Applying Bioinformatics Workflows to Proteogenomics

The Genomes Online (www.genomesonline.org) (27^th February 2013) database contains an estimated 4,136 “complete genomes” from the reported 21,323 genomes. The true number of “complete” genomes is most likely much lower than this due to the potential for missed or incorrect annotations. Many manual/semi-automatic gene annotation systems have been implemented in the past, which use comparative evidence (EST, RNA-seq, FL-cDNA and protein), and then rely on ab-initio gene prediction tools when comparative evidence is lacking. Many systems are not capable of handling the sheer number of genomes required, to achieve high-level curation. Additionally, these approaches lack sensitivity resulting in erroneously predicted genes. Recently, a new methodology coined ‘proteogenomics’ has been devised which incorporates peptide mass spectrometry to improve genomic annotations. In this presentation we incorporate proteogenomic pipelines and demonstrate the utility of proteogenomics to improve genomic annotations. For instance, in Wheat genomics, we were able to map peptides to 76 putative genes and also uncovered 6 novel annotations. In Grape genomics, we mapped peptides to 2,587 known genes and uncovered 127 novel annotations. We also describe a strategy to further mine proteomic data towards the identification of sites of post-translational modifications (PTMs), post- translational processing (e.g. signal cleavage) and gene expression analysis through protein abundance calculations. We outline our intentions to incorporate these pipelines with high performance computing clusters and to utilise a workflow environment such as Yabi (ccg.murdoch.edu.au/yabi) to allow these tools easier access to the wider proteomics and genomics community.