Prediction and Verification of miRNA Expression in Human and Rat Retinas.

PURPOSE: MicroRNAs (miRNAs) play a global role in regulating gene expression and have important tissue-specific functions. Little is known about their role in the retina. The purpose of this study was to establish the retinal expression of those miRNAs predicted to target genes involved in vision. METHODS: miRNAs potentially targeting important "retinal" genes, as defined by expression pattern and implication in disease, were predicted using a published algorithm (TargetScan; Envisioneering Medical Technologies, St. Louis, MO). The presence of candidate miRNAs in human and rat retinal RNA was assessed by RT-PCR. cDNA levels for each miRNA were determined by quantitative PCR. The ability to discriminate between miRNAs varying by a single nucleotide was assessed. The activity of miR-124 and miR-29 against predicted target sites in Rdh10 and Impdh1 was tested by cotransfection of miRNA mimics and luciferase reporter plasmids. RESULTS: Sixty-seven miRNAs were predicted to target one or more of the 320 retinal genes listed herein. All 11 candidate miRNAs tested were expressed in the retina, including miR-7, miR-124, miR135a, and miR135b. Relative levels of individual miRNAs were similar between rats and humans. The Rdh10 3'UTR, which contains a predicted miR-124 target site, mediated the inhibition of luciferase activity by miR-124 mimics in cell culture. CONCLUSIONS: Many miRNAs likely to regulate genes important for retinal function are present in the retina. Conservation of miRNA retinal expression patterns from rats to humans supports evidence from other tissues that disruption of miRNAs is a likely cause of a range of visual abnormalities.

Applying a QSPR correlation to the prediction of surface tensions of ionic liquids

Ionic liquids (ILs) have attracted large amount of interest due to their unique properties. Although large effort has been focused on the investigation of their potential application, characterization of ILs properties and structure–property relationships of ILs are poorly understood. Computer aided molecular design (CAMD) of ionic liquids (ILs) can only be carried if predictive computational methods for the ILs properties are available. The limited availability of experimental data and their quality have been preventing the development of such tools. Based on experimental surface tension data collected from the literature and measured at our laboratory, it is here shown how a quantitative structure–property relationship (QSPR) correlation for parachors can be used along with an estimation method for the densities to predict the surface tensions of ILs. It is shown that a good agreement with literature data is obtained. For circa 40 ionic liquids studied a mean percent deviation (MPD) of 5.75% with a maximum deviation inferior to 16% was observed. A correlation of the surface tensions with the molecular volumes of the ILs was developed for estimation of the surface tensions at room temperature. It is shown that it can describe the experimental data available within a 4.5% deviation. The correlations here developed can thus be used to evaluate the surface tension of ILs for use in process design or in the CAMD of new ionic liquids.

Prediction- and simulation-error based perceptron training: Solution space analysis and a novel combined training scheme

Previous papers have noted the difficulty in obtaining neural models which are stable under simulation when trained using prediction-error-based methods. Here the differences between series-parallel and parallel identification structures for training neural models are investigated. The effect of the error surface shape on training convergence and simulation performance is analysed using a standard algorithm operating in both training modes. A combined series-parallel/parallel training scheme is proposed, aiming to provide a more effective means of obtaining accurate neural simulation models. Simulation examples show the combined scheme is advantageous in circumstances where the solution space is known or suspected to be complex. (c) 2006 Elsevier B.V. All rights reserved.