Neural network analysis of protein tertiary structure

Wilcox, George L.; Poliac, Marius; Liebman, Michael

doi:10.1016/0898-5529(90)90052-a

Cited by 28 publications

(9 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…BigNet performed well in acquisition of this new, homologous training set of larger proteins using roughly the same number of hidden units required for our previous heterologous training set of smaller proteins (Wilcox et al, 1991). Associations between protein sequence and structure in the training set were learned to better than 99% precision within 100 iterations.…”

Section: Resultsmentioning

confidence: 92%

“…We have used one "alphabet" to describe the amino acids: hydrophobicity values adapted from Liebman et al (1986) that ranged from -3.4 for the hydrophobic amino acid tyrosine to +3.3 for the hydrophilic amino acid lysine and were normalized into the range of f l (described in Wilcox et al, 1991). Since several amino acids have almost identical hydrophobicities, this alphabet is degenerate.…”

Section: Methodsmentioning

confidence: 99%

“…Limitations in computer memory and time limited their attention to these near-diagonal elements of distance matrices relating mostly to secondary structure, the local elements of protein folding. Our approach, by contrast, involves We have demonstrated in previous studies the utility of large neural network simulations for The three dimensional structure of a protein is intimately related to its biological function, and CH 3065-0,91/0000-1323 $1.00 OIEEE applying a very large neural network to learn entire distance matrices which encode tertiary as well as secondary structure elements (Wilcox et al, 1991).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Prediction of protein folding using the shift-learning method with a large scale neural network

Poliac

Wilcox

Xin

et al. 1991

[Proceedings] 1991 IEEE International Joint Conference on Neural Networks

View full text Add to dashboard Cite

encoding the association between protein sequence and three dimensional structure for a small heterologous training set of small proteins. In the present study, we report the application of this approach to a selected homologous training set of 8 proteins using the Cray 2 supercompter at the Minnesota Supercomputer Center, Minneapolis USA. The large memory of this machine allowed us to configure a network with more than .3 million connections and 30.000 neural units; a network of this size was necessary to accommodate a new training/testing set with 8 proteins of up to 140 amino acid residues. This training set was constructed to investigate the performance of the neural network approach in prediction of structures within the protease class of proteins; proteases are enzymes which cleave the peptide bonds which join individual amino acid residues of other proteins. The network learned the sequence-structure association for 4 of the proteins within 100 iterations selected in a random order and shifted by a random offset to the left or to the right. When presented with novel sequences from related proteins, the network was able to predict three dimensional structures of the four proteins in the testing set. The results of this study suggest that a neural network trained to recognize the entire sequence of a protein using the shift-learn method can retain some of the rules of protein folding in a form which allows prediction of three dimensional structures. Our findings indicate that large scalar or vector supercomputer architectures are ideal for implementation of useful backpropagation neural networks.

show abstract

Section: Resultsmentioning

confidence: 92%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Prediction of protein folding using the shift-learning method with a large scale neural network

Poliac

Wilcox

Xin

et al. 1991

[Proceedings] 1991 IEEE International Joint Conference on Neural Networks

View full text Add to dashboard Cite

show abstract

“…Recent neural-network applications which use "windows" of amino acids, i.e. , local sequence information, for predicting distance matrices [19,20] of realistic polypeptides and protein structures would be deficient in two respects. First, prediction accuracy is lost to some significant extent.…”

Section: Results and Discussidnmentioning

confidence: 99%

Optimal neural networks for protein-structure prediction

Head‐Gordon

Stillinger

1993

Phys. Rev. E

View full text Add to dashboard Cite

The successful application of neural-network algorithms for prediction of protein structure is stymied by three problem areas: the sparsity of the database of known protein structures, poorly devised network architectures which make the input-output mapping opaque, and a global optimization problem in the multiple-minima space of the network variables. We present a simplified polypeptide model residing in two dimensions with only two amino-acid types, A and B, which allows the determination of the global energy structure for all possible sequences of pentamer, hexamer, and heptamer lengths. This model simplicity allows us to compile a complete structural database and to devise neural networks that reproduce the tertiary structure of all sequences with absolute accuracy and with the smallest number of network variables. These optimal networks reveal that the three problem areas are convoluted, but that thoughtful network designs can actually deconvolute these detrimental traits to provide network algorithms that genuinely impact on the ability of the network to generalize or learn the desired rnappings. Furthermore, the two-dimensional polypeptide model shows sufficient chemical complexity so that transfer of neural-network technology to more realistic three-dimensional proteins is evident.

show abstract

“…Each amino acid consists of a rigid plane formed by single nitrogen, carbon, alpha-carbon (Ca), oxygen, and hydrogen atoms, and a distinguishing side chain. The individual amino acids are distinguished from each other by a number of physical chemical properties that give rise to the three dimensional structure [4].…”

Section: Introductionmentioning

confidence: 99%

Computational Methods for Protein Structure Prediction and Its Application in Drug Design

Toomula¹

2011

J Bacteriol Parasitol

View full text Add to dashboard Cite

Drug design and drug discovery are of critical importance in human health care. Computational approaches have become a major part of structure based drug design. Structure-based drug design utilizes the three dimensional structure of a protein target to design candidate drugs that are predicted to bind with high affinity and selectivity to the target. In this review computational methods for prediction of the protein structure are described and their use towards the drug design is discussed.

show abstract

Neural network analysis of protein tertiary structure

Cited by 28 publications

References 14 publications

Prediction of protein folding using the shift-learning method with a large scale neural network

Prediction of protein folding using the shift-learning method with a large scale neural network

Optimal neural networks for protein-structure prediction

Computational Methods for Protein Structure Prediction and Its Application in Drug Design

Contact Info

Product

Resources

About