A comprehensive simulation study on classification of RNA-Seq data

Abstract

RNA sequencing (RNA-Seq) is a powerful technique for the gene-expression profiling of organisms that uses the capabilities of next-generation sequencing technologies. Developing gene-expression-based classification algorithms is an emerging powerful method for diagnosis, disease classification and monitoring at molecular level, as well as providing potential markers of diseases. Most of the statistical methods proposed for the classification of geneexpression data are either based on a continuous scale (eg. microarray data) or require a normal distribution assumption. Hence, these methods cannot be directly applied to RNASeq data since they violate both data structure and distributional assumptions. However, it is possible to apply these algorithms with appropriate modifications to RNA-Seq data. One way is to develop count-based classifiers, such as Poisson linear discriminant analysis and negative binomial linear discriminant analysis. Another way is to bring the data closer to microarrays and apply microarray-based classifiers. In this study, we compared several classifiers including PLDA with and without power transformation, NBLDA, single SVM, bagging SVM (bagSVM), classification and regression trees (CART), and random forests (RF). We also examined the effect of several parameters such as overdispersion, sample size, number of genes, number of classes, differential-expression rate, and the transformation method on model performances. A comprehensive simulation study is conducted and the results are compared with the results of two miRNA and two mRNA experimental datasets. The results revealed that increasing the sample size, differential-expression rate and decreasing the dispersion parameter and number of groups lead to an increase in classification accuracy. Similar with differential-expression studies, the classification of RNA-Seq data requires careful attention when handling data overdispersion. We conclude that, as a count-based classifier, the power transformed PLDA and, as a microarray-based classifier, vst or rlog transformed RF and SVM classifiers may be a good choice for classification. An R/BIOCONDUCTOR package, MLSeq, is freely available at https://www.bioconductor.org/packages/release/bioc/ html/MLSeq.html.

Rewiring of gene expression in circulating white blood cells is associated with pregnancy outcome in heifers (Bos taurus)

“…We tested this hypothesis by applying machine learning algorithms on two groups of heifers that were bred in 2015 (year one) and 2016 (year two). Parallel random forest emerged as the algorithm with over 90% efficiency of classification nearly all trials executed, which confirms the potential of accurate classification of samples using RNA-seq data under the case-control framework 69,70 . The results show that while not one single gene emerges as a potential biomarker, the accumulated information of transcript abundance from multiple genes can be powerful for the identification of fertility potential in cattle.…”

A Western-Type Dietary Pattern Induces an Atherogenic Gene Expression Profile in the Coronary Arteries of the Ossabaw Pig

“…Pigs from all groups were pooled ( n = 15) and Spearman's correlation coefficients were calculated to determine associations between the expression of differentially expressed genes with concentrations of cardiometabolic risk indicators (LDL cholesterol, HDL cholesterol, and hsCRP) and atherosclerotic lesion severity. A random forest algorithm, produced by the R package MLSeq, was used to identify genes differentially expressed by the WD, with the top variable importance score based on ability to classify the pooled group of pigs by the presence of atherosclerosis in the proximal LAD coronary artery ( 26 ). Raw gene count data underwent a deseq normalization and voom transformation, and a random forest model was run with a k -fold cross-validation (CV = 10) and repeated 1000 times.…”

Parallel bimodal single-cell sequencing of transcriptome and chromatin accessibility

“…The confusion matrixes were generated using MLSeq (Zararsiz et al 2017). Details can be found in the Supplemental Methods, and the detailed script can be found in Supplemental Code.…”