Motivation: MicroRNAs (miRNAs) are endogenous, small (~ 20 nt), single-stranded, non-coding RNAs (ncRNAs) that result from the nuclear and cytoplasmic processing of transcribed precursor hairpin structures. They are increasingly recognized as playing crucial roles as post-transcriptional antisense regulators of gene expression through regulation of mRNA stability or translational efficiency. miRNAs, first reported in Caenorhabditis elegans, have been identified in the genomes of most higher organisms, including worms, flies, plants, mammals and recently in viruses. Functional studies have shown that miRNAs play important roles in processes such as, cell proliferation, fat metabolism, apoptosis, neuronal cell fate, insulin secretion, haematopoietic differentiation and developmental regulation. The detection of homologs of known miRNAs through comparative genomic approaches has proved relatively tractable. However, the ab-initio prediction of miRNA precursors through computational methods poses several additional difficulties, not least the fact that not all thermodynamically plausible transcribed hairpins are processed to yield mature miRNAs. It has not until now been possible to identify conserved sequence or structural elements that define consensus recognition elements for the enzymes that process miRNA precursors. In the light of these observations we wished to develop and improve methods for the discrimination of true miRNA precursor hairpins from spurious hairpins Methods: We have developed a SVM (Support Vector Machine) that considers up to 74 features associated with the primary and secondary structures and thermodynamic characteristics of candidate hairpin structures. We use a standard heuristic approach to optimize combinations of features used and train the SVM with sets of characterized hairpin miRNA precursors and known non-miRNA hairpins. Results: Our SVM shows highly promising results in the discrimination of true miRNA precursors from “spurious” hairpins (typically around 95% sensitivity) in various species. In particular, our levels of false positive predictions appear to be low relative to comparable methods.
A Support Vector Machine for the Discrimination of MicroRNA Precursors from Other Genomic Hairpin Structures / V. Piccolo, G. Pesole, D. Horner. ((Intervento presentato al convegno Annual Meeting Bioinformatics Italian Society (BITS) tenutosi a Napoli nel 2007.
A Support Vector Machine for the Discrimination of MicroRNA Precursors from Other Genomic Hairpin Structures
V. PiccoloPrimo
;G. PesoleSecondo
;D. HornerUltimo
2007
Abstract
Motivation: MicroRNAs (miRNAs) are endogenous, small (~ 20 nt), single-stranded, non-coding RNAs (ncRNAs) that result from the nuclear and cytoplasmic processing of transcribed precursor hairpin structures. They are increasingly recognized as playing crucial roles as post-transcriptional antisense regulators of gene expression through regulation of mRNA stability or translational efficiency. miRNAs, first reported in Caenorhabditis elegans, have been identified in the genomes of most higher organisms, including worms, flies, plants, mammals and recently in viruses. Functional studies have shown that miRNAs play important roles in processes such as, cell proliferation, fat metabolism, apoptosis, neuronal cell fate, insulin secretion, haematopoietic differentiation and developmental regulation. The detection of homologs of known miRNAs through comparative genomic approaches has proved relatively tractable. However, the ab-initio prediction of miRNA precursors through computational methods poses several additional difficulties, not least the fact that not all thermodynamically plausible transcribed hairpins are processed to yield mature miRNAs. It has not until now been possible to identify conserved sequence or structural elements that define consensus recognition elements for the enzymes that process miRNA precursors. In the light of these observations we wished to develop and improve methods for the discrimination of true miRNA precursor hairpins from spurious hairpins Methods: We have developed a SVM (Support Vector Machine) that considers up to 74 features associated with the primary and secondary structures and thermodynamic characteristics of candidate hairpin structures. We use a standard heuristic approach to optimize combinations of features used and train the SVM with sets of characterized hairpin miRNA precursors and known non-miRNA hairpins. Results: Our SVM shows highly promising results in the discrimination of true miRNA precursors from “spurious” hairpins (typically around 95% sensitivity) in various species. In particular, our levels of false positive predictions appear to be low relative to comparable methods.File | Dimensione | Formato | |
---|---|---|---|
Abstract_BITS2007.doc
accesso aperto
Tipologia:
Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione
28.5 kB
Formato
Microsoft Word
|
28.5 kB | Microsoft Word | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.