1. Introduction¶
MicroRNAs (miRNAs) have been characterized as important regulators in almost all animals and plants, as well as in unicellular eukaryotes [13]. Since their discovery in 1993 [11] and subsequently their recognition as a biological entity later from 2000 [17], microRNAs were identified as key regulators on the temporal control of heterochronic genes on the nematode Caenorhabditis elegans as well as in another metazoan species, too [15]. This recognition was complemented by a complete characterization of their typical features as: a stem-loop structure, well-conservation over multiple metazoan clades and typical expression patterns as isolated locus or co-expressed as polycistronic miRNA transcripts [8, 9, 10, 17].
Currently, metazoan miRNAs have been recognized as a conserved group of short ncRNAs, typically ~ 22 nt, with important roles as post-transcriptional regulators of the gene expression affecting a sizeable number of mRNAs and expressed in all developmental process and diseases [2]. Their canonical biogenesis starts in the form of primary precursors (pri-miRNAs) transcribed from long non-coding RNAs or protein-coding transcripts, mostly from introns [22]. Later, derived from hairpin-like precursors excised in the nucleus (pre-miRNAs), their acting form is subsequently further processed as miRNA/miRNA* duplexes on the cytoplasm and incorporated into the RISC complex. Target specificity is achieved by complementarity between the miR and mRNA sequence. (see more details in [2]).
Current classification of annotated miRNAs into families are available in miRBase 1 and mirGeneDB 2 databases. As an example, the human genome reported 1984 miRNA precursors in the miRBase v.22.1 [7] and the corresponding mature products were estimated ~2300 [1]. Focusing on the confidently canonical miRNAs reported in [5], the number of miRNAs is 519. Those differences are explained on the basis of multiple miRNA detection methods as well as the intrinsic definitions to define a canonical miRNA.
Despite the small size of the precursors (80-100 nt), sequence comparison methods are able to detected them, due a high level of sequence conservation [16]. In one hand, it is important to point that the use of blast-based homology searches alone tend to produce false positives that require extensive curation, which relies on properties obtained from miRNAs [18]. On the other hand, the inclusion of the consensus structure complements the homology search methods, for example using covariance models (CMs) [4, 6]. The accuracy and sensitivity of CMs depends critically on the sequence alignment and the annotated consensus used to build the model. Those observations call for an integrated workflow to perform homology search and to evaluate their results consistently.
In this computational approach, focused on the canonical miRNAs processed by Drosha and Dicer,
we improve on ideas from MIRfix [21] and integrate it with homology search.
miRNAture is used to identify and annotate homologs of metazoan microRNAs in a homology-based
setting.
Footnotes