TopoFun: a machine learning method to improve the functional similarity of gene co-expression modules

A comprehensive, accurate functional annotation of genes is key to systems-level approaches. As functionally related genes tend to be co-expressed, one possible approach to identify functional modules or supplement existing gene annotations is to analyse gene co-expression. We describe TopoFun, a machine learning method that combines topological and functional information to improve the functional similarity of gene co-expression modules. Using LASSO, we selected topological descriptors that discriminated modules made of functionally related genes and random modules. Using the selected topological descriptors, we performed linear discriminant analysis to construct a topological score that predicted the type of a module, random-like or functional-like. We combined the topological score with a functional similarity score in a fitness function that we used in a genetic algorithm to explore the co-expression network. To illustrate the use of TopoFun, we started from a subset of the Gene Ontology Biological Processes (GO-BPs) and showed that TopoFun efficiently retrieved genes that we omitted, and aggregated a number of novel genes to the initial GO-BP while improving module topology and functional similarity. Using an independent protein-protein interaction database, we confirmed that the novel genes gathered by TopoFun were functionally related to the original gene set.

ISoLDE: a data-driven statistical method for the inference of allelic imbalance in datasets with reciprocal crosses

MOTIVATION

Allelic imbalance (AI), i.e. the unequal expression of the alleles of the same gene in a single cell, affects a subset of genes in diploid organisms. One prominent example of AI is parental genomic imprinting, which results in parent-of-origin-dependent, mono-allelic expression of a limited number of genes in metatherian and eutherian mammals and in angiosperms. Currently available methods for identifying AI rely on data modeling and come with the associated limitations.

RESULTS

We have designed ISoLDE (Integrative Statistics of alleLe Dependent Expression), a novel nonparametric statistical method that takes into account both AI and the characteristics of RNA-seq data to infer allelic expression bias when at least two biological replicates are available for reciprocal crosses. ISoLDE learns the distribution of a specific test statistic from the data and calls genes 'allelically imbalanced', 'bi-allelically expressed' or 'undetermined'. Depending on the number of replicates, predefined thresholds or permutations are used to make calls. We benchmarked ISoLDE against published methods, and showed that ISoLDE compared favorably with respect to sensitivity, specificity and robustness to the number of replicates. Using ISoLDE on different RNA-seq datasets generated from hybrid mouse tissues, we did not discover novel imprinted genes (IGs), confirming the most conservative estimations of IG number.

AVAILABILITY AND IMPLEMENTATION

ISoLDE has been implemented as a Bioconductor package available at http://bioconductor.org/packages/ISoLDE/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

In Vitro Corticogenesis from Embryonic Stem Cells Recapitulates the In Vivo Epigenetic Control of Imprinted Gene Expression

In vitro corticogenesis from embryonic stem cells (ESCs) is an attractive model of cortical development and a promising tool for cortical therapy. It is unknown to which extent epigenetic mechanisms crucial for cortex development and function, such as parental genomic imprinting, are recapitulated by in vitro corticogenesis. Here, using genome-wide transcriptomic and methylation analyses on hybrid mouse tissues and cells, we find a high concordance of imprinting status between in vivo and ESC-derived cortices. Notably, in vitro corticogenesis strictly reproduced the in vivo parent-of-origin-dependent expression of 41 imprinted genes (IGs), including Mest and Cdkn1c known to control corticogenesis. Parent-of-origin-dependent DNA methylation was also conserved at 14 of 18 imprinted differentially methylated regions. The least concordant imprinted locus was Gpr1-Zdbf2, where the aberrant bi-allelic expression of Zdbf2 and Adam23 was concomitant with a gain of methylation on the maternal allele in vitro. Combined, our data argue for a broad conservation of the epigenetic mechanisms at imprinted loci in cortical cells derived from ESCs. We propose that in vitro corticogenesis helps to define the still poorly understood mechanisms that regulate imprinting in the brain and the roles of IGs in cortical development.

Insights into an original pocket-ligand pair classification: a promising tool for ligand profile prediction

Pockets are today at the cornerstones of modern drug discovery projects and at the crossroad of several research fields, from structural biology to mathematical modeling. Being able to predict if a small molecule could bind to one or more protein targets or if a protein could bind to some given ligands is very useful for drug discovery endeavors, anticipation of binding to off- and anti-targets. To date, several studies explore such questions from chemogenomic approach to reverse docking methods. Most of these studies have been performed either from the viewpoint of ligands or targets. However it seems valuable to use information from both ligands and target binding pockets. Hence, we present a multivariate approach relating ligand properties with protein pocket properties from the analysis of known ligand-protein interactions. We explored and optimized the pocket-ligand pair space by combining pocket and ligand descriptors using Principal Component Analysis and developed a classification engine on this paired space, revealing five main clusters of pocket-ligand pairs sharing specific and similar structural or physico-chemical properties. These pocket-ligand pair clusters highlight correspondences between pocket and ligand topological and physico-chemical properties and capture relevant information with respect to protein-ligand interactions. Based on these pocket-ligand correspondences, a protocol of prediction of clusters sharing similarity in terms of recognition characteristics is developed for a given pocket-ligand complex and gives high performances. It is then extended to cluster prediction for a given pocket in order to acquire knowledge about its expected ligand profile or to cluster prediction for a given ligand in order to acquire knowledge about its expected pocket profile. This prediction approach shows promising results and could contribute to predict some ligand properties critical for binding to a given pocket, and conversely, some key pocket properties for ligand binding.

In silico prediction of aqueous solubility: a multimodel protocol based on chemical similarity

Aqueous solubility is one of the most important ADMET properties to assess and to optimize during the drug discovery process. At present, accurate prediction of solubility remains very challenging and there is an important need of independent benchmarking of the existing in silico models such as to suggest solutions for their improvement. In this study, we developed a new protocol for improved solubility prediction by combining several existing models available in commercial or free software packages. We first performed an evaluation of ten in silico models for aqueous solubility prediction on several data sets in order to assess the reliability of the methods, and we proposed a new diverse data set of 150 molecules as relevant test set, SolDiv150. We developed a random forest protocol to evaluate the performance of different fingerprints for aqueous solubility prediction based on molecular structure similarity. Our protocol, called a "multimodel protocol", allows selecting the most accurate model for a compound of interest among the employed models or software packages, achieving r(2) of 0.84 when applied to SolDiv150. We also found that all models assessed here performed better on druglike molecules than on real drugs, thus additional improvement is needed in this direction. Overall, our approach enlarges the applicability domain as demonstrated by the more accurate results for solubility prediction obtained using our protocol in comparison to using individual models.

Quantification of the methylation at the GNAS locus identifies subtypes of sporadic pseudohypoparathyroidism type Ib

BACKGROUND

Pseudohypoparathyroidism type Ib (PHP-Ib) is due to epigenetic changes at the imprinted GNAS locus, including loss of methylation at the A/B differentially methylated region (DMR) and sometimes at the XL and AS DMRs and gain of methylation at the NESP DMR.

OBJECTIVE

To investigate if quantitative measurement of the methylation at the GNAS DMRs identifies subtypes of PHP-Ib.

DESIGN AND METHODS

In 19 patients with PHP-Ib and 7 controls, methylation was characterised at the four GNAS DMRs through combined bisulfite restriction analysis and quantified through cytosine specific real-time PCR in blood lymphocyte DNA.

RESULTS

A principal component analysis using the per cent of methylation at seven cytosines of the GNAS locus provided three clusters of subjects (controls n=7, autosomal dominant PHP-Ib with loss of methylation restricted to the A/B DMR n=3, and sporadic PHP-Ib with broad GNAS methylation changes n=16) that matched perfectly the combined bisulfite restriction analysis classification. Furthermore, three sub-clusters of patients with sporadic PHP-Ib, that displayed different patterns of methylation, were identified: incomplete changes at all DMRs compatible with somatic mosaicism (n=5), profound epigenetic changes at all DMRs (n=8), and unmodified methylation at XL in contrast with the other DMRs (n=3). Interestingly, parathyroid hormone concentration at the time of diagnosis correlated with the per cent of methylation at the A/B DMR.

CONCLUSION

Quantitative assessment of the methylation in blood lymphocyte DNA is of clinical relevance, allows the diagnosis of PHP-Ib, and identifies subtypes of PHP-Ib. These epigenetic findings suggest mosaicism at least in some patients.

Designing focused chemical libraries enriched in protein-protein interaction inhibitors using machine-learning methods

Protein-protein interactions (PPIs) may represent one of the next major classes of therapeutic targets. So far, only a minute fraction of the estimated 650,000 PPIs that comprise the human interactome are known with a tiny number of complexes being drugged. Such intricate biological systems cannot be cost-efficiently tackled using conventional high-throughput screening methods. Rather, time has come for designing new strategies that will maximize the chance for hit identification through a rationalization of the PPI inhibitor chemical space and the design of PPI-focused compound libraries (global or target-specific). Here, we train machine-learning-based models, mainly decision trees, using a dataset of known PPI inhibitors and of regular drugs in order to determine a global physico-chemical profile for putative PPI inhibitors. This statistical analysis unravels two important molecular descriptors for PPI inhibitors characterizing specific molecular shapes and the presence of a privileged number of aromatic bonds. The best model has been transposed into a computer program, PPI-HitProfiler, that can output from any drug-like compound collection a focused chemical library enriched in putative PPI inhibitors. Our PPI inhibitor profiler is challenged on the experimental screening results of 11 different PPIs among which the p53/MDM2 interaction screened within our own CDithem platform, that in addition to the validation of our concept led to the identification of 4 novel p53/MDM2 inhibitors. Collectively, our tool shows a robust behavior on the 11 experimental datasets by correctly profiling 70% of the experimentally identified hits while removing 52% of the inactive compounds from the initial compound collections. We strongly believe that this new tool can be used as a global PPI inhibitor profiler prior to screening assays to reduce the size of the compound collections to be experimentally screened while keeping most of the true PPI inhibitors. PPI-HitProfiler is freely available on request from our CDithem platform website, www.CDithem.com.

Peptides reproducibly released by in vivo digestion of beef meat and trout flesh in pigs

Characterisation and identification of peptides (800 to 5000 Da) generated by intestinal digestion of fish or meat were performed using MS analyses (matrix-assisted laser desorption ionisation time of flight and nano-liquid chromatography electrospray-ionisation ion trap MS/MS). Four pigs fitted with cannulas at the duodenum and jejunum received a meal exclusively made of cooked Pectoralis profundus beef meat or cooked trout fillets. A protein-free meal, made of free amino acids, starch and fat, was used to identify peptides of endogenous origin. Peptides reproducibly detected in digesta (i.e. from at least three pigs) were evidenced predominantly in the first 3 h after the meal. In the duodenum, most of the fish- and meat-derived peptides were characteristic of a peptic digestion. In the jejunum, the majority of peptides appeared to result from digestion by chymotrypsin and trypsin. Despite slight differences in gastric emptying kinetics and overall peptide production, possibly in relation to food structure and texture, six and four similar peptides were released after ingestion of fish or meat in the duodenum and jejunum. A total of twenty-six different peptides were identified in digesta. All were fragments of major structural (actin, myosin) or sarcoplasmic (creatine kinase, glyceraldehyde-3-phosphate dehydrogenase and myoglobin) muscle proteins. Peptides were short ( < 2000 Da) and particularly rich in proline residues. Nineteen of them contained bioactive sequences corresponding mainly to an antihypertensive activity. The present work showed that after fish or meat ingestion, among the wide variety of peptides produced by enzymic digestion, some of them can be reproducibly observed in intestinal digesta.