MICRORNA DISCOVERY AND CHARACTERIZATION IN VITIS VINIFERA USING SMALLRNA DEEP SEQUENCING AND SUPPORT VECTOR MACHINE

MiRNAs are small non coding RNAs that play an important role in the regulation of multiple cell events. They inhibit gene expression at post transcriptional level by binding mRNA targets that are degraded or squestred from translation. Vitis vinifera is the first whole genome sequenced for a commercially important fruit species. Here we present the development and implementation of diverse strategies for the identification and validation of miRNAs of the grapevine. Many putative conserved microRNA precursors were identified by comparative methods and subsequently validated through high throughput smallRNA sequencing and oligonucleotide array technology. Additional bioinformatics tools were implemented for the ab-initio prediction of miRNAs and for the identification of lineage-specific miRNAs from smallRNA deep sequence data. Materials and methods Software to assist in the design of oligonucleotide arrays for the validation of miRNA expression in grape was developed and oligonucleotide array and deep sequencing experiments were used to confirm the expression of conserved mature miRNAs from most of these loci in at least one tissue or developmental stage. Support Vector Machine - based software to predict novel miRNAs and to study their evolution was developed and shown to outperform similar published methods. This classifier was also incorporated into a novel approach to the analysis of smallRNA deep sequence utilizing patterns of mapping of reads on the genome. Our method performs well in the identification novel miRNAs and non-canonical miRNA-like loci. Results Many conserved miRNAs were identified and show strong patterns of tissue specific expression. We have shown that for many, but by no means all known miRNA precursors, evidence for primary transcript expression can be obtained from high throughput transc-riptome analysis, classically performed to follow expression levels of protein coding genes. We estimated patterns of splicing and alternative splicing of known pri-miRNA transcripts The method developed for the identification of plant miRNA precursors from smallRNA NGS data recovers many novel, canonical miRNAs from Vitis and is capable of identifying loci producing miRNA-like smallRNAs with characteristics that are atypical of most conserved miRNAs. The patterns of smallRNA generated from putatively lineage specific loci have been considered in the context of a current model of miRNA gene evolution.