Integrated Analysis and Annotation of Splicing Variants and Antisense Transcripts Using RSVP and NASTI-seq

The eukaryotic transcriptome consists of many RNA species including thousands of alternative splicing variants and antisense transcripts. High throughput sequencing has been applied to systematically characterize transcriptome at high coverage and base-pair resolution in many organisms. To translate raw sequencing data into biological relevant discoveries, we have recently developed two computational tools: RNAseq Variant Prediction (RSVP) and Natural Anti-Sense Transcripts Identification using RNA-seq (NASTI-seq). RSVP predicts coding and non-coding RNA splicing variants using a Generalized Hidden Markov (HMM) model which incorporates RNA-seq read coverage and spliced read information. NASTI-seq identifies natural antisense transcripts using a statistical model to account for protocol efficiency in strand specific RNA-seq data. Both RSVP and NASTI-seq were evaluated using the model species Arabidopsis and perform better than competing methods in benchmark experiments. These two computational tools were used to analyze over 1.5 billion reads from cell type-specific transcriptome of Arabidopsis roots. We identified ~50,000 splicing variants for coding and non-coding genes and ~2,400 antisense transcripts in our samples. Both computational tools can be easily applied to identify novel splicing variants and antisense transcripts using RNA-seq data from other plant and animal species.