MiRFinder: an improved approach and software implementation for genome-wide fast microRNA precursor scans

Abstract

Background: MicroRNAs (miRNAs) are recognized as one of the most important families of non-coding RNAs that serve as important sequence-specific post-transcriptional regulators of gene expression. Identification of miRNAs is an important requirement for understanding the mechanisms of post-transcriptional regulation. Hundreds of miRNAs have been identified by direct cloning and computational approaches in several species. However, there are still many miRNAs that remain to be identified due to lack of either sequence features or robust algorithms to efficiently identify them.

Results: We have evaluated features valuable for pre-miRNA prediction, such as the local secondary structure differences of the stem region of miRNA and non-miRNA hairpins. We have also established correlations between different types of mutations and the secondary structures of pre-miRNAs. Utilizing these features and combining some improvements of the current pre-miRNA prediction methods, we implemented a computational learning method SVM (support vector machine) to build a high throughput and good performance computational pre-miRNA prediction tool called MiRFinder. The tool was designed for genome-wise, pair-wise sequences from two related species. The method built into the tool consisted of two major steps: 1) genome wide search for hairpin candidates and 2) exclusion of the non-robust structures based on analysis of 18 parameters by the SVM method. Results from applying the tool for chicken/human and D. melanogaster/D. pseudoobscura pair-wise genome alignments showed that the tool can be used for genome wide pre-miRNA predictions.

Conclusion: The MiRFinder can be a good alternative to current miRNA discovery software. This tool is available at http://www.bioinformatics.org/mirfinder/.

Figure 1

Mutation profile of miRNA. There are three types of mutations that cause slight disturbance of the secondary structure of pre-miRNA: (1) mutations in the loop region; (2) mutations between A and G, U and C in the stem region; (3) mutations in the interrelated region of both arms that do not break the base-pairing. The three types of mutations are marked by the numbers 1, 2 and 3, respectively, under the alignments. The conserved nucleotides are marked as "*".

Figure 2

An example of how a hairpin is represented using the new syntax. Symbols of "=", ":", ".", "-"and "^" indicate states of paired, unpaired, insertion, deletion and bulge, respectively. The frequency of each element (combinations of every two adjacent symbols) of the pseudo code of the structure will be used as input vectors for SVM.

Figure 3

The pipeline of the miRFinder. It consists of 5 steps: (1) Smith-Waterman like algorithm searches the genome of short hairpins; (2) The sequence is folded by RNALfold (Hofacker et al., 2004) to get the local structure; (3) the extended good loops is picked out by schLoop; (4) the good loops are re-folded by RNAfold (Zuker & Stiegler, 1981) to get the MFE and secondary structure; (5) the Punish program calculates the punish score of each paired sequence segments; (6) the sequence is then predicted to be miRNAs or non-miRNA hairpins using the SVM (support vector machine) classifier.

Figure 4

The ROC-curve. The solid line shows the ROC-curve for the miRFinder that was trained on miRNAs versus non-miRNA hairpins. The points for Sewer, ProMir, Triplet-SVM, miRNA-SVM, miPred, and RNAmicro are the sensitivities and specificities reported by Sewer et al. (2005), Nam et al. (2005), Xue et al. (2005), Hertel et al. (2006), Helvik et al. (2007) and Kwang Loong et al. (2007). The sensitivity and specificity of miRScan are 0.5 and 0.7, respectively, and are not included in the figure. Panel (A) is a detailed excerpt of Panel (B).