Analyzing Possibly Misannotated Coding lncRNAs Discovered Through Deep Learning Models

Although lncRNAs -by definition- do not code for proteins, it has been reported that some small open reading frames (ORFs) within some lncRNAs are translated into functional micropeptides.
We recently proposed a computational method to identify possibly misannotated lncRNAs from sequence information alone. Our approach first builds deep sequential learning models to discriminate coding and noncoding transcripts and leverages these models' training dynamics to identify coding RNAs that are possibly misannotated as lncRNAs. In this project, students are expected to investigate the candidate set discovered and their possible biological relevance. Biological background (or motivation to learn) and good programming skills are required for this project.