In this work, we propose an original method for aggregating multiple clustering coming from different sources of information.Each partition is encoded by a co-membership matrix between observations.Our approach uses a mixture of Stochastic Block Models (SBM) to group co-membership matrices with similar information into components and to partition observations into different clusters, taking into account their specificities within the components.The parameters are estimated using a Variational Bayesian EM algorithm.The Bayesian framework allows for selecting an optimal numbers of clusters and components.

[This corrects the article DOI: 10.1371/journal.pcbi.1007732.].

The use of comparative genomics for functional, evolutionary, and epidemiological studies requires methods to classify gene families in terms of occurrence in a given species. These methods usually lack multivariate statistical models to infer the partitions and the optimal number of classes and don't account for genome organization. We introduce a graph structure to model pangenomes in which nodes represent gene families and edges represent genomic neighborhood. Our method, named PPanGGOLiN, partitions nodes using an Expectation-Maximization algorithm based on multivariate Bernoulli Mixture Model coupled with a Markov Random Field. This approach takes into account the topology of the graph and the presence/absence of genes in pangenomes to classify gene families into persistent, cloud, and one or several shell partitions. By analyzing the partitioned pangenome graphs of isolate genomes from 439 species and metagenome-assembled genomes from 78 species, we demonstrate that our method is effective in estimating the persistent genome. Interestingly, it shows that the shell genome is a key element to understand genome dynamics, presumably because it reflects how genes present at intermediate frequencies drive adaptation of species, and its proportion in genomes is independent of genome size. The graph-based approach proposed by PPanGGOLiN is useful to depict the overall genomic diversity of thousands of strains in a compact structure and provides an effective basis for very large scale comparative genomics. The software is freely available at https://github.com/labgem/PPanGGOLiN.

BACKGROUND: Genome-Wide Association Studies (GWAS) seek to identify causal genomic variants associated with rare human diseases. The classical statistical approach for detecting these variants is based on univariate hypothesis testing, with healthy individuals being tested against affected individuals at each locus. Given that an individual's genotype is characterized by up to one million SNPs, this approach lacks precision, since it may yield a large number of false positives that can lead to erroneous conclusions about genetic associations with the disease. One way to improve the detection of true genetic associations is to reduce the number of hypotheses to be tested by grouping SNPs. RESULTS: We propose a dimension-reduction approach which can be applied in the context of GWAS by making use of the haplotype structure of the human genome. We compare our method with standard univariate and group-based approaches on both synthetic and real GWAS data. CONCLUSION: We show that reducing the dimension of the predictor matrix by aggregating SNPs gives a greater precision in the detection of associations between the phenotype and genomic regions.

BACKGROUND: A large amount of research has been devoted to the detection and investigation of epistatic interactions in genome-wide association studies (GWASs). Most of the literature focuses on low-order interactions between single-nucleotide polymorphisms (SNPs) with significant main effects. RESULTS: In this paper we propose an original approach for detecting epistasis at the gene level, without systematically filtering on significant genes. We first compute interaction variables for each gene pair by finding its Eigen-Epistasis component, defined as the linear combination of Gene SNPs having the highest correlation with the phenotype. The selection of significant effects is done using a penalized regression method based on Group Lasso controlling the False Discovery Rate. CONCLUSION: The method is tested against two recent alternative proposals from the literature using synthetic data, and shows good performances in different settings. We demonstrate the power of our approach by detecting new gene-gene interactions on three genome-wide association studies.

Assessing the uncertainty pertaining to the conclusions derived from experimental data is challenging when there is a high number of possible explanations compared to the number of experiments. We propose a new two-stage “screen and clean” procedure for assessing the uncertainties pertaining to the selection of relevant variables in high-dimensional regression problems. In this two-stage method, screening consists in selecting a subset of candidate variables by a sparsity-inducing penalized regression, while cleaning consists in discarding all variables that do not pass a significance test. This test was originally based on ordinary least squares regression. We propose to improve the procedure by conveying more information from the screening stage to the cleaning stage. Our cleaning stage is based on an adaptively penalized regression whose weights are adjusted in the screening stage. Our procedure is amenable to the computation of p-values, allowing to control the False Discovery Rate. Our experiments show the benefits of our procedure, as we observe a systematic improvement of sensitivity compared to the original procedure.

BACKGROUND: Genome-wide association studies (GWAS) aim at finding genetic markers that are significantly associated with a phenotype of interest. Single nucleotide polymorphism (SNP) data from the entire genome are collected for many thousands of SNP markers, leading to high-dimensional regression problems where the number of predictors greatly exceeds the number of observations. Moreover, these predictors are statistically dependent, in particular due to linkage disequilibrium (LD). We propose a three-step approach that explicitly takes advantage of the grouping structure induced by LD in order to identify common variants which may have been missed by single marker analyses (SMA). In the first step, we perform a hierarchical clustering of SNPs with an adjacency constraint using LD as a similarity measure. In the second step, we apply a model selection approach to the obtained hierarchy in order to define LD blocks. Finally, we perform Group Lasso regression on the inferred LD blocks. We investigate the efficiency of this approach compared to state-of-the art regression methods: haplotype association tests, SMA, and Lasso and Elastic-Net regressions. RESULTS: Our results on simulated data show that the proposed method performs better than state-of-the-art approaches as soon as the number of causal SNPs within an LD block exceeds 2. Our results on semi-simulated data and a previously published HIV data set illustrate the relevance of the proposed method and its robustness to a real LD structure. The method is implemented in the R package BALD (Blockwise Approach using Linkage Disequilibrium), available from http://www.math-evry.cnrs.fr/publications/logiciels . CONCLUSIONS: Our results show that the proposed method is efficient not only at the level of LD blocks by inferring well the underlying block structure but also at the level of individual SNPs. Thus, this study demonstrates the importance of tailored integration of biological knowledge in high-dimensional genomic studies such as GWAS.

SUMMARY: The R package SIMoNe (Statistical Inference for MOdular NEtworks) enables inference of gene-regulatory networks based on partial correlation coefficients from microarray experiments. Modelling gene expression data with a Gaussian graphical model (hereafter GGM), the algorithm estimates non-zero entries of the concentration matrix, in a sparse and possibly high-dimensional setting. Its originality lies in the fact that it searches for a latent modular structure to drive the inference procedure through adaptive penalization of the concentration matrix. AVAILABILITY: Under the GNU General Public Licence at http://cran.r-project.org/web/packages/simone/

Most similarity or dissimilarity measures used in merging and splitting segmentation methods include in almost all cases a single radiometrical information, integrate rarely geometrical information and ignore the high level knowledge on the image. Consequently, the region hierarchies issued from these approaches may suffer from a structural instability and deficiency in the semantic of the regions due to the image content, its high variability and the complexity of the meaningful regions which compose this image. In this paper, we propose to enhance the "semantic" content of the hierarchy by means of an additional term called "contextual cost". This term integrates the high level knowledge on the image which is derived from a classifier after a supervised learning on the semantic classes composing the image. Its purpose is to better guide the merging process towards the construction of meaningful regions.

This paper proposes a feature selection method for content based image retrieval. The original method is developed in the context of semisupervised learning. It is illustrated using medical image database for diagnoses purposes.

In many discrimination problems a large amount of data is available but only a few of them are labeled. This provides a strong motivation to improve or develop methods for semi-supervised learning. In this paper, boosting is generalized to this task within the optimization framework of MarginBoost . We extend the margin definition to unlabeled data and develop the gradient descent algorithm that corresponds to the resulting margin cost function. This meta-learning scheme can be applied to any base classifier able to benefit from unlabeled data. We propose here to apply it to mixture models trained with an Expectation-Maximization algorithm. Promising results are presented on benchmarks with different rates of labeled data.