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.     

Microarray phenotyping places cyclase associated protein CAP at the crossroad of signaling pathways reorganizing the actin cytoskeleton in Dictyostelium

Large-scale gene expression analysis has been applied recently to uncover groups of genes that are co-regulated in particular processes. Here we undertake such an analysis on CAP, a protein that participates in the regulation of the actin cytoskeleton and in cAMP signaling in Dictyostelium. microarray analysis revealed that loss of CAP altered the expression of many cytoskeletal components. One of these, the Rho GDP-dissociation inhibitor RhoGDI1, was analyzed further. RhoGDI1 null cells expressed lower amounts of CAP, which failed to accumulate predominantly at the cell cortex. To further position CAP in the corresponding signal transduction pathways we studied CAP localization and cellular functioning in mutants that have defects in several signaling components. CAP showed correct localization and dynamics in all analyzed strains except in mutants with deficient cAMP dependent protein kinase A activity, where CAP preferentially accumulated in crown shaped structures. Ectopic expression of CAP improved the efficiency of phagocytosis in Gβ-deficient cells and restored the pinocytosis, morphology and actin distribution defects in a PI3 kinase double mutant (pi3k1/2 null). Our results show that CAP acts at multiple crossroads and links signaling pathways to the actin cytoskeleton either by physical interaction with cytoskeletal components or through regulation of their gene expression.     

Insights into the taxonomy, genetics and physiology of bifidobacteria

Despite the generally accepted importance of bifidobacteria as probiotic components of the human intestinal microflora and their use in health promoting foods, there is only limited information about their phylogenetic position, physiology and underlying genetics. In the last few years numerous molecular approaches have emerged for the identification and characterization of bifidobacterial strains. Their use, in conjunction with traditional culturing methods, has led to a polyphasic taxonomy which has significantly enhanced our knowledge of the role played by these bacteria in the human intestinal ecosystem. The recent adaptation of culture-independent molecular tools to the fingerprinting of intestinal and food communities offers an exciting opportunity for revealing a more detailed picture of the true complexity of these environments. Furthermore, the availability of bifidobacterial genome sequences has advanced knowledge on the genetics of bifidobacteria and the effects of their metabolic activities on the intestinal ecosystem. The release of a complete Bifidobacterium longum genome sequence and the recent initiative to sequence additional strains are expected to open up a new era of comparative genomics in bifidobacterial biology. Moreover, the use of genomotyping allows a global comparative analysis of gene content between different bifidobacterial isolates of a given species without the necessity of sequencing many strains. Genomotyping provides useful information about the degree of relatedness among various strains of Bifidobacterium species and consequently can be used in a polyphasic identification approach. This review will deal mainly with the molecular tools described for bifidobacterial identification and the first insights into the underlying genetics involved in bifidobacterial physiology as well as genome variability.     

Aberrant epigenetic and genetic marks are seen in myelodysplastic leukocytes and reveal Dock4 as a candidate pathogenic gene on chromosome 7q

Myelodysplastic syndromes (MDS) are characterized by abnormal and dysplastic maturation of all blood lineages. Even though epigenetic alterations have been seen in MDS marrow progenitors, very little is known about the molecular alterations in dysplastic peripheral blood cells. We analyzed the methylome of MDS leukocytes by the HELP assay and determined that it was globally distinct from age-matched controls and was characterized by numerous novel, aberrant hypermethylated marks that were located mainly outside of CpG islands and preferentially affected GTPase regulators and other cancer-related pathways. Additionally, array comparative genomic hybridization revealed that novel as well as previously characterized deletions and amplifications could also be visualized in peripheral blood leukocytes, thus potentially reducing the need for bone marrow samples for future studies. Using integrative analysis, potentially pathogenic genes silenced by genetic deletions and aberrant hypermethylation in different patients were identified. DOCK4, a GTPase regulator located in the commonly deleted 7q31 region, was identified by this unbiased approach. Significant hypermethylation and reduced expression of DOCK4 in MDS bone marrow stem cells was observed in two large independent datasets, providing further validation of our findings. Finally, DOCK4 knockdown in primary marrow CD34(+) stem cells led to decreased erythroid colony formation and increased apoptosis, thus recapitulating the bone marrow failure seen in MDS. These findings reveal widespread novel epigenetic alterations in myelodysplastic leukocytes and implicate DOCK4 as a pathogenic gene located on the 7q chromosomal region.