Dataset Information

A novel strategy for classifying the output from an in silico vaccine discovery pipeline for eukaryotic pathogens using machine learning algorithms.

ABSTRACT: BACKGROUND:An in silico vaccine discovery pipeline for eukaryotic pathogens typically consists of several computational tools to predict protein characteristics. The aim of the in silico approach to discovering subunit vaccines is to use predicted characteristics to identify proteins which are worthy of laboratory investigation. A major challenge is that these predictions are inherent with hidden inaccuracies and contradictions. This study focuses on how to reduce the number of false candidates using machine learning algorithms rather than relying on expensive laboratory validation. Proteins from Toxoplasma gondii, Plasmodium sp., and Caenorhabditis elegans were used as training and test datasets. RESULTS:The results show that machine learning algorithms can effectively distinguish expected true from expected false vaccine candidates (with an average sensitivity and specificity of 0.97 and 0.98 respectively), for proteins observed to induce immune responses experimentally. CONCLUSIONS:Vaccine candidates from an in silico approach can only be truly validated in a laboratory. Given any in silico output and appropriate training data, the number of false candidates allocated for validation can be dramatically reduced using a pool of machine learning algorithms. This will ultimately save time and money in the laboratory.

SUBMITTER: Goodswen SJ

PROVIDER: S-EPMC3826511 | biostudies-literature | 2013 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A novel strategy for classifying the output from an in silico vaccine discovery pipeline for eukaryotic pathogens using machine learning algorithms.

Goodswen Stephen J SJ Kennedy Paul J PJ Ellis John T JT

BMC bioinformatics 20131102

<h4>Background</h4>An in silico vaccine discovery pipeline for eukaryotic pathogens typically consists of several computational tools to predict protein characteristics. The aim of the in silico approach to discovering subunit vaccines is to use predicted characteristics to identify proteins which are worthy of laboratory investigation. A major challenge is that these predictions are inherent with hidden inaccuracies and contradictions. This study focuses on how to reduce the number of false can ...[more]

PMID: 24180526

Similar Datasets

Project description:Given thousands of proteins constituting a eukaryotic pathogen, the principal objective for a high-throughput in silico vaccine discovery pipeline is to select those proteins worthy of laboratory validation. Accurate prediction of T-cell epitopes on protein antigens is one crucial piece of evidence that would aid in this selection. Prediction of peptides recognised by T-cell receptors have to date proved to be of insufficient accuracy. The in silico approach is consequently reliant on an indirect method, which involves the prediction of peptides binding to major histocompatibility complex (MHC) molecules. There is no guarantee nevertheless that predicted peptide-MHC complexes will be presented by antigen-presenting cells and/or recognised by cognate T-cell receptors. The aim of this study was to determine if predicted peptide-MHC binding scores could provide contributing evidence to establish a protein's potential as a vaccine. Using T-Cell MHC class I binding prediction tools provided by the Immune Epitope Database and Analysis Resource, peptide binding affinity to 76 common MHC I alleles were predicted for 160 Toxoplasma gondii proteins: 75 taken from published studies represented proteins known or expected to induce T-cell immune responses and 85 considered less likely vaccine candidates. The results show there is no universal set of rules that can be applied directly to binding scores to distinguish a vaccine from a non-vaccine candidate. We present, however, two proposed strategies exploiting binding scores that provide supporting evidence that a protein is likely to induce a T-cell immune response-one using random forest (a machine learning algorithm) with a 72% sensitivity and 82.4% specificity and the other, using amino acid conservation scores with a 74.6% sensitivity and 70.5% specificity when applied to the 160 benchmark proteins. More importantly, the binding score strategies are valuable evidence contributors to the overall in silico vaccine discovery pool of evidence.

Project description:Background and objectiveSmoking is the largest preventable cause of death and diseases in the developed world, and advances in modern electronics and machine learning can help us deliver real-time intervention to smokers in novel ways. In this paper, we examine different machine learning approaches to use situational features associated with having or not having urges to smoke during a quit attempt in order to accurately classify high-urge states.MethodsTo test our machine learning approaches, specifically, Bayes, discriminant analysis and decision tree learning methods, we used a dataset collected from over 300 participants who had initiated a quit attempt. The three classification approaches are evaluated observing sensitivity, specificity, accuracy and precision.ResultsThe outcome of the analysis showed that algorithms based on feature selection make it possible to obtain high classification rates with only a few features selected from the entire dataset. The classification tree method outperformed the naive Bayes and discriminant analysis methods, with an accuracy of the classifications up to 86%. These numbers suggest that machine learning may be a suitable approach to deal with smoking cessation matters, and to predict smoking urges, outlining a potential use for mobile health applications.ConclusionsIn conclusion, machine learning classifiers can help identify smoking situations, and the search for the best features and classifier parameters significantly improves the algorithms' performance. In addition, this study also supports the usefulness of new technologies in improving the effect of smoking cessation interventions, the management of time and patients by therapists, and thus the optimization of available health care resources. Future studies should focus on providing more adaptive and personalized support to people who really need it, in a minimum amount of time by developing novel expert systems capable of delivering real-time interventions.

Project description:Machine learning methods are attracting considerable attention from the pharmaceutical industry for use in drug discovery and applications beyond. In recent studies, we and others have applied multiple machine learning algorithms and modeling metrics and, in some cases, compared molecular descriptors to build models for individual targets or properties on a relatively small scale. Several research groups have used large numbers of datasets from public databases such as ChEMBL in order to evaluate machine learning methods of interest to them. The largest of these types of studies used on the order of 1400 datasets. We have now extracted well over 5000 datasets from CHEMBL for use with the ECFP6 fingerprint and in comparison of our proprietary software Assay Central with random forest, k-nearest neighbors, support vector classification, naïve Bayesian, AdaBoosted decision trees, and deep neural networks (three layers). Model performance was assessed using an array of fivefold cross-validation metrics including area-under-the-curve, F1 score, Cohen's kappa, and Matthews correlation coefficient. Based on ranked normalized scores for the metrics or datasets, all methods appeared comparable, while the distance from the top indicated that Assay Central and support vector classification were comparable. Unlike prior studies which have placed considerable emphasis on deep neural networks (deep learning), no advantage was seen in this case. If anything, Assay Central may have been at a slight advantage as the activity cutoff for each of the over 5000 datasets representing over 570,000 unique compounds was based on Assay Central performance, although support vector classification seems to be a strong competitor. We also applied Assay Central to perform prospective predictions for the toxicity targets PXR and hERG to further validate these models. This work appears to be the largest scale comparison of these machine learning algorithms to date. Future studies will likely evaluate additional databases, descriptors, and machine learning algorithms and further refine the methods for evaluating and comparing such models.

Project description:MOTIVATION:Artificial intelligence, trained via machine learning (e.g. neural nets, random forests) or computational statistical algorithms (e.g. support vector machines, ridge regression), holds much promise for the improvement of small-molecule drug discovery. However, small-molecule structure-activity data are high dimensional with low signal-to-noise ratios and proper validation of predictive methods is difficult. It is poorly understood which, if any, of the currently available machine learning algorithms will best predict new candidate drugs. RESULTS:The quantile-activity bootstrap is proposed as a new model validation framework using quantile splits on the activity distribution function to construct training and testing sets. In addition, we propose two novel rank-based loss functions which penalize only the out-of-sample predicted ranks of high-activity molecules. The combination of these methods was used to assess the performance of neural nets, random forests, support vector machines (regression) and ridge regression applied to 25 diverse high-quality structure-activity datasets publicly available on ChEMBL. Model validation based on random partitioning of available data favours models that overfit and 'memorize' the training set, namely random forests and deep neural nets. Partitioning based on quantiles of the activity distribution correctly penalizes extrapolation of models onto structurally different molecules outside of the training data. Simpler, traditional statistical methods such as ridge regression can outperform state-of-the-art machine learning methods in this setting. In addition, our new rank-based loss functions give considerably different results from mean squared error highlighting the necessity to define model optimality with respect to the decision task at hand. AVAILABILITY AND IMPLEMENTATION:All software and data are available as Jupyter notebooks found at https://github.com/owatson/QuantileBootstrap. SUPPLEMENTARY INFORMATION:Supplementary data are available at Bioinformatics online.

Dataset Information

A novel strategy for classifying the output from an in silico vaccine discovery pipeline for eukaryotic pathogens using machine learning algorithms.

Publications

A novel strategy for classifying the output from an in silico vaccine discovery pipeline for eukaryotic pathogens using machine learning algorithms.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure