Systematical detection of significant genes in microarray data by incorporating gene interaction relationship in biological systems

SEGtool: a specifically expressed gene detection tool and applications in human tissue and single-cell sequencing data

Different tissues and diseases have distinct transcriptional profilings with specifically expressed genes (SEGs). So, the identification of SEGs is an important issue in the studies of gene function, biological development, disease mechanism and biomarker discovery. However, few accurate and easy-to-use tools are available for RNA sequencing (RNA-seq) data to detect SEGs. Here, we presented SEGtool, a tool based on fuzzy c-means, Jaccard index and greedy annealing method for SEG detection automatically and self-adaptively ignoring data distribution. Testing result showed that our SEGtool outperforms the existing tools, which was mainly developed for microarray data. By applying SEGtool to Genotype-Tissue Expression (GTEx) human tissue data set, we detected 3181 SEGs with tissue-related functions. Regulatory networks reveal tissue-specific transcription factors regulating many SEGs, such as ETV2 in testis, HNF4A in liver and NEUROD1 in brain. Applied to a case study of single-cell sequencing (SCS) data from embryo cells, we identified many SEGs in specific stages of human embryogenesis. Notably, SEGtool is suitable for RNA-seq data and even SCS data with high specificity and accuracy. An implementation of SEGtool R package is freely available at http://bioinfo.life.hust.edu.cn/SEGtool/.

Q-GDEMAR: a general method for the identification of differentially expressed genes in microarrays with unbalanced groups

Microarray analysis is a powerful tool to simultaneously determine the pattern of transcription of large amounts of genes. For data post-processing distinct computational methods are currently used that, however, lead to different results regarding the genes expressed differentially. Herein, a new methodology for microarray data analysis named Q-GDEMAR is presented. It combines the quantile characterization of the entire distribution together with the Gaussian deconvolution of the central region of the microarray data distribution. Three discriminant variable variants are proposed that allow us to summarize data and compare groups even when their size is strongly unbalanced. In addition, a simple procedure to compute the false discovery rate (FDR) is also presented. The performance of the method is compared with that observed when using LIMMA (Linear Models Microarray) software as reference. In 58 out of 68 cases, Q-GDEMAR showed a higher sensitivity than LIMMA to detect differentially expressed genes (p = 1 × 10(-10)). The proposed method does not produce biased information, detecting genes with high sensitivity equally well at both tails of the distribution (p = 0.7428). Moreover, all detected genes were associated with very low levels of FDR (median value = 0.67%, interquartile range = 0.87%). Q-GDEMAR can be used as a general method for microarray analysis, but is particularly indicated when the conditions to be compared are unbalanced. The superior performance of Q-GEDEMAR is the consequence of its higher discriminative power and, the fact that it yields a univocal correspondence between the p-values and the values of the discriminating variable. Q-GDEMAR was tested only using Affymetrix microarrays. However, given that it operates after the step of data standardization, it can be used with the same quality features on any of the available mono- or dual-channel microarray platforms.

SpeCond: a method to detect condition-specific gene expression

Transcriptomic studies routinely measure expression levels across numerous conditions. These datasets allow identification of genes that are specifically expressed in a small number of conditions. However, there are currently no statistically robust methods for identifying such genes. Here we present SpeCond, a method to detect condition-specific genes that outperforms alternative approaches. We apply the method to a dataset of 32 human tissues to determine 2,673 specifically expressed genes. An implementation of SpeCond is freely available as a Bioconductor package at http://www.bioconductor.org/packages/release/bioc/html/SpeCond.html.