The quantification of transcriptomic features is the basis of the analysis of RNA-seq data. Researchers at the Novartis Institutes for Biomedical Research have developed an integrated alignment workflow and a simple counting-based approach to derive estimates for gene, exon and exon-exon junction expression. In contrast to previous counting-based approaches, EQP takes into account only reads whose alignment pattern agrees with the splicing pattern of the features of interest. This leads to improved gene expression estimates as well as to the generation of exon counts that allow disambiguating reads between overlapping exons. Unlike other methods that quantify skipped introns, EQP offers a novel way to compute junction counts based on the agreement of the read alignments with the exons on both sides of the junction, thus providing a uniformly derived set of counts.
The researchers evaluated the performance of EQP on both simulated and real Illumina RNA-seq data and compared it with other quantification tools. Their results suggest that EQP provides superior gene expression estimates and they illustrate the advantages of EQP’s exon and junction counts. The provision of uniformly derived high-quality counts makes EQP an ideal quantification tool for differential expression and differential splicing studies.
The distribution of RMSD values for different quantification methods and aligners
Each box plot reflects 32 values computed between the Taqman mean gene expression fold changes and the gene count fold changes of sample SEQC-A versus SEQC-B for different lanes of the RNA-seq data. For EQP we consider four parameter settings: counting reads with up to 100 or up to 10 genomic alignments (EQP and EQP-quasi-unique), with unique genomic alignments (EQP-unique), and with unambiguous genomic alignments (EQP-unambiguous).
Availability – EQP is freely available for download at https://github.com/Novartis/EQP-cluster