@article{Mendes:Heyne:Freitas:Navig_the_unexp:2012,
author = {Mendes, Nuno D. and Heyne, Steffen and Freitas, Ana T. and 
          Sagot, Marie-France and Backofen, Rolf},
title = {Navigating the unexplored seascape of pre-{miRNA} 
         candidates in single-genome approaches},
journal = {Bioinformatics},
year = {2012},
doi = {10.1093/bioinformatics/bts574},
volume = {},
user = {backofen},
pmid = {23052038},
pages = {},
number = {},
issn = {1367-4811},
abstract = {MOTIVATION: The computational search for novel miRNA 
            precursors often involves some sort of structural analysis 
            with the aim of identifying which type of structures are 
            prone to being recognised and processed by the cellular 
            miRNA-maturation machinery. A natural way to tackle this 
            problem is to perform clustering over the candidate 
            structures along with known miRNA precursor structures. 
            Mixed clusters allows then the identification of candidates 
            that are similar to known precursors. Given the large number 
            of pre-miRNA candidates that can be identified in 
            single-genome approaches, even after applying several 
            filters for precursor robustness and stability, a 
            conventional structural clustering approach is unfeasible. 
            RESULTS: We propose a method to represent candidate 
            structures in a feature space which summarises key 
            sequence/structure characteristics of each candidate. We 
            demonstrate that proximity in this feature space is related 
            to sequence/structure similarity, and we select candidates 
            which have a high similarity to known precursors. Additional 
            filtering steps are then applied to further reduce the 
            number of candidates to those with greater transcriptional 
            potential. Our method is compared to another single-genome 
            method (TripletSVM) in two datasets, showing better 
            performance in one and comparable performance in the other, 
            for larger training sets. Additionally, we show that our 
            approach allows for a better interpretation of the 
            results.Availability and Implementation: The MinDist method 
            is implemented using Perl scripts and is freely available at 
            http://www.cravela.org/?mindist=1. CONTACT: 
            backofen@uni-freiburg.de.}
}