Antigenic and genetic evolution of swine influenza A (H3N2) viruses in Europe.

In the early 1970s, a human influenza A/Port Chalmers/1/73 (H3N2)-like virus colonized the European swine population. Analyses of swine influenza A (H3N2) viruses isolated in The Netherlands and Belgium revealed that in the early 1990s, antigenic drift had occurred, away from A/Port Chalmers/1/73, the strain commonly used in influenza vaccines for pigs. Here we show that Italian swine influenza A (H3N2) viruses displayed antigenic and genetic changes similar to those observed in Northern European viruses in the same period. We used antigenic cartography methods for quantitative analyses of the antigenic evolution of European swine H3N2 viruses and observed a clustered virus evolution as seen for human viruses. Although the antigenic drift of swine and human H3N2 viruses has followed distinct evolutionary paths, potential cluster-differentiating amino acid substitutions in the influenza virus surface protein hemagglutinin (HA) were in part the same. The antigenic evolution of swine viruses occurred at a rate approximately six times slower than the rate in human viruses, even though the rates of genetic evolution of the HA at the nucleotide and amino acid level were similar for human and swine H3N2 viruses. Continuous monitoring of antigenic changes is recommended to give a first indication as to whether vaccine strains may need updating. Our data suggest that humoral immunity in the population plays a smaller role in the evolutionary selection processes of swine H3N2 viruses than in human H3N2 viruses.

Complex systems analysis: a tool for shock research.

For the past century, students of shock have focused research efforts to illuminate specific mechanisms that cause, or fail as a consequence of, circulatory collapse. Although clinical strategies aimed at supporting or restoring individual organ systems have proven effective, many patients succumb to more generalized multiple organ system failure. We suggest that general biological systems failure cannot be interpreted through reliance on reductionist science. We propose that complex systems analysis is an essential tool for shock research and we evaluate its application to genomic technologies.

Exploring protein sequence space using knowledge-based potentials.

Knowledge-based potentials can be used to decide whether an amino acid sequence is likely to fold into a prescribed native protein structure. We use this idea to survey the sequence-structure relations in protein space. In particular, we test the following two propositions which were found to be important for efficient evolution: the sequences folding into a particular native fold form extensive neutral networks that percolate through sequence space. The neutral networks of any two native folds approach each other to within a few point mutations. Computer simulations using two very different potential functions, M. Sippl's PROSA pair potential and a neural network based potential, are used to verify these claims.

The geometry of shape space: application to influenza.

Shape space was proposed over 20 years ago as a conceptual formalism in which to represent antibody/antigen binding. It has since played a key role in computational immunology. Antigens and antibodies are considered to be points in an abstract "shape space", where coordinates of points in this space represent generalized physico-chemical properties associated with various (unspecified) physical properties related to binding, such as geometric shape, hydrophobicity, charge, etc. Distances in shape space between points representing antibodies and (the shape complement) of antigens are assumed to be related to their affinity, with small distances corresponding to high affinity. In this paper, we provide algorithms, related to metric and ordinal multidimensional scaling algorithms first developed in the mathematical psychology literature, which construct explicit, quantitative coordinates for points in shape space given experimental data such as hemagglutination inhibition assays, or other general affinity assays. Previously, such coordinates had been conceptual constructs and totally implicit. The dimension of shape space deduced from hemagglutination inhibition assays for influenza is low, approximately five dimensional. The deduction of the explicit geometry of shape space given experimental affinity data provides new ways to quantify the similarity of antibodies to antibodies, antigens to antigens, and the affinity of antigens to antibodies. This has potential utility in, e.g. strain selection decisions for annual influenza vaccines, among other applications. The analysis techniques presented here are not restricted to the analysis of antibody-antigen interactions and are generally applicable to affinity data resulting from binding assays.

SAMIE: statistical algorithm for modeling interaction energies.

We are investigating the rules that govern protein-DNA interactions, using a statistical mechanics based formalism that is related to the Boltzmann Machine of the neural net literature. Our approach is data-driven, in which probabilistic algorithms are used to model protein-DNA interactions, given SELEX and/or phage data as input. In the current report, we trained the network using SELEX data, under the "one-to-one" model of interactions (i.e. one amino acid contacts one base). The trained network was able to successfully identify the wild-type binding sites of EGR and MIG protein families. The predictions using our method are the same or better than that of methods existing in the literature. However our methodology offers the potential to capitalise in quantitative detail, as well as to be used to explore more general model of interactions, given availability of data.

Timing the ancestor of the HIV-1 pandemic strains.

HIV-1 sequences were analyzed to estimate the timing of the ancestral sequence of the main group of HIV-1, the strains responsible for the AIDS pandemic. Using parallel supercomputers and assuming a constant rate of evolution, we applied maximum-likelihood phylogenetic methods to unprecedented amounts of data for this calculation. We validated our approach by correctly estimating the timing of two historically documented points. Using a comprehensive full-length envelope sequence alignment, we estimated the date of the last common ancestor of the main group of HIV-1 to be 1931 (1915-41). Analysis of a gag gene alignment, subregions of envelope including additional sequences, and a method that relaxed the assumption of a strict molecular clock also supported these results.

Superadditive correlation.

The fact that correlation does not imply causation is well known. Correlation between variables at two sites does not imply that the two sites directly interact, because, e.g., correlation between distant sites may be induced by chaining of correlation between a set of intervening, directly interacting sites. Such "noncausal correlation" is well understood in statistical physics: an example is long-range order in spin systems, where spins which have only short-range direct interactions, e.g., the Ising model, display correlation at a distance. It is less well recognized that such long-range "noncausal" correlations can in fact be stronger than the magnitude of any causal correlation induced by direct interactions. We call this phenomenon superadditive correlation (SAC). We demonstrate this counterintuitive phenomenon by explicit examples in (i) a model spin system and (ii) a model continuous variable system, where both models are such that two variables have multiple intervening pathways of indirect interaction. We apply the technique known as decimation to explain SAC as an additive, constructive interference phenomenon between the multiple pathways of indirect interaction. We also explain the effect using a definition of the collective mode describing the intervening spin variables. Finally, we show that the SAC effect is mirrored in information theory, and is true for mutual information measures in addition to correlation measures. Generic complex systems typically exhibit multiple pathways of indirect interaction, making SAC a potentially widespread phenomenon. This affects, e.g., attempts to deduce interactions by examination of correlations, as well as, e.g., hierarchical approximation methods for multivariate probability distributions, which introduce parameters based on successive orders of correlation.

Genetic correlations in mutation processes.

We study the role of phylogenetic trees on correlations in mutation processes. Generally, correlations decay exponentially with the generation number. We find that two distinct regimes of behavior exist. For mutation rates smaller than a critical rate, the underlying tree morphology is almost irrelevant, while mutation rates higher than this critical rate lead to strong tree-dependent correlations. We show analytically that identical critical behavior underlies all multiple point correlations. This behavior generally characterizes branching processes undergoing mutation.

Neural net representations of empirical protein potentials.

Recently, there has been considerable interest in deriving and applying knowledge-based, empirical potential functions for proteins. These empirical potentials have been derived from the statistics of interacting, spatially neighboring residues, as may be obtained from databases of known protein crystal structures. In this paper we employ neural networks to redefine empirical potential functions from the point of view of discrimination functions. This approach generalizes previous work, in which simple frequency counting statistics are used on a database of known protein structures. This generalization allows us to avoid restriction to strictly pairwise interactions. Instead of frequency counting to fix adjustable parameters, one now optimizes an objective function involving a neural network parameterized probability distribution. We show how our method reduces to previous work in special situations, but also allows extensions to include orders of interaction beyond pairwise interaction. Given the close packing of proteins, steric interactions etc., the inclusion of higher order interactions is critical for developing an accurate potential. A key feature in the approach we advocate is the development of a representation to describe the spatial location of interacting residues that exist in a sphere of small fixed radius around each residue. This is a "shape representation" problem that has a natural solution for the interaction neighborhoods of protein residues. We demonstrate in a series of numerical experiments that the neural network approach improves discrimination over that obtained by previous methodologies limited to pair-wise interactions.

Neural networks for determining protein specificity and multiple alignment of binding sites.

We use a quantitative definition of specificity to develop a neural network for the identification of common protein binding sites in a collection of unaligned DNA fragments. We demonstrate the equivalence of the method to maximizing Information Content of the aligned sites when simple models of the binding energy and the genome are employed. The network method subsumes those simple models and is capable of working with more complicated ones. This is demonstrated using a Markov model of the E. coli genome and a sampling method to approximate the partition function. A variation of Gibbs' sampling aids in avoiding local minima.

Covariation of mutations in the V3 loop of human immunodeficiency virus type 1 envelope protein: an information theoretic analysis.

The V3 loop of the human immunodeficiency virus type 1 (HIV-1) envelope protein is a highly variable region that is both functionally and immunologically important. Using available amino acid sequences from the V3 region, we have used an information theoretic quantity called mutual information, a measure of covariation, to quantify dependence between mutations in the loop. Certain pairs of sites, including non-contiguous sites along the sequence, do not have independent mutations but display considerable, statistically significant, covarying mutations as measured by mutual information. For the pairs of sites with the highest mutual information, specific amino acids were identified that were highly predictive of amino acids in the linked site. The observed interdependence between variable sites may have implications for structural or functional relationships; separate experimental evidence indicates functional linkage between some of the pairs of sites with high mutual information. Further specific mutational studies of the V3 loop's role in determining viral phenotype are suggested by our analyses. Also, the implications of our results may be important to consider for V3 peptide vaccine design. The methods used here are generally applicable to the study of variable proteins.

Determination of eukaryotic protein coding regions using neural networks and information theory.

Our previous work applied neural network techniques to the problem of discriminating open reading frame (ORF) sequences taken from introns versus exons. The method counted the codon frequencies in an ORF of a specified length, and then used this codon frequency representation of DNA fragments to train a neural net (essentially a Perceptron with a sigmoidal, or "soft step function", output) to perform this discrimination. After training, the network was then applied to a disjoint "predict" set of data to assess accuracy. The resulting accuracy in our previous work was 98.4%, exceeding accuracies reported in the literature at that time for other algorithms. Here, we report even higher accuracies stemming from calculations of mutual information (a correlation measure) of spatially separated codons in exons, and in introns. Significant mutual information exists in exons, but not in introns, between adjacent codons. This suggests that dicodon frequencies of adjacent codons are important for intron/exon discrimination. We report that accuracies obtained using a neural net trained on the frequency of dicodons is significantly higher at smaller fragment lengths than even our original results using codon frequencies, which were already higher than simple statistical methods that also used codon frequencies. We also report accuracies obtained from including codon and dicodon statistics in all six reading frames, i.e. the three frames on the original and complement strand. Inclusion of six-frame statistics increases the accuracy still further. We also compare these neural net results to a Bayesian statistical prediction method that assumes independent codon frequencies in each position. The performance of the Bayesian scheme is poorer than any of the neural based schemes, however many methods reported in the literature either explicitly, or implicitly, use this method. Specifically, Bayesian prediction schemes based on codon frequencies achieve 90.9% accuracy on 90 codon ORFs, while our best neural net scheme reaches 99.4% accuracy on 60 codon ORFs. "Accuracy" is defined as the average of the exon and intron sensitivities. Achievement of sufficiently high accuracies on short fragment lengths can be useful in providing a computational means of finding coding regions in unannotated DNA sequences such as those arising from the mega-base sequencing efforts of the Human Genome Project. We caution that the high accuracies reported here do not represent a complete solution to the problem of identifying exons in "raw" base sequences. The accuracies are considerably lower from exons of small length, although still higher than accuracies reported in the literature for other methods. Short exon lengths are not uncommon.(ABSTRACT TRUNCATED AT 400 WORDS)

Predicting protein secondary structure using neural net and statistical methods.

A comparison of neural network methods and Bayesian statistical methods is presented for prediction of the secondary structure of proteins given their primary sequence. The Bayesian method makes the unphysical assumption that the probability of an amino acid occurring in each position in the protein is independent of the amino acids occurring elsewhere. However, we find the predictive accuracy of the Bayesian method to be only minimally less than the accuracy of the most sophisticated methods used to date. We present the relationship of neural network methods to Bayesian statistical methods and show that, in principle, neural methods offer considerable power, although apparently they are not particularly useful for this problem. In the process, we derive a neural formalism in which the output neurons directly represent the conditional probabilities of structure class. The probabilistic formalism allows introduction of a new objective function, the mutual information, which translates the notion of correlation as a measure of predictive accuracy into a useful training measure. Although a similar accuracy to other approaches (utilizing a mean-square error) is achieved using this new measure, the accuracy on the training set is significantly and tantalizingly higher, even though the number of adjustable parameters remains the same. The mutual information measure predicts a greater fraction of helix and sheet structures correctly than the mean-square error measure, at the expense of coil accuracy, precisely as it was designed to do. By combining the two objective functions, we obtain a marginally improved accuracy of 64.4%, with Matthews coefficients C alpha, C beta and Ccoil of 0.40, 0.32 and 0.42, respectively. However, since all methods to date perform only slightly better than the Bayes algorithm, which entails the drastic assumption of independence of amino acids, one is forced to conclude that little progress has been made on this problem, despite the application of a variety of sophisticated algorithms such as neural networks, and that further advances will require a better understanding of the relevant biophysics.