Sparse feature selection methods identify unexpected global cellular response to strontium-containing materials.

Despite the increasing sophistication of biomaterials design and functional characterization studies, little is known regarding cells' global response to biomaterials. Here, we combined nontargeted holistic biological and physical science techniques to evaluate how simple strontium ion incorporation within the well-described biomaterial 45S5 bioactive glass (BG) influences the global response of human mesenchymal stem cells. Our objective analyses of whole gene-expression profiles, confirmed by standard molecular biology techniques, revealed that strontium-substituted BG up-regulated the isoprenoid pathway, suggesting an influence on both sterol metabolite synthesis and protein prenylation processes. This up-regulation was accompanied by increases in cellular and membrane cholesterol and lipid raft contents as determined by Raman spectroscopy mapping and total internal reflection fluorescence microscopy analyses and by an increase in cellular content of phosphorylated myosin II light chain. Our unexpected findings of this strong metabolic pathway regulation as a response to biomaterial composition highlight the benefits of discovery-driven nonreductionist approaches to gain a deeper understanding of global cell-material interactions and suggest alternative research routes for evaluating biomaterials to improve their design.

Relevance Vector Machines: Sparse Classification Methods for QSAR.

Sparse machine learning methods have provided substantial benefits to quantitative structure property modeling, as they make model interpretation simpler and generate models with improved predictivity. Sparsity is usually induced via Bayesian regularization using sparsity-inducing priors and by the use of expectation maximization algorithms with sparse priors. The focus to date has been on using sparse methods to model continuous data and to carry out sparse feature selection. We describe the relevance vector machine (RVM), a sparse version of the support vector machine (SVM) that is one of the most widely used classification machine learning methods in QSAR and QSPR. We illustrate the superior properties of the RVM by modeling eight data sets using SVM, RVM, and another sparse Bayesian machine learning method, the Bayesian regularized artificial neural network with Laplacian prior (BRANNLP). We show that RVM models are substantially sparser than the SVM models and have similar or superior performance to them.

Predicting the complex phase behavior of self-assembling drug delivery nanoparticles.

Amphiphilic lyotropic liquid crystalline self-assembled nanomaterials have important applications in the delivery of therapeutic and imaging agents. However, little is known about the effect of the incorporated drug on the structure of nanoparticles. Predicting these properties is widely considered intractable. We present computational models for three drug delivery carriers, loaded with 10 drugs at six concentrations and two temperatures. These models predicted phase behavior for 11 new drugs. Subsequent synchrotron small-angle X-ray scattering experiments validated the predictions.

Modeling biological activities of nanoparticles.

Products are increasingly incorporating nanomaterials, but we have a poor understanding of their adverse effects. To assess risk, regulatory authorities need more experimental testing of nanoparticles. Computational models play a complementary role in allowing rapid prediction of potential toxicities of new and modified nanomaterials. We generated quantitative, predictive models of cellular uptake and apoptosis induced by nanoparticles for several cell types. We illustrate the potential of computational methods to make a contribution to nanosafety.

Nonlinear predictive modeling of MHC class II-peptide binding using Bayesian neural networks.

Methods for predicting the binding affinity of peptides to the MHC have become more sophisticated in the past 5-10 years. It is possible to use computational quantitative structure-activity methods to build models of peptide affinity that are truly predictive. Two of the most useful methods for building models are Bayesian regularized neural networks for continuous or discrete (categorical) data and support vector machines (SVMs) for discrete data. We illustrate the application of Bayesian regularized neural networks to modeling MHC class II-binding affinity of peptides. Training data comprised sequences and binding data for nonamer (nine amino acid) peptides. Peptides were characterized by mathematical representations of several types. Independent test data comprised sequences and binding data for peptides of length < or = 25. We also internally validated the models by using 30% of the data in an internal test set. We obtained robust models, with near-identical statistics for multiple training runs. We determined how predictive our models were using statistical tests and area under the receiver operating characteristic (ROC) graphs (A(ROC)). Some mathematical representations of the peptides were more efficient than others and were able to generalize to unknown peptides outside of the training space. Bayesian neural networks are robust, efficient "universal approximators" that are well able to tackle the difficult problem of correctly predicting the MHC class II-binding activities of a majority of the test set peptides.

Predictive Bayesian neural network models of MHC class II peptide binding.

We used Bayesian regularized neural networks to model data on the MHC class II-binding affinity of peptides. Training data consisted of sequences and binding data for nonamer (nine amino acid) peptides. Independent test data consisted of sequences and binding data for peptides of length 0.8. We also used both amino acid indicator variables (bin20) and property-based descriptors to generate models for MHC class II-binding of peptides. The property-based descriptors were more parsimonious than the indicator variable descriptors, making them applicable to larger peptides, and their design makes them able to generalize to unknown peptides outside of the training space. None of the external test data sets contained any of the nonamer sequences in the training sets. Consequently, the models attempted to predict the activity of truly unknown peptides not encountered in the training sets. Our models were well able to tackle the difficult problem of correctly predicting the MHC class II-binding activities of a majority of the test set peptides. Exceptions to the assumption that nonamer motif activities were invariant to the peptide in which they were embedded, together with the limited coverage of the test data, and the fuzziness of the classification procedure, are likely explanations for some misclassifications.

Sparse feature selection identifies H2A.Z as a novel, pattern-specific biomarker for asymmetrically self-renewing distributed stem cells.

There is a long-standing unmet clinical need for biomarkers with high specificity for distributed stem cells (DSCs) in tissues, or for use in diagnostic and therapeutic cell preparations (e.g., bone marrow). Although DSCs are essential for tissue maintenance and repair, accurate determination of their numbers for medical applications has been problematic. Previous searches for biomarkers expressed specifically in DSCs were hampered by difficulty obtaining pure DSCs and by the challenges in mining complex molecular expression data. To identify such useful and specific DSC biomarkers, we combined a novel sparse feature selection method with combinatorial molecular expression data focused on asymmetric self-renewal, a conspicuous property of DSCs. The analysis identified reduced expression of the histone H2A variant H2A.Z as a superior molecular discriminator for DSC asymmetric self-renewal. Subsequent molecular expression studies showed H2A.Z to be a novel "pattern-specific biomarker" for asymmetrically self-renewing cells, with sufficient specificity to count asymmetrically self-renewing DSCs in vitro and potentially in situ.

Broad-based quantitative structure-activity relationship modeling of potency and selectivity of farnesyltransferase inhibitors using a Bayesian regularized neural network.

Inhibitors of the enzyme farnesyltransferase show potential as novel anticancer agents. There are many known inhibitors, but efforts to build predictive SAR models have been hampered by the structural diversity and flexibility of inhibitors. We have undertaken for the first time a QSAR study of the potency and selectivity of a large, diverse data set of farnesyltransferase inhibitors. We used novel molecular descriptors based on binned atomic properties and invariants of molecular matrices and a robust, nonlinear QSAR mapping paradigm, the Bayesian regularized neural network. We have built robust QSAR models of farnesyltransferase inhibition, geranylgeranyltransferase inhibition, and in vivo data. We have derived a novel selectivity index that allows us to model potency and selectivity simultaneously and have built robust QSAR models using this index that have the potential to discover new potent and selective inhibitors.

Modelling blood-brain barrier partitioning using Bayesian neural nets.

We have employed three families of molecular molecular descriptors, together with Bayesian regularized neural nets, to model the partitioning of a diverse range of drugs and other small molecules across the blood-brain barrier (BBB). The relative efficacy of each descriptors class is compared, and the advantages of flexible, parsimonious, model free mapping methods, like Bayesian neural nets, illustrated. The relative importance of the molecular descriptors for the most predictive BBB model were determined by use of automatic relevance determination (ARD), and compared with the important descriptors from other literature models of BBB partitioning.

Robust, quantitative tools for modelling ex-vivo expansion of haematopoietic stem cells and progenitors.

Despite substantial research activity on bioreactor design and experiments, there are very few reports of modelling tools that can be used to generate predictive models describing how bioreactor parameters affect performance. New developments in mathematics, such as sparse Bayesian feature selection methods and nonlinear model-free modelling regression methods, offer considerable promise for modelling diverse types of data. The utility of these mathematical tools in stem cell biology are demonstrated by analysis of a large set of bioreactor data derived from the literature. In spite of the diversity of the data sources, and the inherent difficulty in representing bioreactor variables, these modelling methods were able to develop robust, quantitative, predictive models. These models relate bioreactor operational parameters to the degree of expansion of haematopoietic stem cells or their progenitors, and also identify the bioreactor variables that are most likely to affect performance across many experiments. These methods show substantial promise in assisting the design and optimisation of stem cell bioreactors.

Modelling inhalational anaesthetics using bayesian feature selection and QSAR modelling methods.

The development of robust and predictive QSAR models is highly dependent on the use of molecular descriptors that contain information relevant to the property being modelled. Selection of these relevant features from a large pool of possibilities is difficult to achieve effectively. Modern Bayesian methods provide substantial advantages over conventional feature selection methods for feature selection and QSAR modelling. We illustrate the importance of descriptor choice and the beneficial properties of Bayesian methods to select context-dependent relevant descriptors and build robust QSAR models, using data on anaesthetics. Our results show the effectiveness of Bayesian feature selection methods in choosing the best descriptors when these are mixed with less informative descriptors. They also demonstrate the efficacy of the Abraham descriptors and identify deficiencies in ParaSurf descriptors for modelling anaesthetic action.

Stem cell decision making and critical-like exploratory networks.

A sound theoretical or conceptual model of gene regulatory processes that control stem cell fate is still lacking, compromising our ability to manipulate stem cells for therapeutic benefit. The complexity of the regulatory and signaling pathways limits development of useful, predictive models that employ solely reductionist methods using molecular components. However, there is clear evidence from other complex systems that coarse-grained or mesoscale models can yield useful insights and provide workable models for the prediction of some emergent properties such as cell phenotype. We present such a coarse-grained model of stem cell decision making, utilizing the concept of self-organized criticality, which is an order that propagates in some nonequilibrium systems. The model proposes that stochastic gene expression within a stem cell gene regulatory network self-organizes to a critical-like state, characterized by cascades of gene expression that prime various transcriptional programs associated with different cell fates. This diversity of cell fate options is reduced during the decision-making process, which involves a supercritical connectivity in the gene regulatory network as a stem cell leaves its niche microenvironment and an overall increase in transcription occurs. As modules of genes that correspond to specific cell fates approach their critical points, competitive interactions occur between them that are influenced by prevailing microenvironmental conditions. The conceptual model incorporates both intrinsic and extrinsic factors governing stem cell fate and provides a logical pathway to the development of a computational model. We further suggest that rapid self-organized criticality, rather than self-organized criticality, best describes the mesoscale organization of gene regulatory networks.

Toward novel universal descriptors: charge fingerprints.

Although there are a myriad of molecular descriptors for QSAR described in the literature, many descriptors contain similar information as others or are information poor. Recent work has suggested that it may be possible to discover a relatively small pool of 'universal' descriptors from which subsets can be drawn to build a diverse variety of models. We describe a new type of descriptor of this type, the charge fingerprint. This descriptor family can build good QSAR models of a diverse range of physicochemical and biological properties and can be calculated quickly and easily. It appears to be useful for modeling large data sets and has potential for screening large virtual libraries.

Predictive mesoscale network model of cell fate decisions during C. elegans embryogenesis.

The differentiation pathway of the nematode worm model organism C. elegans has been studied as a surrogate for future work on the human embryonic stem cell genetic networks. We extend earlier work on recursive networks by the introduction of a regularizer and more robust convergence algorithms, and by training the model to recapitulate experimental gene expression patterns rather than random expression patterns. We also assess the ability of the model to predict the expression profile on the next cell(s) in the lineage. The weight matrix from the model may be interpreted as a set of rules that guides the differentiation of the cells via a set of regulatory factors: internal genes or external entities. The activity of the regulatory factors shows patterns across the differentiation pathway that reflect the left- or right-hand split. Using these patterns, it may be possible to identify the actual factors responsible for the differentiation and to interpret the associated weights. The model was able to predict expression profiles of cells not used in training the model with a relatively low error rate.

Toward a Rosetta stone for the stem cell genome: stochastic gene expression, network architecture, and external influences.

We review literature relating to three types of factors known to influence stem cell behavior. These factors are stochastic gene expression, regulatory network architecture, and the influence of external signals, such as those emanating from the niche. Although these factors are considered separately, their shared evolutionary history necessitates integration. Stochastic gene expression pervades network components; network architecture controls, modulates, or exploits this noise while performing additional computation; and such complexity also interplays with factors external to cells. Adequate understanding of each of these components, and how they interact, will lead to a conceptual model of the stem cell regulatory system that can be used to drive hypothesis-driven research and facilitate interpretation of experimental data.

Influence of benign cellular changes in diagnosis of cervical cancer using IR microspectroscopy.

IR spectroscopy and principal components analysis (PCA) of endocervical cells and smears diagnosed with benign cellular changes were investigated to determine the influence of these potential confounding variables in the diagnosis of cervical cancer. Spectral differences in all cell and diagnostic types investigated were found in the phosphodiester and carbohydrate regions. However, the spectral differences in other bands were not distinct enough to allow differentiation between groups. The PCA was successfully used to obtain a separation of normal ectocervical smears from normal endocervical cells and smears diagnosed with inflammation, Candida albicans, and bacterial vaginosis. A separation with a slight overlap of abnormal ectocervical smears from normal endocervical cells, inflammation, and bacterial vaginosis was obtained with PCA. Candida was not separated from abnormal ectocervical smears with any success.

Comparison of linear and nonlinear classification algorithms for the prediction of drug and chemical metabolism by human UDP-glucuronosyltransferase isoforms.

Partial least squares discriminant analysis (PLSDA), Bayesian regularized artificial neural network (BRANN), and support vector machine (SVM) methodologies were compared by their ability to classify substrates and nonsubstrates of 12 isoforms of human UDP-glucuronosyltransferase (UGT), an enzyme "superfamily" involved in the metabolism of drugs, nondrug xenobiotics, and endogenous compounds. Simple two-dimensional descriptors were used to capture chemical information. For each data set, 70% of the data were used for training, and the remainder were used to assess the generalization performance. In general, the SVM methodology was able to produce models with the best predictive performance, followed by BRANN and then PLSDA. However, a small number of data sets showed either equivalent or better predictability using PLSDA, which may indicate relatively linear relationships in these data sets. All SVM models showed predictive ability (>60% of test set predicted correctly) and five out of the 12 test sets showed excellent prediction (>80% prediction accuracy). These models represent the first use of pattern recognition methods to discriminate between substrates and nonsubstrates of human drug metabolizing enzymes and the first thorough assessment of three classification algorithms using multiple metabolic data sets.