A Comparison of Methods for RNA-Seq Differential Expression Analysis and a New Empirical Bayes Approach

Abstract

Transcriptome-based biosensors are expected to have a large impact on the future of biotechnology. However, a central aspect of transcriptomics is differential expression analysis, where, currently, deep RNA sequencing (RNA-seq) has the potential to replace the microarray as the standard assay for RNA quantification. Our contributions here to RNA-seq differential expression analysis are two-fold. First, given the high cost of an RNA-seq run, biological replicates are rare, and therefore, information sharing across genes to obtain variance estimates is crucial. To handle such information sharing in a rigorous manner, we propose an hierarchical, empirical Bayes approach (R-EBSeq) that combines the Cufflinks model for generating relative transcript abundance measurements, known as FPKM (fragments per kilobase of transcript length per million mapped reads) with the EBArrays framework, which was previously developed for empirical Bayes analysis of microarray data. A desirable feature of R-EBSeq is easy-to-implement analysis of more than pairwise comparisons, as we illustrate with experimental data. Secondly, we develop the standard RNA-seq test data set, on the level of reads, where 79 transcripts are artificially differentially expressed and, therefore, explicitly known. This test data set allows us to compare the performance, in terms of the true discovery rate, of R-EBSeq to three other widely used RNAseq data analysis packages: Cuffdiff, DEseq and BaySeq. Our analysis indicates that DESeq identifies the first half of the differentially expressed transcripts well, but then is outperformed by Cuffdiff and R-EBSeq. Cuffdiff and R-EBSeq are the two top performers. Thus, R-EBSeq offers good performance, while allowing flexible and rigorous comparison of multiple biological conditions.

Comparative transcriptome expression analysis in susceptible and resistant potato (Solanum tuberosum) cultivars to common scab (Streptomyces scabies) revealed immune priming responses in the incompatible interaction

“…By using RNA transcriptomic analysis of two potato cultivars grown under field conditions, shared transcript information across gene sets from three biological replications was produced to ensure direct comparisons. Accurate and precise transcript variance estimates as well as multiple pairwise comparisons were inferred for generating a robust dataset [ 52 ]. Our data showed clear and consistent DEG patterns that distinguished HB from GM.…”

Transcriptome analysis of the epidermis of the purple quail-like (q-lp) mutant of silkworm, Bombyx mori

“…The sequences were aligned to the silkworm genome database SilkDB ( http://silkworm.swu.edu.cn/silkdb/ ); after a second quality analysis for alignment, analysis of the distribution and coverage of the clean reads on the reference sequence was conducted. RPKM (Reads Per Kb per Million reads) [ 24 ] was used to calculate the expression level of genes, with RPKM = mapped reads of gene/(the total mapped reads of all genes*the length of this gene)*10^9. The RPKM of a gene ranged up to 5, and difference in expression was considered at P < 0.05; the fold change of q-l p and 932VR RPKMs was greater than 2.…”

RNA sequencing in post‐mortem human brains of neuropsychiatric disorders

“…18 Comparison of methods for differential expression analysis has been conducted recently to address this issue. [55][56][57][58] On average, algorithms based on negative binomial modeling, such as DESeq2 59,60 and edgeR 61 have improved sensitivities and better control of false discovery rate. 58 Limma voom function 62 performs relatively well and runs faster than DESeq2 and edgeR.…”

“…Using two or more approaches may help reduce the bias induced by one single algorithm . Comparison of methods for differential expression analysis has been conducted recently to address this issue . On average, algorithms based on negative binomial modeling, such as DESeq2 and edgeR have improved sensitivities and better control of false discovery rate .…”