miRNEST 2.0, an integrative microRNA resource

miRNEST 2.0, an integrative microRNA resource :: miRNA gene structures

We developed a pipeline for miRNA discovery in a genomic scale using small RNA libraries. The algorithm performs multiple filtering steps to obtain high-quality candidates; in particular, much attention is paid to the profile of reads mapped to the hairpin. Using this approach, we predicted hundreds of novel miRNAs and confirmed a large number of known miRNAs in 21 plant and animal species. Altogether there are 36 468 miRNA predictions (often one miRNA predicted from multiple small RNA libraries).

We modified the abovementioned pipeline to search for mirtrons, i.e. miRNAs with their pre-miRNA sequence spanning the entire intron. We identified mirtrons in five animal species. Some of them were already known (miRBase), while other represent high-quality mirtron candidates.

We analysed degradomes from ten plant species using PAREsnip (Folkes et al., 2012) to identify experimentally supported miRNA targets. Altogether, we found 2,041 miRNA-target associations.

We used HuntMi, a new machine learning miRNA classifiaction tool, to analyze all hairpins stored in miRNEST. Each sequence was assigned "-1" (not a miRNA) or "1" (true miRNA).

We added miRNA gene structure predictions to miRNEST. Here, ERISdb predictions were used (five species) and complemented with predictions for five more plant species: Brachypodium distachyon, Malus domestica, Medicago truncatula, Populus trichocarpa, and Solanum lycopersicum.

We took care to integrate the abovementioned data with miRNAs already stored in miRNEST 1.0. We also updated the user interface to make it more intuitive.