Antonio Reverter

Combining partial correlation and an information theory approach to the reversed engineering of gene co-expression networks

MOTIVATION We present PCIT, an algorithm for the reconstruction of gene co-expression networks (GCN) that combines the concept partial correlation coefficient with information theory to identify significant gene to gene associations defining edges in the reconstruction of GCN. The properties of PCIT are examined in the context of the topology of the reconstructed network including connectivity structure, clustering coefficient and sensitivity. RESULTS We apply PCIT to a series of simulated datasets with varying levels of complexity in terms of number of genes and experimental conditions, as well as to three real datasets. Results show that, as opposed to the constant cutoff approach commonly used in the literature, the PCIT algorithm can identify and allow for more moderate, yet not less significant, estimates of correlation (r) to still establish a connection in the GCN. We show that PCIT is more sensitive than established methods and capable of detecting functionally validated gene-gene interactions coming from absolute r values as low as 0.3. These bona fide associations, which often relate to genes with low variation in expression patterns, are beyond the detection limits of conventional fixed-threshold methods, and would be overlooked by studies relying on those methods. AVAILABILITY FORTRAN 90 source code to perform the PCIT algorithm is available as Supplementary File 1.

A validated whole-genome association study of efficient food conversion in cattle.

The genetic factors that contribute to efficient food conversion are largely unknown. Several physiological systems are likely to be important, including basal metabolic rate, the generation of ATP, the regulation of growth and development, and the homeostatic control of body mass. Using whole-genome association, we found that DNA variants in or near proteins contributing to the background use of energy of the cell were 10 times as common as those affecting appetite and body-mass homeostasis. In addition, there was a genic contribution from the extracellular matrix and tissue structure, suggesting a trade-off between efficiency and tissue construction. Nevertheless, the largest group consisted of those involved in gene regulation or control of the phenotype. We found that the distribution of micro-RNA motifs was significantly different for the genetic variants associated with residual feed intake than for the genetic variants in total, although the distribution of promoter sequence motifs was not different. This suggests that certain subsets of micro-RNA are more important for the regulation of this trait. Successful validation depended on the sign of the allelic association in different populations rather than on the strength of the initial association or its size of effect.

A differential wiring analysis of expression data correctly identifies the gene containing the causal mutation.

Transcription factor (TF) regulation is often post-translational. TF modifications such as reversible phosphorylation and missense mutations, which can act independent of TF expression level, are overlooked by differential expression analysis. Using bovine Piedmontese myostatin mutants as proof-of-concept, we propose a new algorithm that correctly identifies the gene containing the causal mutation from microarray data alone. The myostatin mutation releases the brakes on Piedmontese muscle growth by translating a dysfunctional protein. Compared to a less muscular non-mutant breed we find that myostatin is not differentially expressed at any of ten developmental time points. Despite this challenge, the algorithm identifies the myostatin ‘smoking gun’ through a coordinated, simultaneous, weighted integration of three sources of microarray information: transcript abundance, differential expression, and differential wiring. By asking the novel question “which regulator is cumulatively most differentially wired to the abundant most differentially expressed genes?” it yields the correct answer, “myostatin”. Our new approach identifies causal regulatory changes by globally contrasting co-expression network dynamics. The entirely data-driven ‘weighting’ procedure emphasises regulatory movement relative to the phenotypically relevant part of the network. In contrast to other published methods that compare co-expression networks, significance testing is not used to eliminate connections.

Gene expression studies of developing bovine longissimus muscle from two different beef cattle breeds

BackgroundThe muscle fiber number and fiber composition of muscle is largely determined during prenatal development. In order to discover genes that are involved in determining adult muscle phenotypes, we studied the gene expression profile of developing fetal bovine longissimus muscle from animals with two different genetic backgrounds using a bovine cDNA microarray. Fetal longissimus muscle was sampled at 4 stages of myogenesis and muscle maturation: primary myogenesis (d 60), secondary myogenesis (d 135), as well as beginning (d 195) and final stages (birth) of functional differentiation of muscle fibers. All fetuses and newborns (total n = 24) were from Hereford dams and crossed with either Wagyu (high intramuscular fat) or Piedmontese (GDF8 mutant) sires, genotypes that vary markedly in muscle and compositional characteristics later in postnatal life.ResultsWe obtained expression profiles of three individuals for each time point and genotype to allow comparisons across time and between sire breeds. Quantitative reverse transcription-PCR analysis of RNA from developing longissimus muscle was able to validate the differential expression patterns observed for a selection of differentially expressed genes, with one exception. We detected large-scale changes in temporal gene expression between the four developmental stages in genes coding for extracellular matrix and for muscle fiber structural and metabolic proteins. FSTL1 and IGFBP5 were two genes implicated in growth and differentiation that showed developmentally regulated expression levels in fetal muscle. An abundantly expressed gene with no functional annotation was found to be developmentally regulated in the same manner as muscle structural proteins. We also observed differences in gene expression profiles between the two different sire breeds. Wagyu-sired calves showed higher expression of fatty acid binding protein 5 (FABP5) RNA at birth. The developing longissimus muscle of fetuses carrying the Piedmontese mutation shows an emphasis on glycolytic muscle biochemistry and a large-scale up-regulation of the translational machinery at birth. We also document evidence for timing differences in differentiation events between the two breeds.ConclusionTaken together, these findings provide a detailed description of molecular events accompanying skeletal muscle differentiation in the bovine, as well as gene expression differences that may underpin the phenotype differences between the two breeds. In addition, this study has highlighted a non-coding RNA, which is abundantly expressed and developmentally regulated in bovine fetal muscle.

Gene expression profiling of muscle tissue in Brahman steers during nutritional restriction

Expression profiling using microarrays allows for the detailed characterization of the gene networks that regulate an animals response to environmental stresses. During nutritional restriction, processes such as protein turnover, connective tissue remodeling, and muscle atrophy take place in the skeletal muscle of the animal. These processes and their regulation are of interest in the context of managing livestock for optimal production efficiency and product quality. Here we expand on recent research applying complementary DNA (cDNA) microarray technology to the study of the effect of nutritional restriction on bovine skeletal muscle. Using a custom cDNA microarray of 9,274 probes from cattle muscle and s.c. fat libraries, we examined the differential gene expression profile of the LM from 10 Brahman steers under three different dietary treatments. The statistical approach was based on mixed-model ANOVA and model-based clustering of the BLUP solutions for the gene x diet interaction effect. From the results, we defined a transcript profile of 156 differentially expressed array elements between the weight loss and weight gain diet substrates. After sequence and annotation analyses, the 57 upregulated elements represented 29 unique genes, and the 99 downregulated elements represented 28 unique genes. Most of these co-regulated genes cluster into groups with distinct biological function related to protein turnover and cytoskeletal metabolism and contribute to our mechanistic understanding of the processes associated with remodeling of muscle tissue in response to nutritional stress.

Association weight matrix for the genetic dissection of puberty in beef cattle

We describe a systems biology approach for the genetic dissection of complex traits based on applying gene network theory to the results from genome-wide associations. The associations of single-nucleotide polymorphisms (SNP) that were individually associated with a primary phenotype of interest, age at puberty in our study, were explored across 22 related traits. Genomic regions were surveyed for genes harboring the selected SNP. As a result, an association weight matrix (AWM) was constructed with as many rows as genes and as many columns as traits. Each {i, j} cell value in the AWM corresponds to the z-score normalized additive effect of the ith gene (via its neighboring SNP) on the jth trait. Columnwise, the AWM recovered the genetic correlations estimated via pedigree-based restricted maximum-likelihood methods. Rowwise, a combination of hierarchical clustering, gene network, and pathway analyses identified genetic drivers that would have been missed by standard genome-wide association studies. Finally, the promoter regions of the AWM-predicted targets of three key transcription factors (TFs), estrogen-related receptor γ (ESRRG), Pal3 motif, bound by a PPAR-γ homodimer, IR3 sites (PPARG), and Prophet of Pit 1, PROP paired-like homeobox 1 (PROP1), were surveyed to identify binding sites corresponding to those TFs. Applied to our case, the AWM results recapitulate the known biology of puberty, captured experimentally validated binding sites, and identified candidate genes and gene–gene interactions for further investigation.

Genome-wide association studies of female reproduction in tropically adapted beef cattle

The genetics of reproduction is poorly understood because the heritabilities of traits currently recorded are low. To elucidate the genetics underlying reproduction in beef cattle, we performed a genome-wide association study using the bovine SNP50 chip in 2 tropically adapted beef cattle breeds, Brahman and Tropical Composite. Here we present the results for 3 female reproduction traits: 1) age at puberty, defined as age in days at first observed corpus luteum (CL) after frequent ovarian ultrasound scans (AGECL); 2) the postpartum anestrous interval, measured as the number of days from calving to first ovulation postpartum (first rebreeding interval, PPAI); and 3) the occurrence of the first postpartum ovulation before weaning in the first rebreeding period (PW), defined from PPAI. In addition, correlated traits such as BW, height, serum IGF1 concentration, condition score, and fatness were also examined. In the Brahman and Tropical Composite cattle, 169 [false positive rate (FPR) = 0.262] and 84 (FPR = 0.581) SNP, respectively, were significant (P < 0.001) for AGECL. In Brahman, 41% of these significant markers mapped to a single chromosomal region on BTA14. In Tropical Composites, 16% of these significant markers were located on BTA5. For PPAI, 66 (FPR = 0.67) and 113 (FPR = 0.432) SNP were significant (P < 0.001) in Brahman and Tropical Composite, respectively, whereas for PW, 68 (FPR = 0.64) and 113 (FPR = 0.432) SNP were significant (P < 0.01). In Tropical Composites, the largest concentration of PPAI markers were located on BTA5 [19% (PPAI) and 23% (PW)], and BTA16 [17% (PPAI) and 18% (PW)]. In Brahman cattle, the largest concentration of markers for postpartum anestrus was located on BTA3 (14% for PPAI and PW) and BTA14 (17% PPAI). Very few of the significant markers for female reproduction traits for the Brahman and Tropical Composite breeds were located in the same chromosomal regions. However, fatness and BW traits as well as serum IGF1 concentration were found to be associated with similar genome regions within and between breeds. Clusters of SNP associated with multiple traits were located on BTA14 in Brahman and BTA5 in Tropical Composites.

Regulatory impact factors

MOTIVATION Although transcription factors (TF) play a central regulatory role, their detection from expression data is limited due to their low, and often sparse, expression. In order to fill this gap, we propose a regulatory impact factor (RIF) metric to identify critical TF from gene expression data. RESULTS To substantiate the generality of RIF, we explore a set of experiments spanning a wide range of scenarios including breast cancer survival, fat, gonads and sex differentiation. We show that the strength of RIF lies in its ability to simultaneously integrate three sources of information into a single measure: (i) the change in correlation existing between the TF and the differentially expressed (DE) genes; (ii) the amount of differential expression of DE genes; and (iii) the abundance of DE genes. As a result, RIF analysis assigns an extreme score to those TF that are consistently most differentially co-expressed with the highly abundant and highly DE genes (RIF1), and to those TF with the most altered ability to predict the abundance of DE genes (RIF2). We show that RIF analysis alone recovers well-known experimentally validated TF for the processes studied. The TF identified confirm the importance of PPAR signaling in adipose development and the importance of transduction of estrogen signals in breast cancer survival and sexual differentiation. We argue that RIF has universal applicability, and advocate its use as a promising hypotheses generating tool for the systematic identification of novel TF not yet documented as critical.

Accuracy of prediction of genomic breeding values for residual feed intake and carcass and meat quality traits in Bos taurus, Bos indicus, and composite beef cattle

The aim of this study was to assess the accuracy of genomic predictions for 19 traits including feed efficiency, growth, and carcass and meat quality traits in beef cattle. The 10,181 cattle in our study had real or imputed genotypes for 729,068 SNP although not all cattle were measured for all traits. Animals included Bos taurus, Brahman, composite, and crossbred animals. Genomic EBV (GEBV) were calculated using 2 methods of genomic prediction [BayesR and genomic BLUP (GBLUP)] either using a common training dataset for all breeds or using a training dataset comprising only animals of the same breed. Accuracies of GEBV were assessed using 5-fold cross-validation. The accuracy of genomic prediction varied by trait and by method. Traits with a large number of recorded and genotyped animals and with high heritability gave the greatest accuracy of GEBV. Using GBLUP, the average accuracy was 0.27 across traits and breeds, but the accuracies between breeds and between traits varied widely. When the training population was restricted to animals from the same breed as the validation population, GBLUP accuracies declined by an average of 0.04. The greatest decline in accuracy was found for the 4 composite breeds. The BayesR accuracies were greater by an average of 0.03 than GBLUP accuracies, particularly for traits with known genes of moderate to large effect mutations segregating. The accuracies of 0.43 to 0.48 for IGF-I traits were among the greatest in the study. Although accuracies are low compared with those observed in dairy cattle, genomic selection would still be beneficial for traits that are hard to improve by conventional selection, such as tenderness and residual feed intake. BayesR identified many of the same quantitative trait loci as a genomewide association study but appeared to map them more precisely. All traits appear to be highly polygenic with thousands of SNP independently associated with each trait.

Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets

BackgroundAltered networks of gene regulation underlie many complex conditions, including cancer. Inferring gene regulatory networks from high-throughput microarray expression data is a fundamental but challenging task in computational systems biology and its translation to genomic medicine. Although diverse computational and statistical approaches have been brought to bear on the gene regulatory network inference problem, their relative strengths and disadvantages remain poorly understood, largely because comparative analyses usually consider only small subsets of methods, use only synthetic data, and/or fail to adopt a common measure of inference quality.MethodsWe report a comprehensive comparative evaluation of nine state-of-the art gene regulatory network inference methods encompassing the main algorithmic approaches (mutual information, correlation, partial correlation, random forests, support vector machines) using 38 simulated datasets and empirical serous papillary ovarian adenocarcinoma expression-microarray data. We then apply the best-performing method to infer normal and cancer networks. We assess the druggability of the proteins encoded by our predicted target genes using the CancerResource and PharmGKB webtools and databases.ResultsWe observe large differences in the accuracy with which these methods predict the underlying gene regulatory network depending on features of the data, network size, topology, experiment type, and parameter settings. Applying the best-performing method (the supervised method SIRENE) to the serous papillary ovarian adenocarcinoma dataset, we infer and rank regulatory interactions, some previously reported and others novel. For selected novel interactions we propose testable mechanistic models linking gene regulation to cancer. Using network analysis and visualization, we uncover cross-regulation of angiogenesis-specific genes through three key transcription factors in normal and cancer conditions. Druggabilty analysis of proteins encoded by the 10 highest-confidence target genes, and by 15 genes with differential regulation in normal and cancer conditions, reveals 75% to be potential drug targets.ConclusionsOur study represents a concrete application of gene regulatory network inference to ovarian cancer, demonstrating the complete cycle of computational systems biology research, from genome-scale data analysis via network inference, evaluation of methods, to the generation of novel testable hypotheses, their prioritization for experimental validation, and discovery of potential drug targets.