1 - 5 of 148
- PublicationIn Silico Protein Motif Discovery and Structural Analysis(Springer, 2011-06-30)A wealth of in silico tools is available for protein motif discovery and structural analysis. The aim of this chapter is to collect some of the most common and useful tools and to guide the biologist in their use. A detailed explanation is provided for the use of Distill, a suite of web servers for the prediction of protein structural features and the prediction of full-atom 3D models from a protein sequence. Besides this, we also provide pointers to many other tools available for motif discovery and secondary and tertiary structure prediction from a primary amino acid sequence. The prediction of protein intrinsic disorder and the prediction of functional sites and SLiMs are also briefly discussed. Given that user queries vary greatly in size, scope and character, the trade-offs in speed, accuracy and scale need to be considered when choosing which methods to adopt.
Scopus© Citations 1 91
- PublicationIn silico approaches to predict the potential of milk protein-derived peptides as dipeptidyl peptidase IV (DPP-IV) inhibitors(Elsevier, 2014-07)Molecular docking of a library of all 8000 possible tripeptides to the active site of DPP-IV was used to determine their binding potential. A number of tripeptides were selected for experimental testing, however, there was no direct correlation between the Vina score and their in vitro DPP-IV inhibitory properties. While Trp-Trp-Trp, the peptide with the best docking score, was a moderate DPP-IV inhibitor (IC50 216 μM), Lineweaver and Burk analysis revealed its action to be non-competitive. This suggested that it may not bind to the active site of DPP-IV as assumed in the docking prediction. Furthermore, there was no significant link between DPP-IV inhibition and the physicochemical properties of the peptides (molecular mass, hydrophobicity, hydrophobic moment (μH), isoelectric point (pI) and charge). LIGPLOTs indicated that competitive inhibitory peptides were predicted to have both hydrophobic and hydrogen bond interactions with the active site of DPP-IV. DPP-IV inhibitory peptides generally had a hydrophobic or aromatic amino acid at the N-terminus, preferentially a Trp for non-competitive inhibitors and a broader range of residues for competitive inhibitors (Ile, Leu, Val, Phe, Trp or Tyr). Two of the potent DPP-IV inhibitors, Ile-Pro-Ile and Trp-Pro (IC 50 values of 3.5 and 44.2 μM, respectively), were predicted to be gastrointestinally/intestinally stable. This work highlights the needs to test the assumptions (i.e. competitive binding) of any integrated strategy of computational and experimental screening, in optimizing screening. Future strategies targeting allosteric mechanisms may need to rely more on structure-activity relationship modeling, rather than on docking, in computationally selecting peptides for screening.
Scopus© Citations 116 142
- PublicationVariational Bayesian inference for the Latent Position Cluster Model for network data(Elsevier, 2013-01)A number of recent approaches to modeling social networks have focussed on embedding the nodes in a latent “social space”. Nodes that are in close proximity are more likely to form links than those who are distant. This naturally accounts for reciprocal and transitive relationships which are commonly found in many network datasets. The Latent Position Cluster Model is one such model that also explicitly incorporates clustering by modeling the locations using a finite Gaussian mixture model. Observed covariates and sociality random effects may also be modeled. However, inference for the model via MCMC is cumbersome and thus scaling to large networks is a challenge. Variational Bayesian methods offer an alternative inference methodology for this problem. Sampling based MCMC is replaced by an optimization that requires many orders of magnitude fewer iterations to converge. A Variational Bayesian algorithm for the Latent Position Cluster Model is therefore developed and demonstrated.
Scopus© Citations 50 210
- PublicationPairwise Interaction Field Neural Networks For Drug Discovery(University College Dublin. School of Computer Science and Informatics, 2013-06)Automatically mapping small, drug-like molecules into their biological activity is an open problem in chemioinformatics. Numerous approaches to solve the problem have been attempted, which typically rely on different machine learning tools and, critically, depend on the how a molecule is represented (be it as a one-dimensional string, a two-dimensional graph, its three-dimensional structure, or a feature vector of some kind). In fact arguably the most critical bottleneck in the process is how to encode the molecule in a way that is both informative and can be dealt with by the machine learning algorithms downstream. Recently we have introduced an algorithm which entirely does away with this complex, error-prone and time-consuming encoding step by automatically finding an optimal code for a molecule represented as a twodimensional graph. In this report we introduce a model which we have recently developed (Neural Network Pairwise Interaction Fields) to extend this same approach to molecules represented as their three-dimensional structures. We benchmark the algorithm on a number of public data sets. While our tests confirm that three-dimensional representations are generally less informative than two-dimensional ones (possibly because the former are generally the result of a prediction process, and as such contain noise), the algorithm we introduce compares well with the state of the art in 3D-based prediction, in spite of not requiring any prior knowledge about the domain, or prior encoding of the molecule.
- PublicationPotential utility of docking to identify protein-peptide binding regions(University College Dublin. School of Computer Science and Informatics, 2013-05)Disordered regions of proteins often bind to structured domains, mediating interactions within and between proteins. However, it is difficult to identify a priori the short regions involved in binding. We set out to determine if docking peptides to peptide binding domains would assist in these predictions. First, we investigated the docking of known short peptides to their native and non-native peptide binding domains. We then investigated the docking of overlapping peptides adjacent to the native peptide. We found only weak discrimination of docking scores between native peptide and adjacent peptides in this context with similar results for both ordered and disordered regions. Finally, we trained a bidirectional recurrent neural network using as input the peptide sequence, predicted secondary structure, Vina docking score and Pepsite score.We conclude that docking has only modest power to define the location of a peptide within a larger protein region known to contain it. However, this information can be used in training machine learning methods which may allow for the identification of peptide binding regions within a protein sequence.