Search

Posté dans MedRxiv le 6 juillet 2023; The birth-death exposed-infectious (BDEI) model describes the transmission of pathogens featuring an incubation period (when the host is already infected but not yet infectious), for example Ebola and SARS-CoV-2. In a phylodynamics framework, it serves to infer such epidemiological parameters as the basic reproduction number R 0 , the incubation period and the infectious time from a phylogenetic tree (a genealogy of pathogen sequences). With constantly growing sequencing data, the BDEI model should be extremely useful for unravelling information on pathogen epidemics. However, existing implementations of this model in a phylodynamic framework have not yet caught up with the sequencing speed. While the accuracy of estimations should increase with data set size, existing BDEI implementations are limited to medium data sets of up to 500 samples, for both computing time and numerical instability reasons. We improve accuracy and drastically reduce computing time for the BDEI model by rewriting its differential equations in a highly parallelizable way, and by using a combination of numerical analysis methods for their efficient resolution. Our implementation takes one minute on a phylogenetic tree of 10 000 samples. We compare our parameter estimator to the existing implementations on simulated data. Results show that we are not only much faster (50 000 times), but also more accurate. An application of our method to the 2014 Ebola epidemic in Sierra-Leone is also convincing, with very fast calculation and precise estimates. Our BDEI estimator should become an important tool for routine epidemiological surveillance. It is available at github.com/evolbioinfo/BDEI . Significance Statement Phylodynamics uses pathogen genomes as a source of information for epidemiological parameter inference. With constantly growing genome sequence availability, phylodynamics has a high potential for shedding light on epidemics, especially in the beginning, when classical epidemiological data (e.g. incidence curves) are yet limited. However due to the high complexity of differential equations used in phylodynamic models, current implementations suffer from numerical instability and are only applicable to datasets of limited size (

Evolutionary convergences are observed at all levels, from phenotype to DNA and protein sequences, and the changes observed at these different levels tend to be strongly correlated. Here we propose a simulation-based method to detect positions under convergent evolution in large protein alignments, without prior knowledge on the phenotype and environmental constraints. A phylogeny is inferred from the data and used in simulations to estimate the expected number of amino-acid changes in stable evolutionary constraints (null model) for each position. Similarly, we count the number of mutations towards the same amino acid in the data and test if they are occurring more often than expected. We applied our method to two real datasets: HIV reverse transcriptase and fish rhodopsin, and to HIVlike simulated data. On the latter, with known convergent events and substitution model, we detected on average two third of these events, with a low fraction of false positives. With HIV data, one knows that drug resistance mutations (DRMs) are convergent. Even without any knowledge of patient treatment status, we retrieved more than 70% of positions corresponding to known DRMs. On the rhodopsin dataset, four substitutions are supposed to be convergent, as they change the maximum wavelength absorption of the photoreceptor and occurred several times independently during evolution. We detected three of them. These results demonstrate the potential of the method to target specific mutations to be further studied experimentally or, for example, using a nonsynonymous/synonymous rate ratio approach. Our software named ConDor is available at http://condor.pasteur.cloud.

Drug resistance mutations (DRMs) appear in HIV under treatment pressure. DRMs are commonly transmitted to naive patients. The standard approach to reveal new DRMs is to test for significant frequency differences of mutations between treated and naive patients. However, we then consider each mutation individually and cannot hope to study interactions between several mutations. Here, we aim to leverage the ever-growing quantity of high-quality sequence data and machine learning methods to study such interactions (i.e. epistasis), as well as try to find new DRMs. We trained classifiers to discriminate between Reverse Transcriptase Inhibitor (RTI)-experienced and RTI-naive samples on a large HIV-1 reverse transcriptase (RT) sequence dataset from the UK (n ≈ 55, 000), using all observed mutations as binary representation features. To assess the robustness of our findings, our classifiers were evaluated on independent data sets, both from the UK and Africa. Important representation features for each classifier were then extracted as potential DRMs. To find novel DRMs, we repeated this process by removing either features or samples associated to known DRMs. When keeping all known resistance signal, we detected sufficiently prevalent known DRMs, thus validating the approach. When removing features corresponding to known DRMs, our classifiers retained some prediction accuracy, and six new mutations significantly associated with resistance were identified. These six mutations have a low genetic barrier, are correlated to known DRMs, and are spatially close to either the RT active site or the regulatory binding pocket. When removing both known DRM features and sequences containing at least one known DRM, our classifiers lose all prediction accuracy. These results likely indicate that all mutations directly conferring resistance have been found, and that our newly discovered DRMs are accessory or compensatory mutations. Moreover, apart from the accessory nature of the relationships we found, we did not find any significant signal of further, more subtle epistasis combining several mutations which individually do not seem to confer any resistance., Author summary Almost all drugs to treat HIV target the Reverse Transcriptase (RT) and Drug resistance mutations (DRMs) appear in HIV under treatment pressure. Resistant strains can be transmitted and limit treatment options at the population level. Classically, multiple statistical testing is used to find DRMs, by comparing virus sequences of treated and naive populations. However, with this method, each mutation is considered individually and we cannot hope to reveal any interaction (epistasis) between them. Here, we used machine learning to discover new DRMs and study potential epistasis effects. We applied this approach to a very large UK dataset comprising ≈ 55, 000 RT sequences. Results robustness was checked on different UK and African datasets. Six new mutations associated to resistance were found. All six have a low genetic barrier and show high correlations with known DRMs. Moreover, all these mutations are close to either the active site or the regulatory binding pocket of RT. Thus, they are good candidates for further wet experiments to establish their role in drug resistance. Importantly, our results indicate that epistasis seems to be limited to the classical scheme where primary DRMs confer resistance and associated mutations modulate the strength of the resistance and/or compensate for the fitness cost induced by DRMs.

International audience; SARS-CoV-2 is the virus responsible for the global COVID19 pandemic. We review what is known about the origin of this virus, detected in China at the end of December 2019. The genome of this virus mainly evolves under the effect of point mutations. These are generally neutral and have no impact on virulence and severity, but some appear to influence infectivity, notably the D614G mutation of the Spike protein. To date (30/09/2020) no recombination of the virus has been documented in the human host, and very few insertions and deletions. The worldwide spread of the virus was the subject of controversies that we summarize, before proposing a new approach free from the limitations of previous methods. The results show a complex scenario with, for example, numerous introductions to the USA and returns of the virus from the USA to certain countries including France.; Le SARS-CoV-2 est le virus responsable de la pandémie mondiale de COVID19. On dresse ici un bilan de ce qui est connu sur l’origine de ce virus, détecté en Chine fin décembre 2019. Le génome de ce virus évolue sous l’effet de mutations ponctuelles. Celles-ci sont généralement neutres et sans impact sur la virulence et la sévérité, mais certaines semblent influer sur l’infectiosité, notamment la mutation D614G de la protéine Spike. A l’inverse, on n’a à ce jour (30/09/2020) documenté aucune recombinaison du virus chez l’hôte humain, et très peu d’insertions et de délétions. La propagation mondiale du virus a fait l’objet de polémiques sur lesquelles nous revenons, avant de proposer une nouvelle approche débarrassée des limites des méthodes précédentes. Les résultats montrent une propagation complexe avec, par exemple, de très nombreuses introductions aux USA et des retours du virus depuis les USA vers certains pays dont la France.

International audience

International audience

International audience

International audience

International audience

International audience; Les manuels de biologie ont longtemps écrit que le génome du virus de l’immunodéficience humaine (VIH) était composé de neuf gènes: les trois gènes rétroviraux structuraux ( gag , pol et env ) et un ensemble de six gènes régulateurs ou accessoires. Suggérée à la fin des années 1980 [1] , l’existence d’un dixième gène, asp, restait débattue. Nous venons de confirmer sa présence dans le groupe pandémique du VIH, en analysant près de 23 000 séquences du VIH-1 et du SIV (virus d’immuno- déficience simien)

Inferring epidemiological parameters such as the R0 from time-scaled phylogenies is a timely challenge. Most current approaches rely on likelihood functions, which raise specific issues that range from computing these functions to finding their maxima numerically. Here, we present a new regression-based Approximate Bayesian Computation (ABC) approach, which we base on a large variety of summary statistics intended to capture the information contained in the phylogeny and its corresponding lineage-through-time plot. The regression step involves the Least Absolute Shrinkage and Selection Operator (LASSO) method, which is a robust machine learning technique. It allows us to readily deal with the large number of summary statistics, while avoiding resorting to Markov Chain Monte Carlo (MCMC) techniques. To compare our approach to existing ones, we simulated target trees under a variety of epidemiological models and settings, and inferred parameters of interest using the same priors. We found that, for large phylogenies, the accuracy of our regression-ABC is comparable to that of likelihood-based approaches involving birth-death processes implemented in BEAST2. Our approach even outperformed these when inferring the host population size with a Susceptible-Infected-Removed epidemiological model. It also clearly outperformed a recent kernel-ABC approach when assuming a Susceptible-Infected epidemiological model with two host types. Lastly, by re-analyzing data from the early stages of the recent Ebola epidemic in Sierra Leone, we showed that regression-ABC provides more realistic estimates for the duration parameters (latency and infectiousness) than the likelihood-based method. Overall, ABC based on a large variety of summary statistics and a regression method able to perform variable selection and avoid overfitting is a promising approach to analyze large phylogenies., Author summary Given the rapid evolution of many pathogens, analysing their genomes by means of phylogenies can inform us about how they spread. This is the focus of the field known as “phylodynamics”. Most existing methods inferring epidemiological parameters from virus phylogenies are limited by the difficulty of handling complex likelihood functions, which commonly incorporate latent variables. Here, we use an alternative method known as regression-based Approximate Bayesian Computation (ABC), which circumvents this problem by using simulations and dataset comparisons. Since phylogenies are difficult to compare to one another, we introduce many summary statistics to describe them and take advantage of current machine learning techniques able to perform variable selection. We show that the accuracy we reach is comparable to that of existing methods. This accuracy increases with phylogeny size and can even be higher than that of existing methods for some parameters. Overall, regression-based ABC opens new perspectives to infer epidemiological parameters from large phylogenies.

International audience; Phylogenies provide a useful way to understand the evolutionary history of genetic samples, and data sets with more than a thousand taxa are becoming increasingly common, notably with viruses (e.g., human immunodefi- ciency virus (HIV)). Dating ancestral events is one of the first, essential goals with such data. However, current sophisticated probabilistic approaches struggle to handle data sets of this size. Here,we present very fast dating algorithms, based on a Gaussian model closely related to the Langley-Fitch molecular-clock model. We show that this model is robust to uncorrelated violations of the molecular clock. Our algorithms apply to serial data, where the tips of the tree have been sampled through times. They estimate the substitution rate and the dates of all ancestral nodes. When the input tree is unrooted, they can provide an estimate for the root position, thus representing a new, practical alternative to the standard rooting methods (e.g., midpoint). Our algorithms exploit the tree (recursive) structure of the problem at hand, and the close relationships between least squares and linear algebra. We distinguish between an unconstrained setting and the case where the temporal prece- dence constraint (i.e., an ancestral node must be older that its daughter nodes) is accounted for. With rooted trees, the former is solved using linear algebra in linear computing time (i.e., proportional to the number of taxa), while the resolution of the latter, constrained setting, is based on an active-set method that runs in nearly linear time. With unrooted trees the computing time becomes (nearly) quadratic (i.e., proportional to the square of the number of taxa). In all cases, very large input trees (¿10,000 taxa) can easily be processed and transformed into time-scaled trees. We compare these algorithms to standard methods (root-to-tip, r8s version of Langley-Fitchmethod, and BEAST). Using simulated data, we show that their estimation accuracy is similar to that of the most sophisticated methods, while their computing time is much faster. We apply these algorithms on a large data set comprising 1194 strains of Influenza virus from the pdm09 H1N1 Human pandemic. Again the results show that these algorithms provide a very fast alternative with results similar to those of other computer programs. These algorithms are implemented in the LSD software (least-squares dating), which can be downloaded from http://www.atgc-montpellier.fr/LSD/, along with all our data sets and detailed results.

International audience; Probabilistic models of character change/substitutions; Distance (minimum evolution, NJ, Fitch-Margoliash, etc.) and Parsimony methods; Likelihood and Bayesian methods; Tree search; Sources of error, both systematic and non-systematic.

International audience; A popular approach in phylogenetics consists in estimating a matrix of evolutionary distances between pairs of taxa, and then using this information to reconstruct a phylogenetic tree for those taxa. In this entry, we first explain how distances should be defined and estimated, and then focus on the task of inferring a phylogenetic tree that accounts for the estimated distances. Specifically, we will introduce the classical guiding principles for tree inference – least squares and minimum evolution – and then present the most popular distance-based methods – in particular neighbor joining and a wide array of algorithms inspired by it.

International audience; FastME provides distance algorithms to infer phylogenies. FastME is based on balanced minimum evolution, which is the very principle of Neighbor Joining (NJ). FastME improves over NJ by performing topological moves using fast, sophisticated algorithms. The first version of FastME only included Nearest Neighbor Interchange. The new 2.0 version also includes Subtree Pruning and Regrafting, while remaining as fast as NJ and providing a number of facilities: Distance estimation for DNA and proteins with various models and options, bootstrapping, and parallel computations. FastME is available using several interfaces: Command-line (to be integrated in pipelines), PHYLIP-like, and a Web server (http:// www.atgc-montpellier.fr/fastme/).

International audience; This article introduces the Transitive Consistency Score (TCS) web server; a service making it possible to estimate the local reliability of protein multiple sequence alignments (MSAs) using the TCS index. The evaluation can be used to identify the aligned positions most likely to contain structurally analogous residues and also most likely to support an accurate phylogenetic reconstruction. The TCS scoring scheme has been shown to be accurate predictor of structural alignment correctness among commonly used methods. It has also been shown to outper-form common filtering schemes like Gblocks or tri-mAl when doing MSA post-processing prior to phylo-genetic tree reconstruction. The web server is available from http://tcoffee.crg.cat/tcs.

National audience;

International audience

International audience; Probabilistic models of character change/substitutions; Distance (minimum evolution, NJ, Fitch-Margoliash, etc.) and Parsimony methods; Likelihood and Bayesian methods; Tree search; Sources of error, both systematic and non-systematic, and approaches to address them: Stochastic variation, Model mis-specification (base composition; substitution matrices; others), Long-branch attraction, Taxon-sampling; others

International audience; Maximum-likelihood (ML) methods are more and more used to infer phylogenies, because of their high level of accruracy. I'll present the main components and features of this approach: 1) preamble, basic principle and properties; 2) ML distance estimation; 3) tree likelihood computation via the pruning algorithm; 4) heuristics to search the tree space; 5) simulation results, computing times and topological accuracy compared to other methods; 6) approximate likelihood ratio test for branch supports.

International audience; Amino-acid substitution models are essential in most methods to infer phylogenies from protein data. These models represent the ways in which proteins evolve and substitutions accumulate along the course of time. It is widely accepted that the substitution processes vary depending on the structural configuration of the protein residues. However, this information is not (or is rarely) used in phylogenetic studies, though the three-dimensional structure of dozens of thousands of proteins has been elucidated. Here we reinvestigate the question in order to fill this gap. We use an improved estimation methodology and a very large database comprising 1,471 non-redundant globular protein alignments with structural annotations to estimate new amino-acid substitution models accounting for the secondary structure and solvent accessibility of the residues. These models incorporate a confidence coefficient which is estimated from the data and reflects the reliability of structural annotations in the analyzed sequences. Our results with 300 independent test alignments show an impressive likelihood gain, compared to standard models such as JTT or WAG. Moreover, the use of these models induces significant topological changes in the inferred trees, which should be of primary interest to phylogeneticists. Our data, models and software are available for download from http://atgc.lirmm.fr/phyml-structure/.

International audience

International audience

International audience; I'll present recent developments in inferring phylogenies from sequences by maximum likelihood (ML). In fact, these approaches involve two main components: probabilistic modelling of substitution events, and algorithmic to infer near-optimal trees. I'll show that algorithmic has considerably progressed during the last few years, allowing now for fast ML inference from large datasets. On the other hand, I'll underline the difficulties encountered with modelling, especially of protein evolution, and show that elaborating more realistic and improved substitution models is a major challenge.

International audience; Amino-acid replacement models form an essential basis of protein phylogenetics. They are used to compute the substitution probabilities along the branches of phylogenies, and thus the likelihood of the data. They are also essential in protein alignment. A number of replacement models and methods to estimate these models from protein alignments were proposed since the seminal work of Dayhoff et al. (1978). I'll describe the main tendencies and the advances that our group recently achieved in this field. I'll explain how we inferred a new replacement matrix, called LG (following the authors), which clearly improves the standard JTT (Jones et al. 1992) and WAG (Whelan and Goldman 2001) matrices. I'll also describe other modelling approaches based on mixtures and/or site partitioning (e.g. based on solvent exposition, in the line of Goldman et al. 1998) and show that these further improve over LG. The gain in likelihood value between these compound models and standard JTT or WAG is of the same range as the gain brought by the use a gamma distribution to account for variation of rates among sites. Moreover, maximum-likelihood tree topologies tend to be modified when these new models are used, meaning that they should become routine in phylogenetics.

See our Mol. Biol. Evol. paper to obtain the full paper.; International audience; It is nearly 20 years since the landmark paper (Saitou and Nei, 1987) in MBE introducing Neighbor-Joining (NJ). This method has become the most widely-used method for building phylogenetic trees from distances, and the original paper has been cited about 13,000 times (Science Citation Index ). Yet the question ‘what does the NJ method seek to do?' has until recently proved somewhat elusive, leading to some imprecise claims and misunderstanding. However a rigorous answer to this question has recently been provided by further mathematical investigation. The origins of this story lie in a paper by Pauplin (2000) though its continuation has unfolded in more mathematically-inclined literature (Steel and Semple 2004; Desper and Gascuel, 2005). Moreover, these findings explain the superiority of recent minimum evolution-based algorithms over NJ (Desper and Gascuel, 2004; Vinh and Von Haesler, 2005). The conference will follow the same line as a recent note of Gascuel and Steel (2006), aiming to make these results more widely accessible.

National audience

International audience; Presentation of the main approaches to build phylogenetic trees from sequence data. The focus will be on : (1) probabilistic models of sequence evolution (JC, K2P, HKY, GTR, JTT, WAG and LG) ; (2) modeling of rates accross sites using gama distributions ; (3) likelihood calculations; (4) tree inference approaches ; (5) bootstrap and other branch supports. I'll both explain the theory and train the students with practicals using standard programs and web servers (PHYLIP, PhyML, Phylogeny.fr).

INFO/MAB, INFO/ZENITHFinal Report of the ModSysC2020 Working Group at University Montpellier 2; At University Montpellier 2, the modeling and simulation of complex systems has been identified as a major scientific challenge and one of the priority axes in interdisciplinary research, with major potential impact on training, economy and society. Many research groups and laboratories in Montpellier are already working in that direction, but typically in isolation within their own scientific discipline. Several local actions have been initiated in order to structure the scientific community with interdisciplinary projects, but with little coordination among the actions. The goal of the ModSysC2020 (modeling and simulation of complex systems in 2020) working group was to analyze the local situation (forces and weaknesses, current projects), identify the critical research directions and propose concrete actions in terms of research projects, equipment facilities, human resources and training to be encouraged. To guide this perspective, we decomposed the scientific challenge into four main themes, for which there is strong background in Montpellier: (1) modeling and simulation of complex systems; (2) algorithms and computing; (3) scientific data management; (4) production, storage and archiving of data from the observation of the natural and biological media. In this report, for each theme, we introduce the context and motivations, analyze the situation in Montpellier, identify research directions and propose specific actions in terms of interdisciplinary research projects and training. We also provide an analysis of the socio-economical aspects of modeling and simulation through use cases in various domains such as life science and healthcare, environmental science and energy. Finally, we discuss the importance of revisiting students training in fundamental domains such as modeling, computer programming and database which are typically taught too late, in specialized masters.

National audience; La phylogénie moléculaire est un domaine très actif de la bioinformatique. Dans cet épisode du podcast audio, Olivier Gascuel dresse un panorama de cette discipline.

National audience

National audience

International audience; Proceedings of the “Bamako 2009 Conference on the Bioinformatics of Infectious Diseases”, special issue of Infections, Genetics and Evolution, 11(4), Editorial

National audience; Hidden Markov Models (HMMs) have proved to be powerful for protein domain identification. However, numerous domains may be missed in highly divergent proteins. This is the case for the proteins of Plasmodium falciparum, the main causal agent of human malaria. Here, we propose a method that uses domain co-occurrence to increase the sensitivity of the approach while controlling its false discovery rate. Applied to P. falciparum, our method identify (with an error rate below 20%) 482 new domains (versus 3482 in PlasmoDB), which involve 158 new GO annotations.

International audience; This special issue of Infection, Genetics and Evolution is devoted to the Bioinformatics of African Pathogens and Vectors conference, which was held in Nairobi (Kenya), 28th May-2nd June 2007.

International audience; I'll describe how to use our maximum-likelihood phylogeny software PhyML (Guindon&Gascuel 2003). The main parts of my talk will be related to (1) input data, (2) substitution model, (3) tree searching, (4) branch support, (5) ouput files, (6) others (web sites, command line, manual...)

International audience; We introduce a new type of phylogenetic Hidden Markov Model, combining the strength of usual HMM and the knowledge of the phylogeny of a family of sequences.We use such models to look into the genome of a target species for members of the sequence family. Our results on some 690 protein families show a better sensitivity and a better specificity when compared to standard profile HMM or Blast searchs.

International audience; We introduce a new type of phylogenetic Hidden Markov Model, combining the strength of usual HMM and the knowledge of the phylogeny of a family of sequences. We use such models to look into the genome of a target species for members of the sequence family. Our results on some 690 protein families show a better sensitivity and a better specificity when compared to standard profile HMM or Blast searchs.

International audience; Standard models of amino-acid replacement in proteins are based on Markovian rate matrices. I will review recent efforts to improve these models by (1) enhancing rate matrix estimation from large alignment databases, (2) combining several models (matrices) to account for solvent exposition and secondary structure, (3) designing site-specific models to describe biochemical constraints acting on amino-acid substitutions.

July; International audience

International audience; Amino acid replacement matrices are an essential basis of protein phylogenetics. They are used to compute substitution probabilities along phylogeny branches and thus the likelihood of the data. They are also essential in protein alignment. A number of replacement matrices and methods to estimate these matrices from protein alignments have been proposed since the seminal work of Dayhoff et al. (1972). An important advance was achieved by Whelan and Goldman (2001) and their WAG matrix, thanks to an efficient maximum likelihood estimation approach that accounts for the phylogenies of sequences within each training alignment. We further refine this method by incorporating the variability of evolutionary rates across sites in the matrix estimation and using a much larger and diverse database than BRKALN, which was used to estimate WAG. To estimate our new matrix (called LG after the authors), we use an adaptation of the XRATE software and 3,912 alignments from Pfam, comprising approximately 50,000 sequences and approximately 6.5 million residues overall. To evaluate the LG performance, we use an independent sample consisting of 59 alignments from TreeBase and randomly divide Pfam alignments into 3,412 training and 500 test alignments. The comparison with WAG and JTT shows a clear likelihood improvement. With TreeBase, we find that 1) the average Akaike information criterion gain per site is 0.25 and 0.42, when compared with WAG and JTT, respectively; 2) LG is significantly better than WAG for 38 alignments (among 59), and significantly worse with 2 alignments only; and 3) tree topologies inferred with LG, WAG, and JTT frequently differ, indicating that using LG impacts not only the likelihood value but also the output tree. Results with the test alignments from Pfam are analogous. LG and a PHYML implementation can be downloaded from http://atgc.lirmm.fr/LG

Textes issus de la rencontre intitulée Mathematics of evolution & Phylogenetics, tenue à l'Institut Poincaré à Paris en juin 2005; International audience; Evolution is a complex process, acting at multiple scales, from DNA sequences and proteins to populations of species. Understanding and reconstructing evolution is of major importance in numerous subfields of biology. For example, phylogenetics and sequence evolution is central to comparative genomics, attempts to decipher genomes, and molecular epidemiology. Phylogenetics is also the focal point of large-scale international biodiversity assessment initiatives such as the 'Tree of Life' project, which aims to build the evolutionary tree for all extant species. Since the pioneering work in phylogenetics in the 1960s, models have become increasingly sophisticated to account for the inherent complexity of evolution. They rely heavily on mathematics and aim at modelling and analyzing biological phenomena such as horizontal gene transfer, heterogeneity of mutation, and speciation and extinction processes. This book presents these recent models, their biological relevance, their mathematical basis, their properties, and the algorithms to infer them from data. A number of subfields from mathematics and computer science are involved: combinatorics, graph theory, stringology, probabilistic and Markov models, information theory, statistical inference, Monte Carlo methods, continuous and discrete algorithmics. This book arises from the Mathematics of Evolution & Phylogenetics meeting at the Mathematical Institute Henri Poincaré, Paris, in June 2005 and is based on the outstanding state-of-the-art reports presented by the conference speakers. Ten chapters - based around five themes - provide a detailed overview of key topics, from the underlying concepts to the latest results, some of which are at the forefront of current research.

Title: GONNA: a Gene Ontology Nearest Neighbor Approach for the functional prediction of P.falciparum orphan genes Plasmodium falciparum, the pathogenic agent responsible for malaria, causes close to 3 millions human deaths each year in the world. Its genome, published in 2002, remains poorly understood. Among its ~5000 genes, ~3000 have a totally unknown biological function. Several reasons explain this situation, the first one being a totally atypical genome composition which renders ineffective the usual methods of functional annotation based on sequence comparison (alignment) with homologous genes in nearby organisms. Non-homology methods are thus needed to obtain functional clues for the orphan genes. Notably, transcriptomic analysis using DNA microarrays have been proposed. These approaches are based on the hypothesis that uncharacterized genes may potentially share functional roles with annotated genes of similar transcriptomic profile, a principle known as "Guilt by Association" (GBA). Interestingly, this intra-species principle can also be apply to other type of available postgenomic data, such as proteomic data or protein interaction data. The GO Consortium (http://www.geneontology.org) has developed a systematic and standardized nomenclature for annotating genes in various organisms, including P.falciparum. The GO annotation is a hierarchical structure describing generalization relationships between hundreds of terms. So each gene can have several associated GO terms, due to the hierarchical structure of the ontology, and of the existence of multi-functional genes. Here, we propose a low computing time method that allows the application of the GBA principle on the Gene Ontology in a very extensive way, and for several postgenomic data. Notably, it allows intensive used of a "cross-validation" procedure to provide high quality assessment of the predictions of the method for each GO term individually. The result is an assessed functional draft of the P.falciparum orphan genes. This draft has been integrated in a database available on the web and that can be accessed in different ways. A global view allows to rapidly inspect the branches of the ontology that are more suitable for the GBA principle using the considered type of data (transcriptome, proteome, and protein interaction data). Moreover, user can query the database to search for the potential GO terms of one particular gene, or to search for all the genes that potentially belong to a given GO term.

It has been recognized for a long time that substitution processes vary depending on structural configurations. However, this information is not (or rarely) used in phylogenetic studies, even though the structure of dozen thousands of proteins has been elucidated. Here we reinvestigate the question in order to fill this gap. We used a very large dataset comprising 4,389 protein alignments with structural annotations to estimate new amino-acid substitution matrices for various structural configurations. Moreover, we used an independent sample of 500 alignments to evaluate the gain in tree likelihood brought by these new matrices. Various ways to combine these models (matrices) were envisaged, namely, separate analysis based on available annotations, mixtures (assuming no structural information), and a combination of both based on an estimated parameter that reflects the reliability of structural annotations. Our results show that separate analysis and mixtures are nearly equivalent in average, while our confidence-based approach is best thanks to its ability to detect poorly annotated proteins. Highest likelihood values are obtained with six structural categories combining exposed/buried and alpha/beta/other status of the sites; the average gain is as high as 1.16 AIC points per site, compared to standard WAG model. This six-category model is closely followed by the two-category exposed/buried model, while the secondary structure-based three-category model is worse, but still better than WAG. All these likelihood gains induce significant topological changes in the trees being inferred, indicating that our models should be used routinely by phylogeneticists.

Most genomes contain large amounts of orphan genes. For example, 60\% of the P.falciparum genes (the main causal agent of Malaria) lack functional annotation. New approaches, commonly referred as Guilt by Association (GBA) methods, have been proposed to functionally annotate the orphan genes. These methods are based on classification (non-supervised or supervised) applied on post-genomic data. They typically use transcriptome and protein interaction data, but do not exploit paralogy. However, homologous genes from the same species (i.e. paralogues) tend to share similar function(s). Results: This article focus on the effects of paralogy on GBA methods. We illustrate the strength of the dependence between functional annotations and paralogy, and show that applying any GBA method without accounting for this form of homology has two opposite effects: it leads us to over-estimate the method accuracy with orphan genes, and to lose the benefit of paralogy. We present and discuss a resampling algorithm to correctly estimate the performance of any GBA method, as well as a simple, general scheme that can be combined with any GBA method in order to benefit from paralogy. Both procedures are used to measure the bias and benefit on transcriptomic data from Yeast and P.falciparum, two organisms that are well and poorly annotated, respectively. Our results show that both the bias and the benefit induced by paralogy may be substantial, depending on the GBA methods to be considered, the data and the organisms. Conclusions: The search of annotated paralogues should be incorporated in the design of any GBA methods. Moreover, our resampling procedure should be used routinely to obtain predictions with unbiased performance estimates, which is of first importance, eg, to chose among contradictory predictions, or to select target genes to perform wet experiments.

International audience; The guilt by association (GBA) principle is used in several supervised and non-supervised methods to functionally annotate uncharacterized genes from transcriptomic data or from other information source. However, these methods do not distinguish between genes which have or have not intra-species homologues. We show here that functional annotation and intra-species homology are strongly dependent. We emphasize that applying any GBA method not accounting for this form of homology has two opposite effects: it leads to over-estimating the method performance on the genes with no intra-species homologues, and to losing the benefit of homology on the other genes. Bias and benefit are measured on \Pf\ and Yeast, and a general scheme to properly apply the GBA principle is given. All together, this method improves over previous standard applications of the GBA principle.

National audience; Motivations: Les protéines du complexe majeur d’histocompatibilité (MHC) assurent une fonction essentielle au sein du système immunitaire, en présentant des peptides du soi ou des peptides antigéniques aux récepteurs T. La liaison non covalente de la beta2-microglobuline aux protéines du MHC de classe I est nécessaire à leur expression à la surface cellulaire et à la présentation des peptides. La superfamille du MHC regroupe les protéines de structure homologue aux protéines MHC-I. Ces protéines sont impliquées dans une grande variété de processus biologiques et interagissent ou non avec la beta2-microglobuline. La prédiction de la liaison (ou de l'absence de liaison) à la beta2-microglobuline, pour des protéines de la superfamille du MHC nouvellement identifiées, permettrait d’une part d’indiquer leur mécanisme de reconnaissance moléculaire, et d’autre part de détecter des mutants pathologiques dont l’expression à la surface cellulaire est affectée. La description standardisée des domaines et la méthode d'alignement (pour partie structurale) mises en place au sein d'IMGT s’appliquent avec succès aux protéines de la superfamille du MHC malgré leur faible similarité de séquence, et fournissent une numérotation unique des résidus qui favorise le développement d'une telle approche prédictive.Résultats: La méthode proposée combine un classifieur Bayesien dit naïf et la numérotation unique IMGT. Elle est composée de deux étapes : un ensemble de descripteurs binaires discriminants (associant une position dans l’alignement et un groupe d’acides aminés) est tout d’abord extrait des données ; les fréquences de ces descripteurs sont ensuite estimées conditionnellement aux classes que l’on cherche à séparer, pour construire le classifieur. Nous appliquons cette approche à un jeu de données composé de 807 séquences alignées, correspondant aux allèles de 47 gènes de la superfamille du MHC. 18 descripteurs sont sélectionnés pour leur capacité à discriminer les protéines selon qu’elles se lient ou non à la beta2-microglobuline. L’analyse structurale des protéines du jeu de données montre que ces descripteurs correspondent à des sites potentiels de contact à la beta2-microglobuline ou à des sites impliqués dans le maintien d’une conformation favorable au contact. La performance du classifieur est évaluée par la procédure de ''leave-one-out'', déclinée en 3 modes qui distinguent les cas où la prédiction concerne un nouveau gène, une espèce non référencée au sein des données ou un nouveau type de récepteur, avec respectivement 98%, 93% et 79% de succès. Ces taux élevés de bonne prédiction mettent en évidence l’efficacité de l’approche proposée, qui devrait trouver des applications dans d’autres problématiques biologiques.Données complémentaires: Les séquences alignées de la superfamille du MHC qui composent le jeu de données sont accessibles dans les sections MHC et RPI d’IMGT Repertoire ; les structures actuellement résolues sont accessibles sous forme de fichiers de coordonnées annotés dans IMGT/3Dstructure-DB (http://imgt.cines.fr).

International audience; This book considers evolution at different scales: sequences, genes, gene families, organelles, genomes and species. The focus is on the mathematical and computational tools and concepts, which form an essential basis of evolutionary studies, indicate their limitations, and give them orientation. Recent years have witnessed rapid progress in the mathematics of evolution and phylogeny, with models and methods becoming more realistic, powerful, and complex.

International audience