A feature selection method for classification within functional genomics experiments based on the proportional overlapping score

Background

Microarray technology, as well as other functional genomics experiments, allow simultaneous measurements of thousands of genes within each sample. Both the prediction accuracy and interpretability of a classifier could be enhanced by performing the classification based only on selected discriminative genes. We propose a statistical method for selecting genes based on overlapping analysis of expression data across classes. This method results in a novel measure, called proportional overlapping score (POS), of a feature’s relevance to a classification task.

Results

We apply POS, along‐with four widely used gene selection methods, to several benchmark gene expression datasets. The experimental results of classification error rates computed using the Random Forest, k Nearest Neighbor and Support Vector Machine classifiers show that POS achieves a better performance.

Conclusions

A novel gene selection method, POS, is proposed. POS analyzes the expressions overlap across classes taking into account the proportions of overlapping samples. It robustly defines a mask for each gene that allows it to minimize the effect of expression outliers. The constructed masks along‐with a novel gene score are exploited to produce the selected subset of genes.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-274) contains supplementary material, which is available to authorized users.     

CPORT: a consensus interface predictor and its performance in prediction-driven docking with HADDOCK

Background

Macromolecular complexes are the molecular machines of the cell. Knowledge at the atomic level is essential to understand and influence their function. However, their number is huge and a significant fraction is extremely difficult to study using classical structural methods such as NMR and X-ray crystallography. Therefore, the importance of large-scale computational approaches in structural biology is evident. This study combines two of these computational approaches, interface prediction and docking, to obtain atomic-level structures of protein-protein complexes, starting from their unbound components.

Methodology/Principal Findings

Here we combine six interface prediction web servers into a consensus method called CPORT (Consensus Prediction Of interface Residues in Transient complexes). We show that CPORT gives more stable and reliable predictions than each of the individual predictors on its own. A protocol was developed to integrate CPORT predictions into our data-driven docking program HADDOCK. For cases where experimental information is limited, this prediction-driven docking protocol presents an alternative to ab initio docking, the docking of complexes without the use of any information. Prediction-driven docking was performed on a large and diverse set of protein-protein complexes in a blind manner. Our results indicate that the performance of the HADDOCK-CPORT combination is competitive with ZDOCK-ZRANK, a state-of-the-art ab initio docking/scoring combination. Finally, the original interface predictions could be further improved by interface post-prediction (contact analysis of the docking solutions).

Conclusions/Significance

The current study shows that blind, prediction-driven docking using CPORT and HADDOCK is competitive with ab initio docking methods. This is encouraging since prediction-driven docking represents the absolute bottom line for data-driven docking: any additional biological knowledge will greatly improve the results obtained by prediction-driven docking alone. Finally, the fact that original interface predictions could be further improved by interface post-prediction suggests that prediction-driven docking has not yet been pushed to the limit. A web server for CPORT is freely available at http://haddock.chem.uu.nl/services/CPORT.     

PSPP: A Protein Structure Prediction Pipeline for Computing Clusters

Background

Protein structures are critical for understanding the mechanisms of biological systems and, subsequently, for drug and vaccine design. Unfortunately, protein sequence data exceed structural data by a factor of more than 200 to 1. This gap can be partially filled by using computational protein structure prediction. While structure prediction Web servers are a notable option, they often restrict the number of sequence queries and/or provide a limited set of prediction methodologies. Therefore, we present a standalone protein structure prediction software package suitable for high-throughput structural genomic applications that performs all three classes of prediction methodologies: comparative modeling, fold recognition, and ab initio. This software can be deployed on a user''s own high-performance computing cluster.

Methodology/Principal Findings

The pipeline consists of a Perl core that integrates more than 20 individual software packages and databases, most of which are freely available from other research laboratories. The query protein sequences are first divided into domains either by domain boundary recognition or Bayesian statistics. The structures of the individual domains are then predicted using template-based modeling or ab initio modeling. The predicted models are scored with a statistical potential and an all-atom force field. The top-scoring ab initio models are annotated by structural comparison against the Structural Classification of Proteins (SCOP) fold database. Furthermore, secondary structure, solvent accessibility, transmembrane helices, and structural disorder are predicted. The results are generated in text, tab-delimited, and hypertext markup language (HTML) formats. So far, the pipeline has been used to study viral and bacterial proteomes.

Conclusions

The standalone pipeline that we introduce here, unlike protein structure prediction Web servers, allows users to devote their own computing assets to process a potentially unlimited number of queries as well as perform resource-intensive ab initio structure prediction.     

The evolution of mitochondrial genomes in modern frogs (Neobatrachia): nonadaptive evolution of mitochondrial genome reorganization

Background

Although mitochondrial (mt) gene order is highly conserved among vertebrates, widespread gene rearrangements occur in anurans, especially in neobatrachians. Protein coding genes in the mitogenome experience adaptive or purifying selection, yet the role that selection plays on genomic reorganization remains unclear. We sequence the mitogenomes of three species of Glandirana and hot spots of gene rearrangements of 20 frog species to investigate the diversity of mitogenomic reorganization in the Neobatrachia. By combing these data with other mitogenomes in GenBank, we evaluate if selective pressures or functional constraints act on mitogenomic reorganization in the Neobatrachia. We also look for correlations between tRNA positions and codon usage.

Results

Gene organization in Glandirana was typical of neobatrachian mitogenomes except for the presence of pseudogene trnS (AGY). Surveyed ranids largely exhibited gene arrangements typical of neobatrachian mtDNA although some gene rearrangements occurred. The correlation between codon usage and tRNA positions in neobatrachians was weak, and did not increase after identifying recurrent rearrangements as revealed by basal neobatrachians. Codon usage and tRNA positions were not significantly correlated when considering tRNA gene duplications or losses. Change in number of tRNA gene copies, which was driven by genomic reorganization, did not influence codon usage bias. Nucleotide substitution rates and d_N/d_S ratios were higher in neobatrachian mitogenomes than in archaeobatrachians, but the rates of mitogenomic reorganization and mt nucleotide diversity were not significantly correlated.

Conclusions

No evidence suggests that adaptive selection drove the reorganization of neobatrachian mitogenomes. In contrast, protein-coding genes that function in metabolism showed evidence for purifying selection, and some functional constraints appear to act on the organization of rRNA and tRNA genes. As important nonadaptive forces, genetic drift and mutation pressure may drive the fixation and evolution of mitogenomic reorganizations.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-691) contains supplementary material, which is available to authorized users.     

Inferring the perturbed microRNA regulatory networks from gene expression data using a network propagation based method

Background

MicroRNAs (miRNAs) are a class of endogenous small regulatory RNAs. Identifications of the dys-regulated or perturbed miRNAs and their key target genes are important for understanding the regulatory networks associated with the studied cellular processes. Several computational methods have been developed to infer the perturbed miRNA regulatory networks by integrating genome-wide gene expression data and sequence-based miRNA-target predictions. However, most of them only use the expression information of the miRNA direct targets, rarely considering the secondary effects of miRNA perturbation on the global gene regulatory networks.

Results

We proposed a network propagation based method to infer the perturbed miRNAs and their key target genes by integrating gene expressions and global gene regulatory network information. The method used random walk with restart in gene regulatory networks to model the network effects of the miRNA perturbation. Then, it evaluated the significance of the correlation between the network effects of the miRNA perturbation and the gene differential expression levels with a forward searching strategy. Results show that our method outperformed several compared methods in rediscovering the experimentally perturbed miRNAs in cancer cell lines. Then, we applied it on a gene expression dataset of colorectal cancer clinical patient samples and inferred the perturbed miRNA regulatory networks of colorectal cancer, including several known oncogenic or tumor-suppressive miRNAs, such as miR-17, miR-26 and miR-145.

Conclusions

Our network propagation based method takes advantage of the network effect of the miRNA perturbation on its target genes. It is a useful approach to infer the perturbed miRNAs and their key target genes associated with the studied biological processes using gene expression data.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-255) contains supplementary material, which is available to authorized users.     

Bimodal signatures of germline methylation are linked with gene expression plasticity in the coral Acropora millepora

Background

In invertebrates, genes belonging to dynamically regulated functional categories appear to be less methylated than “housekeeping” genes, suggesting that DNA methylation may modulate gene expression plasticity. To date, however, experimental evidence to support this hypothesis across different natural habitats has been lacking.

Results

Gene expression profiles were generated from 30 pairs of genetically identical fragments of coral Acropora millepora reciprocally transplanted between distinct natural habitats for 3 months. Gene expression was analyzed in the context of normalized CpG content, a well-established signature of historical germline DNA methylation. Genes with weak methylation signatures were more likely to demonstrate differential expression based on both transplant environment and population of origin than genes with strong methylation signatures. Moreover, the magnitude of expression differences due to environment and population were greater for genes with weak methylation signatures.

Conclusions

Our results support a connection between differential germline methylation and gene expression flexibility across environments and populations. Studies of phylogenetically basal invertebrates such as corals will further elucidate the fundamental functional aspects of gene body methylation in Metazoa.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-1109) contains supplementary material, which is available to authorized users.     

Map of open and closed chromatin domains in Drosophila genome

Background

Chromatin compactness has been considered a major determinant of gene activity and has been associated with specific chromatin modifications in studies on a few individual genetic loci. At the same time, genome-wide patterns of open and closed chromatin have been understudied, and are at present largely predicted from chromatin modification and gene expression data. However the universal applicability of such predictions is not self-evident, and requires experimental verification.

Results

We developed and implemented a high-throughput analysis for general chromatin sensitivity to DNase I which provides a comprehensive epigenomic assessment in a single assay. Contiguous domains of open and closed chromatin were identified by computational analysis of the data, and correlated to other genome annotations including predicted chromatin “states”, individual chromatin modifications, nuclear lamina interactions, and gene expression. While showing that the widely trusted predictions of chromatin structure are correct in the majority of cases, we detected diverse “exceptions” from the conventional rules. We found a profound paucity of chromatin modifications in a major fraction of closed chromatin, and identified a number of loci where chromatin configuration is opposite to that expected from modification and gene expression patterns. Further, we observed that chromatin of large introns tends to be closed even when the genes are expressed, and that a significant proportion of active genes including their promoters are located in closed chromatin.

Conclusions

These findings reveal limitations of the existing predictive models, indicate novel mechanisms of epigenetic regulation, and provide important insights into genome organization and function.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-988) contains supplementary material, which is available to authorized users.     

Relocation of genes generates non-conserved chromosomal segments in Fusarium graminearum that show distinct and co-regulated gene expression patterns

Background

Genome comparisons between closely related species often show non-conserved regions across chromosomes. Some of them are located in specific regions of chromosomes and some are even confined to one or more entire chromosomes. The origin and biological relevance of these non-conserved regions are still largely unknown. Here we used the genome of Fusarium graminearum to elucidate the significance of non-conserved regions.

Results

The genome of F. graminearum harbours thirteen non-conserved regions dispersed over all of the four chromosomes. Using RNA-Seq data from the mycelium of F. graminearum, we found weakly expressed regions on all of the four chromosomes that exactly matched with non-conserved regions. Comparison of gene expression between two different developmental stages (conidia and mycelium) showed that the expression of genes in conserved regions is stable, while gene expression in non-conserved regions is much more influenced by developmental stage. In addition, genes involved in the production of secondary metabolites and secreted proteins are enriched in non-conserved regions, suggesting that these regions could also be important for adaptations to new environments, including adaptation to new hosts. Finally, we found evidence that non-conserved regions are generated by sequestration of genes from multiple locations. Gene relocations may lead to clustering of genes with similar expression patterns or similar biological functions, which was clearly exemplified by the PKS2 gene cluster.

Conclusions

Our results showed that chromosomes can be functionally divided into conserved and non-conserved regions, and both could have specific and distinct roles in genome evolution and regulation of gene expression.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-191) contains supplementary material, which is available to authorized users.     

seqCNA: an R package for DNA copy number analysis in cancer using high-throughput sequencing

Background

Deviations in the amount of genomic content that arise during tumorigenesis, called copy number alterations, are structural rearrangements that can critically affect gene expression patterns. Additionally, copy number alteration profiles allow insight into cancer discrimination, progression and complexity. On data obtained from high-throughput sequencing, improving quality through GC bias correction and keeping false positives to a minimum help build reliable copy number alteration profiles.

Results

We introduce seqCNA, a parallelized R package for an integral copy number analysis of high-throughput sequencing cancer data. The package includes novel methodology on (i) filtering, reducing false positives, and (ii) GC content correction, improving copy number profile quality, especially under great read coverage and high correlation between GC content and copy number. Adequate analysis steps are automatically chosen based on availability of paired-end mapping, matched normal samples and genome annotation.

Conclusions

seqCNA, available through Bioconductor, provides accurate copy number predictions in tumoural data, thanks to the extensive filtering and better GC bias correction, while providing an integrated and parallelized workflow.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-178) contains supplementary material, which is available to authorized users.     

PhylDiag: identifying complex synteny blocks that include tandem duplications using phylogenetic gene trees

Background

Extant genomes share regions where genes have the same order and orientation, which are thought to arise from the conservation of an ancestral order of genes during evolution. Such regions of so-called conserved synteny, or synteny blocks, must be precisely identified and quantified, as a prerequisite to better understand the evolutionary history of genomes.

Results

Here we describe PhylDiag, a software that identifies statistically significant synteny blocks in pairwise comparisons of eukaryote genomes. Compared to previous methods, PhylDiag uses gene trees to define gene homologies, thus allowing gene deletions to be considered as events that may break the synteny. PhylDiag also accounts for gene orientations, blocks of tandem duplicates and lineage specific de novo gene births. Starting from two genomes and the corresponding gene trees, PhylDiag returns synteny blocks with gaps less than or equal to the maximum gap parameter gap_max. This parameter is theoretically estimated, and together with a utility to graphically display results, contributes to making PhylDiag a user friendly method. In addition, putative synteny blocks are subject to a statistical validation to verify that they are unlikely to be due to a random combination of genes.

Conclusions

We benchmark several known metrics to measure 2D-distances in a matrix of homologies and we compare PhylDiag to i-ADHoRe 3.0 on real and simulated data. We show that PhylDiag correctly identifies small synteny blocks even with insertions, deletions, incorrect annotations or micro-inversions. Finally, PhylDiag allowed us to identify the most relevant distance metric for 2D-distance calculation between homologies.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-268) contains supplementary material, which is available to authorized users.     

SPoRE: a mathematical model to predict double strand breaks and axis protein sites in meiosis

Background

Meiotic recombination between homologous chromosomes provides natural combinations of genetic variations and is a main driving force of evolution. It is initiated via programmed DNA double-strand breaks (DSB) and involves a specific axial chromosomal structure. So far, recombination regions have been mainly determined by experiments, both expensive and time-consuming.

Results

SPoRE is a mathematical model that describes the non-uniform localisation of DSB and axis proteins sites, and distinguishes high versus low protein density. It is based on a combination of genomic signals, based on what is known from wet-lab experiments, whose contribution is precisely quantified. It models axis proteins accumulation at gene 5’-ends with a discrete approximation of their diffusion and convection along genes. It models DSB accumulation at approximated gene promoter positions with intergenic region length and GC-content. SPoRE can be used for prediction and it is parameterised in an obvious way that makes it easy to understand from a biological viewpoint.

Conclusions

When compared to Saccharomyces cerevisiae experimental data, SPoRE predicts axis protein and DSB positions with high sensitivity and precision, axis protein density with an average local correlation r=0.63 and DSB density with an average local correlation r=0.62. SPoRE outbreaks previous DSB predictors, which are based on nucleotide patterning, and it reaches 85% of success rate in DSB prediction compared to 54% obtained by available tools on a benchmarked dataset.SPoRE is available at the address http://www.lcqb.upmc.fr/SPoRE/.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-014-0391-1) contains supplementary material, which is available to authorized users.     

Epitopemap: a web application for integrated whole proteome epitope prediction

Background

Predictions of MHC binding affinity are commonly used in immunoinformatics for T cell epitope prediction. There are multiple available methods, some of which provide web access. However there is currently no convenient way to access the results from multiple methods at the same time or to execute predictions for an entire proteome at once.

Results

We designed a web application that allows integration of multiple epitope prediction methods for any number of proteins in a genome. The tool is a front-end for various freely available methods. Features include visualisation of results from multiple predictors within proteins in one plot, genome-wide analysis and estimates of epitope conservation.

Conclusions

We present a self contained web application, Epitopemap, for calculating and viewing epitope predictions with multiple methods. The tool is easy to use and will assist in computational screening of viral or bacterial genomes.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0659-0) contains supplementary material, which is available to authorized users.     

Gene rearrangements in gekkonid mitochondrial genomes with shuffling,loss, and reassignment of tRNA genes

Background

Vertebrate mitochondrial genomes (mitogenomes) are 16–18 kbp double-stranded circular DNAs that encode a set of 37 genes. The arrangement of these genes and the major noncoding region is relatively conserved through evolution although gene rearrangements have been described for diverse lineages. The tandem duplication-random loss model has been invoked to explain the mechanisms of most mitochondrial gene rearrangements. Previously reported mitogenomic sequences for geckos rarely included gene rearrangements, which we explore in the present study.

Results

We determined seven new mitogenomic sequences from Gekkonidae using a high-throughput sequencing method. The Tropiocolotes tripolitanus mitogenome involves a tandem duplication of the gene block: tRNA^Arg, NADH dehydrogenase subunit 4L, and NADH dehydrogenase subunit 4. One of the duplicate copies for each protein-coding gene may be pseudogenized. A duplicate copy of the tRNA^Arg gene appears to have been converted to a tRNA^Gln gene by a C to T base substitution at the second anticodon position, although this gene may not be fully functional in protein synthesis. The Stenodactylus petrii mitogenome includes several tandem duplications of tRNA^Leu genes, as well as a translocation of the tRNA^Ala gene and a putative origin of light-strand replication within a tRNA gene cluster. Finally, the Uroplatus fimbriatus and U. ebenaui mitogenomes feature the apparent loss of the tRNA^Glu gene from its original position. Uroplatus fimbriatus appears to retain a translocated tRNA^Glu gene adjacent to the 5’ end of the major noncoding region.

Conclusions

The present study describes several new mitochondrial gene rearrangements from Gekkonidae. The loss and reassignment of tRNA genes is not very common in vertebrate mitogenomes and our findings raise new questions as to how missing tRNAs are supplied and if the reassigned tRNA gene is fully functional. These new examples of mitochondrial gene rearrangements in geckos should broaden our understanding of the evolution of mitochondrial gene arrangements.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-930) contains supplementary material, which is available to authorized users.