Using Intron Position Conservation for Homology-Based Gene Prediction

Next generation sequencing has lead to a rapid increase in the number of sequenced genomes. Initial annotation of protein-coding genes in newly sequenced genomes is typically based on computational predictions. Here, we present a homology-based gene prediction program called GeMoMa, which explicitly incorporates the conservation of intron positions. GeMoMa utilizes gene models from a related species and predicts gene models in the genome of an organism of interest. In contrast to transcriptomics-based gene predictions, GeMoMa is capable of predicting rarely transcribed genes. By design, GeMoMa, provides information about putative homologous gene pairs and allows for transferring information about gene function from one organism to another. We apply GeMoMa to several animal and plant species and compare it with state-of-the-art competitors based on available annotations, using RNA-seq data, and Sanger sequencing. Our key findings are: i) Utilizing intron position conservation improves homology-based gene prediction and ii) predictions of GeMoMa can help to improve existing or add new transcripts in annotated genomes. The development of homology-based gene prediction tools has largely stalled during the last years. However, we demonstrate that the inclusion of additional features may substantially improve prediction performance. Hence, our results might trigger the investigation of further features.