Computational methodologies for studying non-coding RNAs relevant to central nervous system function and dysfunction

Non-coding RNAs (ncRNAs) are a large and diverse group of transcripts that span the eukaryotic genome, of which less than 2% encodes proteins. Several distinct families of ncRNAs have been described and implicated in many aspects of central nervous system (CNS) function including translation, RNA metabolism, gene regulation, and development. The need to distinguish ncRNAs from sequence data, as well as potentially uncovering novel ncRNA families, has ignited the development of customized computational approaches and bioinformatic resources to handle these tasks. In this review, we provide an overview of the numerous procedures developed to predict ncRNAs based on their primary sequence and predicted secondary structure. These methodologies are broadly grouped into genome scanning algorithms, mixed approaches, and machine learning algorithms. Regulatory ncRNAs, particularly microRNAs (miRNAs), are a major focus of current research efforts and this review will therefore center on the prediction of miRNAs and the putative gene targets they act upon. With the advent of ultra high-throughput sequencing technologies 'deep sequencing' has emerged as the cutting-edge method for ncRNA identification and we will also touch on some computational resources that play a key role in analysis of this type of data.