BACKGROUND: Relationships among genetic or epigenetic features can be explored by learning probabilistic networks and unravelling the dependencies among a set of given genetic/epigenetic features. Bayesian networks (BNs) consist of nodes that represent the variables and arcs that represent the probabilistic relationships between the variables. However, practical guidance on how to make choices among the wide array of possibilities in Bayesian network analysis is limited. Our study aimed to apply a BN approach, while clearly laying out our analysis choices as an example for future researchers, in order to provide further insights into the relationships among epigenetic features and a stressful condition in chickens (Gallus gallus). RESULTS: Chickens raised under control conditions (n = 22) and chickens exposed to a social isolation protocol (n = 24) were used to identify differentially methylated regions (DMRs). A total of 60 DMRs were selected by a threshold, after bioinformatic pre-processing and analysis. The treatment was included as a binary variable (control = 0; stress = 1). Thereafter, a BN approach was applied: initially, a pre-filtering test was used for identifying pairs of features that must not be included in the process of learning the structure of the network; then, the average probability values for each arc of being part of the network were calculated; and finally, the arcs that were part of the consensus network were selected. The structure of the BN consisted of 47 out of 61 features (60 DMRs and the stressful condition), displaying 43 functional relationships. The stress condition was connected to two DMRs, one of them playing a role in tight and adhesive intracellular junctions in organs such as ovary, intestine, and brain. CONCLUSIONS: We clearly explain our steps in making each analysis choice, from discrete BN models to final generation of a consensus network from multiple model averaging searches. The epigenetic BN unravelled functional relationships among the DMRs, as well as epigenetic features in close association with the stressful condition the chickens were exposed to. The DMRs interacting with the stress condition could be further explored in future studies as possible biomarkers of stress in poultry species.

Pattern classification methods assign an object to one of several predefined classes/categories based on features extracted from observed attributes of the object (pattern). When L discriminatory features for the pattern can be accurately determined, the pattern classification problem presents no difficulty. However, precise identification of the relevant features for a classification algorithm (classifier) to be able to categorize real world patterns without errors is generally infeasible. In this case, the pattern classification problem is often cast as devising a classifier that minimizes the misclassification rate. One way of doing this is to consider both the pattern attributes and its class label as random variables, estimate the posterior class probabilities for a given pattern and then assign the pattern to the class/category for which the posterior class probability value estimated is maximum. More often than not, the form of the posterior class probabilities is unknown. The so-called Parzen Window approach is widely employed to estimate class-conditional probability (class-specific probability) densities for a given pattern. These probability densities can then be utilized to estimate the appropriate posterior class probabilities for that pattern. However, the Parzen Window scheme can become computationally impractical when the size of the training dataset is in the tens of thousands and L is also large (a few hundred or more). Over the years, various schemes have been suggested to ameliorate the computational drawback of the Parzen Window approach, but the problem still remains outstanding and unresolved. In this paper, we revisit the Parzen Window technique and introduce a novel approach that may circumvent the aforementioned computational bottleneck. The current paper presents the mathematical aspect of our idea. Practical realizations of the proposed scheme will be given elsewhere.

BACKGROUND: In this work we predict enzyme function at the level of chemical mechanism, providing a finer granularity of annotation than traditional Enzyme Commission (EC) classes. Hence we can predict not only whether a putative enzyme in a newly sequenced organism has the potential to perform a certain reaction, but how the reaction is performed, using which cofactors and with susceptibility to which drugs or inhibitors, details with important consequences for drug and enzyme design. Work that predicts enzyme catalytic activity based on 3D protein structure features limits the prediction of mechanism to proteins already having either a solved structure or a close relative suitable for homology modelling. RESULTS: In this study, we evaluate whether sequence identity, InterPro or Catalytic Site Atlas sequence signatures provide enough information for bulk prediction of enzyme mechanism. By splitting MACiE (Mechanism, Annotation and Classification in Enzymes database) mechanism labels to a finer granularity, which includes the role of the protein chain in the overall enzyme complex, the method can predict at 96% accuracy (and 96% micro-averaged precision, 99.9% macro-averaged recall) the MACiE mechanism definitions of 248 proteins available in the MACiE, EzCatDb (Database of Enzyme Catalytic Mechanisms) and SFLD (Structure Function Linkage Database) databases using an off-the-shelf K-Nearest Neighbours multi-label algorithm. CONCLUSION: We find that InterPro signatures are critical for accurate prediction of enzyme mechanism. We also find that incorporating Catalytic Site Atlas attributes does not seem to provide additional accuracy. The software code (ml2db), data and results are available online at http://sourceforge.net/projects/ml2db/ and as supplementary files.

Phylogenomic analysis of the occurrence and abundance of protein domains in proteomes has recently showed that the α/β architecture is probably the oldest fold design. This holds important implications for the origins of biochemistry. Here we explore structure-function relationships addressing the use of chemical mechanisms by ancestral enzymes. We test the hypothesis that the oldest folds used the most mechanisms. We start by tracing biocatalytic mechanisms operating in metabolic enzymes along a phylogenetic timeline of the first appearance of homologous superfamilies of protein domain structures from CATH. A total of 335 enzyme reactions were retrieved from MACiE and were mapped over fold age. We define a mechanistic step type as one of the 51 mechanistic annotations given in MACiE, and each step of each of the 335 mechanisms was described using one or more of these annotations. We find that the first two folds, the P-loop containing nucleotide triphosphate hydrolase and the NAD(P)-binding Rossmann-like homologous superfamilies, were α/β architectures responsible for introducing 35% (18/51) of the known mechanistic step types. We find that these two oldest structures in the phylogenomic analysis of protein domains introduced many mechanistic step types that were later combinatorially spread in catalytic history. The most common mechanistic step types included fundamental building blocks of enzyme chemistry: "Proton transfer," "Bimolecular nucleophilic addition," "Bimolecular nucleophilic substitution," and "Unimolecular elimination by the conjugate base." They were associated with the most ancestral fold structure typical of P-loop containing nucleotide triphosphate hydrolases. Over half of the mechanistic step types were introduced in the evolutionary timeline before the appearance of structures specific to diversified organisms, during a period of architectural diversification. The other half unfolded gradually after organismal diversification and during a period that spanned ∼2 billion years of evolutionary history.

BACKGROUND: Teaching bioinformatics at universities is complicated by typical computer classroom settings. As well as running software locally and online, students should gain experience of systems administration. For a future career in biology or bioinformatics, the installation of software is a useful skill. We propose that this may be taught by running the course on GNU/Linux running on inexpensive Raspberry Pi computer hardware, for which students may be granted full administrator access. RESULTS: We release 4273π, an operating system image for Raspberry Pi based on Raspbian Linux. This includes minor customisations for classroom use and includes our Open Access bioinformatics course, 4273π Bioinformatics for Biologists. This is based on the final-year undergraduate module BL4273, run on Raspberry Pi computers at the University of St Andrews, Semester 1, academic year 2012-2013. CONCLUSIONS: 4273π is a means to teach bioinformatics, including systems administration tasks, to undergraduates at low cost.

BACKGROUND: We present the algorithm PFClust (Parameter Free Clustering), which is able automatically to cluster data and identify a suitable number of clusters to group them into without requiring any parameters to be specified by the user. The algorithm partitions a dataset into a number of clusters that share some common attributes, such as their minimum expectation value and variance of intra-cluster similarity. A set of n objects can be clustered into any number of clusters from one to n, and there are many different hierarchical and partitional, agglomerative and divisive, clustering methodologies available that can be used to do this. Nonetheless, automatically determining the number of clusters present in a dataset constitutes a significant challenge for clustering algorithms. Identifying a putative optimum number of clusters to group the objects into involves computing and evaluating a range of clusterings with different numbers of clusters. However, there is no agreed or unique definition of optimum in this context. Thus, we test PFClust on datasets for which an external gold standard of 'correct' cluster definitions exists, noting that this division into clusters may be suboptimal according to other reasonable criteria. PFClust is heuristic in the sense that it cannot be described in terms of optimising any single simply-expressed metric over the space of possible clusterings. RESULTS: We validate PFClust firstly with reference to a number of synthetic datasets consisting of 2D vectors, showing that its clustering performance is at least equal to that of six other leading methodologies - even though five of the other methods are told in advance how many clusters to use. We also demonstrate the ability of PFClust to classify the three dimensional structures of protein domains, using a set of folds taken from the structural bioinformatics database CATH. CONCLUSIONS: We show that PFClust is able to cluster the test datasets a little better, on average, than any of the other algorithms, and furthermore is able to do this without the need to specify any external parameters. Results on the synthetic datasets demonstrate that PFClust generates meaningful clusters, while our algorithm also shows excellent agreement with the correct assignments for a dataset extracted from the CATH part-manually curated classification of protein domain structures.

BACKGROUND: We investigate the relationships between the EC (Enzyme Commission) class, the associated chemical reaction, and the reaction mechanism by building predictive models using Support Vector Machine (SVM), Random Forest (RF) and k-Nearest Neighbours (kNN). We consider two ways of encoding the reaction mechanism in descriptors, and also three approaches that encode only the overall chemical reaction. Both cross-validation and also an external test set are used. RESULTS: The three descriptor sets encoding overall chemical transformation perform better than the two descriptions of mechanism. SVM and RF models perform comparably well; kNN is less successful. Oxidoreductases and hydrolases are relatively well predicted by all types of descriptor; isomerases are well predicted by overall reaction descriptors but not by mechanistic ones. CONCLUSIONS: Our results suggest that pairs of similar enzyme reactions tend to proceed by different mechanisms. Oxidoreductases, hydrolases, and to some extent isomerases and ligases, have clear chemical signatures, making them easier to predict than transferases and lyases. We find evidence that isomerases as a class are notably mechanistically diverse and that their one shared property, of substrate and product being isomers, can arise in various unrelated ways.The performance of the different machine learning algorithms is in line with many cheminformatics applications, with SVM and RF being roughly equally effective. kNN is less successful, given the role that non-local information plays in successful classification. We note also that, despite a lack of clarity in the literature, EC number prediction is not a single problem; the challenge of predicting protein function from available sequence data is quite different from assigning an EC classification from a cheminformatics representation of a reaction.

MOTIVATION: Accurately predicting the binding affinities of large sets of diverse protein-ligand complexes is an extremely challenging task. The scoring functions that attempt such computational prediction are essential for analysing the outputs of molecular docking, which in turn is an important technique for drug discovery, chemical biology and structural biology. Each scoring function assumes a predetermined theory-inspired functional form for the relationship between the variables that characterize the complex, which also include parameters fitted to experimental or simulation data and its predicted binding affinity. The inherent problem of this rigid approach is that it leads to poor predictivity for those complexes that do not conform to the modelling assumptions. Moreover, resampling strategies, such as cross-validation or bootstrapping, are still not systematically used to guard against the overfitting of calibration data in parameter estimation for scoring functions. RESULTS: We propose a novel scoring function (RF-Score) that circumvents the need for problematic modelling assumptions via non-parametric machine learning. In particular, Random Forest was used to implicitly capture binding effects that are hard to model explicitly. RF-Score is compared with the state of the art on the demanding PDBbind benchmark. Results show that RF-Score is a very competitive scoring function. Importantly, RF-Score's performance was shown to improve dramatically with training set size and hence the future availability of more high-quality structural and interaction data is expected to lead to improved versions of RF-Score. CONTACT: [email protected]; [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Functionally analogous enzymes are those that catalyze similar reactions on similar substrates but do not share common ancestry, providing a window on the different structural strategies nature has used to evolve required catalysts. Identification and use of this information to improve reaction classification and computational annotation of enzymes newly discovered in the genome projects would benefit from systematic determination of reaction similarities. Here, we quantified similarity in bond changes for overall reactions and catalytic mechanisms for 95 pairs of functionally analogous enzymes (non-homologous enzymes with identical first three numbers of their EC codes) from the MACiE database. Similarity of overall reactions was computed by comparing the sets of bond changes in the transformations from substrates to products. For similarity of mechanisms, sets of bond changes occurring in each mechanistic step were compared; these similarities were then used to guide global and local alignments of mechanistic steps. Using this metric, only 44% of pairs of functionally analogous enzymes in the dataset had significantly similar overall reactions. For these enzymes, convergence to the same mechanism occurred in 33% of cases, with most pairs having at least one identical mechanistic step. Using our metric, overall reaction similarity serves as an upper bound for mechanistic similarity in functional analogs. For example, the four carbon-oxygen lyases acting on phosphates (EC 4.2.3) show neither significant overall reaction similarity nor significant mechanistic similarity. By contrast, the three carboxylic-ester hydrolases (EC 3.1.1) catalyze overall reactions with identical bond changes and have converged to almost identical mechanisms. The large proportion of enzyme pairs that do not show significant overall reaction similarity (56%) suggests that at least for the functionally analogous enzymes studied here, more stringent criteria could be used to refine definitions of EC sub-subclasses for improved discrimination in their classification of enzyme reactions. The results also indicate that mechanistic convergence of reaction steps is widespread, suggesting that quantitative measurement of mechanistic similarity can inform approaches for functional annotation.

BACKGROUND: The need for fast and accurate scoring functions has been driven by the increased use of in silico virtual screening twinned with high-throughput screening as a method to rapidly identify potential candidates in the early stages of drug development. We examine the ability of some the most common scoring functions (GOLD, ChemScore, DOCK, PMF, BLEEP and Consensus) to discriminate correctly and efficiently between active and non-active compounds among a library of approximately 3,600 diverse decoy compounds in a virtual screening experiment against heat shock protein 90 (Hsp90). RESULTS: Firstly, we investigated two ranking methodologies, GOLDrank and BestScorerank. GOLDrank is based on ranks generated using GOLD. The various scoring functions, GOLD, ChemScore, DOCK, PMF, BLEEP and Consensus, are applied to the pose ranked number one by GOLD for that ligand. BestScorerank uses multiple poses for each ligand and independently chooses the best ranked pose of the ligand according to each different scoring function. Secondly, we considered the effect of introducing the Thr184 hydrogen bond tether to guide the docking process towards a particular solution, and its effect on enrichment. Thirdly, we considered normalisation to account for the known bias of scoring functions to select larger molecules. All the scoring functions gave fairly similar enrichments, with the exception of PMF which was consistently the poorest performer. In most cases, GOLD was marginally the best performing individual function; the Consensus score usually performed similarly to the best single scoring function. Our best results were obtained using the Thr184 tether in combination with the BestScorerank protocol and normalisation for molecular weight. For that particular combination, DOCK was the best individual function; DOCK recovered 90% of the actives in the top 10% of the ranked list; Consensus similarly recovered 89% of the actives in its top 10%. CONCLUSION: Overall, we demonstrate the validity of virtual screening as a method for identifying new leads from a pool of ligands with similar physicochemical properties and we believe that the outcome of this study provides useful insight into the setting up of a suitable docking and scoring protocol, resulting in enrichment of 'target active' compounds.

SUMMARY: MACiE (mechanism, annotation and classification in enzymes) is a publicly available web-based database, held in CMLReact (an XML application), that aims to help our understanding of the evolution of enzyme catalytic mechanisms and also to create a classification system which reflects the actual chemical mechanism (catalytic steps) of an enzyme reaction, not only the overall reaction. AVAILABILITY: http://www-mitchell.ch.cam.ac.uk/macie/.

Chemical information is now seen as critical for most areas of life sciences. But unlike Bioinformatics, where data is openly available and freely re-usable, most chemical information is closed and cannot be re-distributed without permission. This has led to a failure to adopt modern informatics and software techniques and therefore paucity of chemistry in bioinformatics. New technology, however, offers the hope of making chemical data (compounds and properties) free during the authoring process. We argue that the technology is already available; we require a collective agreement to enhance publication protocols.

The current methods of publishing chemical information in bioscience articles are analysed. Using 3 papers as use-cases, it is shown that conventional methods using human procedures, including cut-and-paste are time-consuming and introduce errors. The meaning of chemical terms and the identity of compounds is often ambiguous. valuable experimental data such as spectra and computational results are almost always omitted. We describe an Open XML architecture at proof-of-concept which addresses these concerns. Compounds are identified through explicit connection tables or links to persistent Open resources such as PubChem. It is argued that if publishers adopt these tools and protocols, then the quality and quantity of chemical information available to bioscientists will increase and the authors, publishers and readers will find the process cost-effective.

SUMMARY: The Protein Ligand Database (PLD) is a publicly available web-based database that aims to provide further understanding of protein-ligand interactions. The PLD contains biomolecular data including calculated binding energies, Tanimoto ligand similarity scores and protein percentage sequence similarities. The database has potential for application as a tool in molecular design. AVAILABILITY: http://www-mitchell.ch.cam.ac.uk/pld/