A feature selection based framework for histology image classification using global and local heterogeneity quantification.

Biopsy remains the gold standard for the diagnosis of chronic liver diseases. However, the concordance between readers is subject to variability causing an increasing need of objective tissue description methods. A complete framework has been implemented to analyze histological images from any kind of tissue. Based on the feature selection approach, it computes the most relevant subset of descriptors in terms of classification from a wide initial list of local and global descriptors. In comparison with equivalent methods, this implementation is able to find lists of descriptors which are significantly shorter for an equivalent accuracy and furthermore it enables the classification of slides using combinations of global and local measurements. The results have pointed that it could reach an accuracy of 82.8% in a human liver fibrosis grading approach by selecting 6 descriptors from an initial set of 258 global and local descriptors.

Dynamics of rabies virus quasispecies during serial passages in heterologous hosts.

To understand the mutations and genetic rearrangements that allow rabies virus infections of new hosts and adaptation in nature, the quasispecies structure of the nucleoprotein and glycoprotein genes as well as two noncoding sequences of a rabies virus genome were determined. Gene sequences were obtained from the brain and from the salivary glands of the original host, a naturally infected European fox, and after serial passages in mice, dogs, cats and cell culture. A relative genetic stasis of the consensus sequences confirmed previous results about the stability of rabies virus. At the quasispecies level, the mutation frequency varies, in the following order: glycoprotein region (21.9 x 10(-4) mutations per bp), noncoding sequence nucleoprotein-phosphoprotein region (7.2-7.9 x 10(-4) mutations per bp) and nucleoprotein gene region (2.9-3.7 x 10(-4) mutations per bp). These frequencies varied according to the number, type of heterologous passages and the genomic region considered. The shape of the quasispecies structure was dramatically modified by passages in mice, in which the mutation frequencies increased by 12-31 x 10(-4) mutations per bp, depending on the region considered. Non-synonymous mutations were preponderant particularly in the glycoprotein gene, stressing the importance of positive selection in the maintenance and fixation of substitutions. Two mechanisms of genomic evolution of the rabies virus quasispecies, while adapting to environmental changes, have been identified: a limited accumulation of mutations with no replacement of the original master sequence and a less frequent but rapid selective overgrowth of favoured variants.

Genetic polymorphism in the rabies virus nucleoprotein gene.

In an attempt to compare intrinsic and extrinsic genetic diversity of the lyssavirus genotypes, 69 rabies virus isolates from various part of the world were partially sequenced and compared to 13 representative isolates of the 6 lyssavirus genotypes. The analysis of their phylogenetic relationships, performed on the complete nucleoprotein (N) coding gene (1350 bases), established that the rabies virus isolates all belonged to genotype 1 and that at least 11 phylogenetic lineages could be identified in accordance with their geographical localization and species of origin. These lineages diverged mostly by the accumulation of synonymous mutations. Stabilizing selection, possibly related to host specificity, limits amino acid sequence and antigenic drift. Analyses were also performed either on the highly variable 400-base region coding for the amino terminus of the N protein or on the 93-base noncoding region corresponding to the 3' end of the N mRNA, the intergenic N-phosphoprotein (M1) region, and the 5' end of the M1 mRNA. These shorter nucleotide sequences were shown to provide phylogenetic data suitable for the completion of large epidemiological studies, but with less robustness. This latter noncoding sequence, despite a 3.1 times higher mutation rate than its adjacent coding N gene, followed a parallel evolutionary pattern.

Taxonomy and evolutionary studies on lyssaviruses with special reference to Africa.

Currently, the Lyssavirus genus is composed of four serotypes and the ungrouped European bat lyssaviruses (EBL). Using previously described PCR amplification and direct sequencing methods, the N genes of 70 representative lyssaviruses collected from 40 different countries were examined. From the results, a phylogenetic tree with six branches (genotypes) was constructed: genotypes 1-4 correlated with the classification of serotypes 1-4 and in addition EBL 1 and EBL 2 were identified as two further genotypes (5 and 6). Analysis of the 51 rabies (genotype 1) viruses in the study identified nine distinct groups which could be correlated with their geographical origins. A comparison of the six genotypes with four principal Vesiculovirus serotypes showed that their relationship was closer than that of even the two most closely related Vesiculovirus serotypes. Antigenic site sequences of isolates from vaccine/treatment failures were similar to vaccine and field strains, suggesting that failures were not due to genetic variation at the N protein level.

Molecular diversity of the Lyssavirus genus.

The sequence of 5568 nucleotides of the 3' moiety of the Mokola virus genome (serotype 3 of lyssaviruses) encompassing the nucleoprotein (N), phosphoprotein, matrix protein, and glycoprotein genes is presented and compared to that of the vaccinal strains of serotype 1. It allowed us to determine consensus sequences derived from the transcriptional start/stop signals and the order of protein conservation (nucleoprotein > matrix protein > phosphoprotein) in lyssaviruses. The sequences of the N gene of a fox rabies virus isolate from France (serotype 1), Lagos bat virus (serotype 2), Duvenhage virus (serotype 4), two European bat lyssaviruses (EBL) subtype 1, and two EBL subtype 2 were also determined to study the genetic diversity throughout the whole Lyssavirus genus and reinvestigate the classification of this genus. Six clearly distinct genotypes can be distinguished according to their percentage of amino acid similarity. Genotypes 2 (Lagos bat virus) and 3 (Mokola virus) are the most phylogenetically distant from the vaccinal and classical rabies viruses of genotype 1. Genotypes 4 (Duvenhage virus) and 5 (EBL1) are closely related to each other. Genotype 6 is represented by EBL2. Compared to the N proteins of the four principal serotypes of the Vesiculovirus genus (vesicular stomatitis virus serotype New Jersey and serotype Indiana, Chandipura virus, and Piry virus), the N gene of lyssaviruses exhibits a lower genetic variability.

Antigenic and molecular characterization of bat rabies virus in Europe.

The predominant role of Eptesicus serotinus in the epizootic of bat rabies in Europe was further outlined by the first isolation of the rabies virus from this species in France. The distribution of the virus was studied in naturally infected E. serotinus bats at the time of death and suggested that the papillae of the tongue and the respiratory mucosa may play a role in virus production and excretion. The analysis of 501 French rabies virus isolates from various animal species by antinucleocapsid monoclonal antibodies indicated that transmission of the disease from bats to terrestrial animals is unlikely. The antigenic profile of two isolates from French bats corresponded to that of European bat lyssavirus type 1 (EBL1). Comparisons of 12 different isolates from bats with antinucleocapsid and antiglycoprotein monoclonal antibodies and by direct sequencing of the polymerase chain reaction amplification product of the N gene indicated that EBL1, EBL2, Duvenhage virus (serotype 4 of lyssavirus), and the European fox rabies virus (serotype 1) are phylogenetically distant. They formed four tight genetic clusters named genotypes. EBL1 was shown to be antigenically and genetically more closely related to Duvenhage virus than to EBL2. We propose that EBL1 and EBL2 constitute two distinct genotypes which further serologic characterization will probably classify as new serotypes. We also report a simple method for the rapid characterization of EBL based on the digestion of the polymerase chain reaction product of the N gene by three restriction endonucleases.