Tailored machine learning models for functional RNA detection in genome-wide screens

<p>The prediction of noncoding RNA and protein coding genetic loci has received<br> &nbsp; considerable attention in comparative genomics aiming in particular<br> &nbsp; at the identification of properties of nucleotide sequences that are<br> &nbsp; informative of their biological role in the cell. We present here a<br> &nbsp; software framework for the alignment-based training, evaluation and<br> &nbsp; application of machine learning models with user-defined<br> &nbsp; parameters. Instead of focusing on the one-size-fits-all approach of<br> &nbsp; pervasive \is annotation pipelines, we offer a framework for the<br> &nbsp; structured generation and evaluation of models based on arbitrary<br> &nbsp; features and input data, focusing on stable and explainable results.<br> &nbsp; Furthermore, we showcase the usage of our software package in a<br> &nbsp; full-genome screen of Drosophila melanogaster and evaluate<br> &nbsp; our results against the well-known but much less flexible program<br> &nbsp; RNAz.</p> <p>&nbsp;</p>