Ilya B. Muchnik

Recognition of a protein fold in the context of the SCOP classification

A computational method has been developed for the assignment of a protein sequence to a folding class in the Structural Classification of Proteins (SCOP). This method uses global descriptors of a primary protein sequence in terms of the physical, chemical, and structural properties of the constituent amino acids. Neural networks are utilized to combine these descriptors in a way to discriminate members of a given fold from members of all other folds. An extensive testing of the method has been performed to evaluate its prediction accuracy. The method is applicable for the fold assignment of any protein sequence with or without significant sequence homology to known proteins. A WWW page for predicting protein folds is available at URL http://cbcg.lbl.gov/. Proteins 1999;35:401–407.

A biologically consistent model for comparing molecular phylogenies.

In the framework of the problem of combining different gene trees into a unique species phylogeny, a model for duplication/speciation/loss events along the evolutionary tree is introduced. The model is employed for embedding a phylogeny tree into another one via the so-called duplication/speciation principle requiring that the gene duplicated evolves in such a way that any of the contemporary species involved bears only one of the gene copies diverged. The number of biologically meaningful elements in the embedding result (duplications, losses, information gaps) is considered a (asymmetric) dissimilarity measure between the trees. The model duplication concept is compared with that one defined previously in terms of a mapping procedure for the trees. A graph-theoretic reformulation of the measure is derived.

Prevalence of wet litter and the associated risk factors in broiler flocks in the United Kingdom

A postal questionnaire was sent to the managers of 857 broiler farms in the UK to determine the prevalence and risk factors for wet litter. The response rate was 75 per cent. Wet litter was reported by 75 per cent (95 per cent confidence interval [CI] 71·3 to 78·3) of the respondents in at least one flock during the year 2001 and 56·1 per cent (95 per cent CI 52·0 to 60·0) of them reported that they had an outbreak of wet litter in their most recently reared flock. Wet litter occurred more often during the winter months and farms using side ventilation systems were at an increased risk (odds ratio 1·74; 95 per cent CI 1·09 to 2·76). A multivariable analysis was carried out using two different definitions of wet litter as outcome variables – all cases of wet litter, and cases of wet litter associated with disease. Consistent risk factors for both outcomes were coccidiosis, feed equipment failures and the availability of separate farm clothing for each house. Cases of wet litter associated with disease were reported by 33·7 per cent (95 per cent CI 28·8 to 39·1) of the managers in their last flock and were associated with the use of hand sanitisers and broiler houses with walls made of concrete.

Ortholog Clustering on a Multipartite Graph

We present a method for automatically extracting groups of orthologous genes from a large set of genomes by a new clustering algorithm on a weighted multipartite graph. The method assigns a score to an arbitrary subset of genes from multiple genomes to assess the orthologous relationships between genes in the subset. This score is computed using sequence similarities between the member genes and the phylogenetic relationship between the corresponding genomes. An ortholog cluster is found as the subset with the highest score, so ortholog clustering is formulated as a combinatorial optimization problem. The algorithm for finding an ortholog cluster runs in time O(|E| + |V| log |V|), where V and E are the sets of vertices and edges, respectively, in the graph. However, if we discretize the similarity scores into a constant number of bins, the runtime improves to O(|E| + |V|). The proposed method was applied to seven complete eukaryote genomes on which the manually curated database of eukaryotic ortholog clusters, KOG, is constructed. A comparison of our results with the manually curated ortholog clusters shows that our clusters are well correlated with the existing clusters

Layered clusters of tightness set functions

A method for structural clustering is proposed involving data on subset-to-entity linkages that can be calculated with structural data such as graphs or sequences or images. The method is based on the layered structure of the problem of maximization of a set function defined as the minimum value of linkages between a set and its elements and referred to as the tightness function. When the linkage function is monotone, the layered cluster can be easily found with a greedy type algorithm.

Early warning indicators for hock burn in broiler flocks

Hock burn is a common disease of broiler chickens affecting flock welfare and farmer income. Here we use hierarchical logistic regression (HLR) models to identify risk factors for hock burn using data from 5895 flocks, collected over 3.5 years by a large UK broiler company. The results suggest that at 2 weeks of age, weight and weight density may be useful predictors of flocks at risk of a high incidence of hock burn. In contrast, stocking density at placement is not. The use of these and other variables in disease prevention add value to routinely collected management data and can assist in improving broiler welfare and farm income.

Logical analysis of Chinese labor productivity patterns

Using data published by the Chinese Statistical Bureau, an elaborated version of theCobb‐Douglas production function was developed in [3] to express the dependence that industrial production has on classic economic factors, ownership‐related variables and geographic location. In this paper, we reexamine the same data using the new Boolean‐based methodology of Logical Analysis of Data (LAD). The LAD models detect numerous characteristic patterns for explaining changes in productivity, strongly confirm and complement the conclusions of [3], and lead to a decision support system aimed at increasing productivity in Chinas provinces.

Featureless Pattern Recognition in an Imaginary Hilbert Space and Its Application to Protein Fold Classification

The featureless pattern recognition methodology based on measuring some numerical characteristics of similarity between pairs of entities is applied to the problem of protein fold classification. In computational biology, a commonly adopted way of measuring the likelihood that two proteins have the same evolutionary origin is calculating the so-called alignment score between two amino acid sequences that shows properties of inner product rather than those of a similarity measure. Therefore, in solving the problem of determining the membership of a protein given by its amino acid sequence (primary structure) in one of preset fold classes (spatial structure), we treat the set of all feasible amino acid sequences as a subset of isolated points in an imaginary space in which the linear operations and inner product are defined in an arbitrary unknown manner, but without any conjecture on the dimension, i.e. as a Hilbert space.

Fixed points approach to clustering

Assume that a dissimilarity measure between elements and subsets of the set being clustered is given. We define the transformation of the set of subsets under which each subset is transformed into the set of all elements whose dissimilarity to it is not greater than a given threshold. Then a cluster is defined as a fixed point of this transformation. Three well-known clustering strategies are considered from this point of view: hierarchical clustering, graph-theoretic methods, and conceptual clustering. For hierarchical clustering generalizations are obtained that allow for overlapping clusters and/or clusters not forming a cover. Three properties of dissimilarity are introduced which guarantee the existence of fixed points for each threshold. We develop the relation to the theory of quasi-concave set functions, to help give an additional interpretation of clusters.

Featureless pattern recognition in an imaginary Hilbert space

The featureless methodology is applied to the class of pattern recognition problems in which the adopted pairwise similarity measure possesses the most fundamental property of inner product to form a nonnegative definite matrix for any finite assembly of objects. It is proposed to treat the set of all feasible objects of recognition as a subset of isolated points in an imaginary Hilbert space. This idea is applied to the problem of determining the membership of a protein given by its amino acid sequence (primary structure) in one of preset fold classes (spatial structure) on the basis of measuring the likelihood that two proteins have the same evolutionary origin by way of calculating the so-called alignment score between two amino acid sequences, as it is commonly adopted in computational biology.