2009/06/03

miRanalyzer, a new web server tool for smallRNAs deep-sequencing analysis, has been developed by a Functional Genomics group at CIC bioGUNE, led by Ana. M. Aransay, in collaboration with German bioinformaticians.

A paper by Michael Hackenberg et al, presenting miRanalyzer, a microRNA detection and analysis tool, was published in May's Web Server issue of Nucleic Acid Research journal. The project was completed in collaboration with bioinformatics experts from the Institute for Bioinformatics and Systems Biology at the German Research Center for Environmental Health.

While next generation sequencing is revolutionizing genomics, new protocols are being developed for the sequencing of smallRNA molecules and the measurement of its expression levels, the detection of this type of RNA is still a daunting task. Very little smallRNA expression data is currently available. This will undoubtedly change drastically over the next years generating new insights into the role of microRNAs in post-transcriptional regulation. The large amount of data created in deep-sequencing experiments (up to several gigabytes per single experiment) will create a high demand for bioinformatics tools capable of dealing with this kind of results.

Functional Genomics Group at CIC bioGUNE in collaboration with the Institute for Bioinformatics and Systems Biology at the German Research Center for Environmental Health developed miRanalyzer, a web server tool for the analysis of smallRNAs deep-sequencing experiments. The tool requires a simple input file containing a list of unique reads and its copy numbers (expression levels). miRanalyzer detects all known microRNA sequences annotated in miRBase, finds all perfect matches against other libraries of transcribed sequences and predicts new microRNAs.

The prediction of new microRNAs is an especially important task, as there are many species with very few known microRNAs. A highly accurate machine learning algorithm for the prediction of new microRNAs was developed; it reaches AUC values of 97.9% and recall values of up to 75% on unseen data. The web tool summarizes all the steps in a single output page, which provides a comprehensive overview of the analysis, adding links to more detailed output pages for each analysis module.