Evolutionary Modeling and Prediction of Non-Coding RNAs in Drosophila

Abstract

We performed benchmarks of phylogenetic grammar-based ncRNA gene prediction, experimenting with eight different models of structural evolution and two different programs for genome alignment. We evaluated our models using alignments of twelve Drosophila genomes. We find that ncRNA prediction performance can vary greatly between different gene predictors and subfamilies of ncRNA gene. Our estimates for false positive rates are based on simulations which preserve local islands of conservation; using these simulations, we predict a higher rate of false positives than previous computational ncRNA screens have reported. Using one of the tested prediction grammars, we provide an updated set of ncRNA predictions for D. melanogaster and compare them to previously-published predictions and experimental data. Many of our predictions show correlations with protein-coding genes. We found significant depletion of intergenic predictions near the 3′ end of coding regions and furthermore depletion of predictions in the first intron of protein-coding genes. Some of our predictions are colocated with larger putative unannotated genes: for example, 17 of our predictions showing homology to the RFAM family snoR28 appear in a tandem array on the X chromosome; the 4.5 Kbp spanned by the predicted tandem array is contained within a FlyBase-annotated cDNA.

Structure-based whole-genome realignment reveals many novel noncoding RNAs

“…This increase may be even more significant in light of recent suggestions that previous screens underestimated their FDRs. For example, Gruber et al (2010) re-estimate the FDR of an RNAz 1.0 screen in ENCODE from the original estimation of 50% ) to 82% ;Bradley et al (2009) report similar observations in Drosophila.…”

“…Since both predictors capture different sets of ncRNAs (Rose et al 2007;Stark et al 2007;Washietl et al 2007;Bradley et al 2009), it is attractive to integrate EvoFold (or related grammar-based predictors from Bradley et al 2009) into the REAPR pipeline in future work. Due to this modularity, REAPR can directly profit from future advances in predicting structural ncRNAs from fixed alignments.…”

“…We apply REAPR to predict ncRNAs in 12 Drosophila genomes, utilizing the same WGA as previous screens (Rose et al 2007;Bradley et al 2009). REAPR with D = 20 produces 70% more highconfidence predictions than directly predicting from the original WGA (Fig.…”

“…In Figure 3, A through D, we considered predictions that contain any combination of fly genomes, whereas previous screens restricted attention to loci containing D. melanogaster (Rose et al 2007;Stark et al 2007;Bradley et al 2009). For comparison, we also restricted attention to predictions containing D. melanogaster.…”

“…By screening multiple whole-genome alignments (WGAs), the more recent approaches of RNAz (Washietl et al 2005b), which efficiently detects structural stability and conservation by energy minimization and support vector machines, and the phylo-SCFG-based EvoFold (Pedersen et al 2006) have significantly increased prediction accuracy. Such methods have been applied to screen WGAs of mammals (Washietl et al 2005a;Pedersen et al 2006) and Drosophila (applying RNAz [Rose et al 2007], EvoFold [Stark et al 2007], and EvoFold-like grammar-based approaches [Bradley et al 2009]). RNAz and EvoFold have further been run on the ENCODE pilot region of the human genome by Washietl et al (2007).…”