Search

Kernel methods have proven to be useful and successful to analyse large-scale multi-omics datasets [Schölkopf et al., 2004]. However, as stated in [Hofmann et al., 2015, Mariette et al., 2017], these methods usually suffer from a lack of interpretability as the information of thousands descriptors is summarized in a few similarity measures, that can be strongly in uenced by a large number of irrelevant descriptors. To address this issue, feature selection is a widely used strategy: it consist in selecting the most promising features during or prior the analysis. However, most existing methods are proposed in a supervised framework [Tibshirani, 1996, Robnik-Sikonja and Kononenko, 2003, Lin and Tang, 2006]. In the unsupervised framework, the number of proposals is much less important, because there is no objective criterion or value on which to tune the quality of a given feature. Proposals thus aim at preserving at best the similarities between individuals like the SPEC approach [Zhao and Liu, 2007] or at recovering a latent cluster structure, like MCFS [Cai et al., 2010], NDFS [Li et al., 2012] and UDFS [Yang et al., 2011]. In this communication, we will present a feature selection algorithm that explicitly takes advantage of the kernel structure in an unsupervised fashion.

Background: The fourth round of the Critical Assessment of Small Molecule Identification (CASMI) Contest ( www.casmi-contest.org ) was held in 2016, with two new categories for automated methods. This article covers the 208 challenges in Categories 2 and 3, without and with metadata, from organization, participation, results and post-contest evaluation of CASMI 2016 through to perspectives for future contests and small molecule annotation/identification. Results: The Input Output Kernel Regression (CSI:IOKR) machine learning approach performed best in "Category 2: Best Automatic Structural Identification - In Silico Fragmentation Only", won by Team Brouard with 41% challenge wins. The winner of "Category 3: Best Automatic Structural Identification - Full Information" was Team Kind (MS-FINDER), with 76% challenge wins. The best methods were able to achieve over 30% Top 1 ranks in Category 2, with all methods ranking the correct candidate in the Top 10 in around 50% of challenges. This success rate rose to 70% Top 1 ranks in Category 3, with candidates in the Top 10 in over 80% of the challenges. The machine learning and chemistry-based approaches are shown to perform in complementary ways. Conclusions: The improvement in (semi-)automated fragmentation methods for small molecule identification has been substantial. The achieved high rates of correct candidates in the Top 1 and Top 10, despite large candidate numbers, open up great possibilities for high-throughput annotation of untargeted analysis for "known unknowns". As more high quality training data becomes available, the improvements in machine learning methods will likely continue, but the alternative approaches still provide valuable complementary information. Improved integration of experimental context will also improve identification success further for "real life" annotations. The true "unknown unknowns" remain to be evaluated in future CASMI contests. Graphical abstract .

In this paper we apply machine learning methods for predicting protein interactions in fungal secretion pathways. We assume an inter-species transfer setting, where training data is obtained from a single species and the objective is to predict protein interactions in other, related species. In our methodology, we combine several state of the art machine learning approaches, namely, multiple kernel learning (MKL), pairwise kernels and kernelized structured output prediction in the supervised graph inference framework. For MKL, we apply recently proposed centered kernel alignment and p-norm path following approaches to integrate several feature sets describing the proteins, demonstrating improved performance. For graph inference, we apply input-output kernel regression (IOKR) in supervised and semi-supervised modes as well as output kernel trees (OK3). In our experiments simulating increasing genetic distance, Input-Output Kernel Regression proved to be the most robust prediction approach. We also show that the MKL approaches improve the predictions compared to uniform combination of the kernels. We evaluate the methods on the task of predicting protein-protein-interactions in the secretion pathways in fungi, S.cerevisiae, baker’s yeast, being the source, T. reesei being the target of the inter-species transfer learning. We identify completely novel candidate secretion proteins conserved in filamentous fungi. These proteins could contribute to their unique secretion capabilities.

National audience; Nous abordons le problème de la prédiction de liens comme une tâche d'apprentissage d'un noyau de sortie par régression à sortie noyau semi-supervisée. En se plaçant dans le cadre de la théorie des espaces de Hilbert à noyau autoreproduisant à valeurs opérateurs pour des fonctions à valeurs vectorielles, nous établissons un nouveau théorème de représentation dédié à la régression semi-supervisée pour un critère des moindres carrés pénalisé. Nous choisissons ensuite un noyau à valeur opérateur défini à partir d'un noyau d'entrée à valeurs scalaires puis nous construisons un espace de Hilbert avec ce noyau comme noyau autoreproduisant. Nous appliquons ensuite le théorème de représentation. La minimisation des moindres carrés pénalisés dans ce cadre conduit à une solution analytique comme dans le cas de la régression ridge qui est donc ici étendue. Nous étudions la pertinence de cette nouvelle approche semi-supervisée dans le cadre de la prédiction de lien transductive. Des jeux de données artificiels étayent notre étude puis deux applications réelles sont traitées en utilisant un très faible pourcentage de données étiquetées.

International audience; Gene regulatory network inference remains a challenging problem in systems biology despite numerous approaches. When substantial knowledge on a gene regulatory network is already available, supervised network inference also is appropriate. Such a method builds a binary classifier able to assign a class (Regulation/No regulation) to an ordered pair of genes. Once learnt, the classifier can be used to predict new regulations. In this work, we explore the framework of Markov Logic Network (MLN) recently introduced by Richardson and Domingos (2004, 2006). A MLN is a random Markov network that codes for a set of weighted formula. It therefore combines features of probabilistic graphical models with the expressivity of first order logic rules. Starting from a known gene regulatory network involved in the switch proliferation/differentiation of keratinocytes cells, a set of experimental transcriptomic data, and description of genes in terms of GO terms encoded into first order logic, we learn a Markov Logic network, e.g. a set of weighted rules that conclude on the predicate "regulates". As a side contribution, we define a list of basic tests for performance assessment, valid for any binary classifier. A first test consists of measuring the average performance on balanced edge prediction problem; a second one deals with the ability of the classifier, once enhanced by asymmetric bagging, to update a given network; finally a third test measures the ability of the method to predict new interactions with new genes. The numerical studies show that MLNs achieve very good prediction while opening the door to some interpretability of the decisions. Additionally to the ability to suggest new regulation, such an approach allows to cross-validate experimental data with existing knowledge.

International audience; Protein-protein interaction network inference is addressed as an output kernel learning task through semi-supervised Output Kernel Regression. Working in the framework of RKHS theory for vector-valued functions, we establish a new representer theorem devoted to semi-supervised least square regression. We then apply it to get a new model and show its relevance using numerical experiments on artificial networks and a protein-protein interaction network dataset using a very low percentage of labeled proteins in a transductive setting.

International audience; Link prediction is addressed as an output kernel learning task through semi-supervised Output Kernel Regression. Working in the framework of RKHS theory with vector- valued functions, we establish a new repre- senter theorem devoted to semi-supervised least square regression. We then apply it to get a new model (POKR: Penalized Output Kernel Regression) and show its relevance us- ing numerical experiments on arti cial net- works and two real applications using a very low percentage of labeled data in a transduc- tive setting.

In small molecule identification from tandem mass (MS/MS) spectra, input&ndash, output kernel regression (IOKR) currently provides the state-of-the-art combination of fast training and prediction and high identification rates. The IOKR approach can be simply understood as predicting a fingerprint vector from the MS/MS spectrum of the unknown molecule, and solving a pre-image problem to find the molecule with the most similar fingerprint. In this paper, we bring forward the following improvements to the IOKR framework: firstly, we formulate the IOKRreverse model that can be understood as mapping molecular structures into the MS/MS feature space and solving a pre-image problem to find the molecule whose predicted spectrum is the closest to the input MS/MS spectrum. Secondly, we introduce an approach to combine several IOKR and IOKRreverse models computed from different input and output kernels, called IOKRfusion. The method is based on minimizing structured Hinge loss of the combined model using a mini-batch stochastic subgradient optimization. Our experiments show a consistent improvement of top-k accuracy both in positive and negative ionization mode data.

In this paper, we introduce a novel approach, called Input Output Kernel Regression (IOKR), for learning mappings between structured inputs and structured outputs. The approach belongs to the family of Output Kernel Regression methods devoted to regression in feature space endowed with some output kernel. In order to take into account structure in input data and benefit from kernels in the input space as well, we use the Reproducing Kernel Hilbert Space theory for vector-valued functions. We first recall the ridge solution for supervised learning and then study the regularized hinge loss-based solution used in Maximum Margin Regression. Both models are also developed in the context of semi-supervised setting. We also derive an extension of Generalized Cross Validation for model selection in the case of the least-square model. Finally we show the versatility of the IOKR framework on two different problems: link prediction seen as a structured output problem and multi-task regression seen as a multiple and interdependent output problem. Eventually, we present a set of detailed numerical results that shows the relevance of the method on these two tasks.

National audience; In this work, we address the problem of protein-protein interaction network inference as a semi-supervised output kernel learning problem. Using the kernel trick in the output space allows one to reduce the problem of learning from pairs to learning a single variable function with values in a Hilbert space. We turn to the Reproducing Kernel Hilbert Space theory devoted to vector- valued functions, which provides us with a general framework for output kernel regression. In this framework, we propose a novel method which allows to extend Output Kernel Regression to semi-supervised learning. We study the relevance of this approach on transductive link prediction using artificial data and a protein-protein interaction network of S. Cerevisiae using a very low percentage of labeled data.

International audience; BackgroundGene regulatory network inference remains a challenging problem in systems biology despite the numerous approaches that have been proposed. When substantial knowledge on a gene regulatory network is already available, supervised network inference is appropriate. Such a method builds a binary classifier able to assign a class (Regulation/No regulation) to an ordered pair of genes. Once learnt, the pairwise classifier can be used to predict new regulations. In this work, we explore the framework of Markov Logic Networks (MLN) that combine features of probabilistic graphical models with the expressivity of first-order logic rules.ResultsWe propose to learn a Markov Logic network, e.g. a set of weighted rules that conclude on the predicate "regulates", starting from a known gene regulatory network involved in the switch proliferation/differentiation of keratinocyte cells, a set of experimental transcriptomic data and various descriptions of genes all encoded into first-order logic. As training data are unbalanced, we use asymmetric bagging to learn a set of MLNs. The prediction of a new regulation can then be obtained by averaging predictions of individual MLNs. As a side contribution, we propose three in silico tests to assess the performance of any pairwise classifier in various network inference tasks on real datasets. A first test consists of measuring the average performance on balanced edge prediction problem; a second one deals with the ability of the classifier, once enhanced by asymmetric bagging, to update a given network. Finally our main result concerns a third test that measures the ability of the method to predict regulations with a new set of genes. As expected, MLN, when provided with only numerical discretized gene expression data, does not perform as well as a pairwise SVM in terms of AUPR. However, when a more complete description of gene properties is provided by heterogeneous sources, MLN achieves the same performance as a black-box model such as a pairwise SVM while providing relevant insights on the predictions.ConclusionsThe numerical studies show that MLN achieves very good predictive performance while opening the door to some interpretability of the decisions. Besides the ability to suggest new regulations, such an approach allows to cross-validate experimental data with existing knowledge.