. Moreover, we show that words of amino acids can have significant propensities for some letters. Perspectives point towards the prediction of the series of letters describing the structure of a protein from its amino acid sequence. D
Hidden Markov model approach for identifying the modular framework of the protein backbone
Protein Engineering Design and Selection, 1999
The hidden Markov model (HMM) was used to identify recurrent short 3D structural building blocks (SBBs) describing protein backbones, independently of any a priori knowledge. Polypeptide chains are decomposed into a series of short segments defined by their inter-α-carbon distances. Basically, the model takes into account the sequentiality of the observed segments and assumes that each one corresponds to one of several possible SBBs. Fitting the model to a database of non-redundant proteins allowed us to decode proteins in terms of 12 distinct SBBs with different roles in protein structure. Some SBBs correspond to classical regular secondary structures. Others correspond to a significant subdivision of their bounding regions previously considered to be a single pattern. The major contribution of the HMM is that this model implicitly takes into account the sequential connections between SBBs and thus describes the most probable pathways by which the blocks are connected to form the framework of the protein structures. Validation of the SBBs code was performed by extracting SBB series repeated in recoding proteins and examining their structural similarities. Preliminary results on the sequence specificity of SBBs suggest promising perspectives for the prediction of SBBs or series of SBBs from the protein sequences.
Classification of Protein 3D Folds by Hidden Markov Learning on Sequences of Structural Alphabets
Proceedings of the 3rd Asia-Pacific Bioinformatics Conference, 2005
Fragment-based analysis of protein three-dimensional (3D) structures has received increased attention in recent years. Here, we used a set of pentamer local structure alphabets (LSAs) recently derived in our laboratory to represent protein structures, i.e. we transformed the 3D structures into one-dimensional (1D) sequences of LSAs. We then applied Hidden Markov Model training to these LSA sequences to assess their ability to capture features characteristic of 43 populated protein folds. In the size range of LSAs examined (5 to 41 alphabets), the performance was optimal using 20 alphabets, giving an accuracy of fold classification of 82% in a 5-fold cross-validation on training-set structures sharing < 40% pairwise sequence identity at the amino acid level. For test-set structures, the accuracy was as high as for the training set, but fell to 65% for those sharing no more than 25% amino acid sequence identity with the training-set structures. These results suggest that sufficient 3D information can be retained during the drastic 3D->1D transformation for use as a framework for developing efficient and useful structural bioinformatics tools.
Predicting protein structure using SAM, UCSC’s hidden Markov model tools
2002
Abstract The protein-folding problem, in its purest form, is too difficult for us to solve in the next several years, but we need structure predictions now. One solution is to try to recognize the similarity between a target protein and one of the thousands of proteins whose structure has been determined experimentally. For very similar proteins, the relationships are easy to find and good models can be built by copying the backbone (and even some sidechains) from the homologous protein of known structure.
Predicting protein structure using hidden Markov models
Proteins: Structure, Function, and Genetics, 1997
We discuss how methods based on hidden Markov models performed in the fold recognition section of the CASP2 experiment. Hidden Markov models were built for a set of about a thousand structures from the PDB database, and each CASP2 target sequence was scored against this library of hidden Markov models. In addition, a hidden Markov model was built for each of the target sequences, and all of the sequences in PDB were scored against that target model. Having high scores from both methods was found to be highly indicative of the target and a structure being homologous.
Proceedings Seventh International Symposium on String Processing and Information Retrieval. SPIRE 2000, 2000
The transformation of protein 1D-sequence to protein 3D-structure is one of the main difficulties of the structural biology. A structural alphabet had been previously defined from dihedral angles describing the protein backbone as structural information by using an unsupervised classifier. The 16 Protein Blocks (PBs), basis element of the structural alphabet, allows a correct 3D structure approximation . Local prediction had been estimated by a Bayesian approach and shown that sequence information induces strongly the local fold, but stays coarse (prediction rate of 40.7% with one PB, 75.8% with the four most probable PBs).