Use of proteomics as a method of enhancing genome annotation.

Abstract

While the human genome was sequenced in 2001, its full annotation continues to be a challenging area of research. Our inability to clearly define the location of a gene and its protein products are major obstacles to completing the annotation. In recent years, alternative splicing (AS) as been recognized as a major biological factor contributing to increased complexity in the genome. It is through AS that a single genomic locus can encode for multiple protein products. In the case of the human genome, AS enables an estimated 40,000 known proteins to be produced from the approximately 23,000 gene loci currently annotated. The number of known proteins is still considered only an estimate and does not take into account, tissue-specific proteins and post-translational modifications. Numerous computational methods have been developed to identify coding regions in genomic sequence. However, even the best software packages available today are only 70-75% accurate. In order to overcome this, it is important to combine computational methods with experimental data in order to achieve better annotations. With the advent of high-throughput proteomics, large amounts of tandem mass spectral data are now being generated. This data can be interrogated to determine what proteins were present in the original biological sample analyzed. However, searching mass spectral data using databases of known proteins would overlook potentially novel AS isoforms. To address this concern it is necessary to search a database that contains both known and potential protein sequences. In this thesis we describe the use of established technologies to identify novel exons to previously annotated genes. We search a large repository of mass spectral data collected from human blood against an exhaustive six-frame translation of the human genome. In addition, the subset of the mass spectral data was searched using a database of putative alternatively spliced transcripts constructed from expression data to validate poorly characterized transcripts. Through our six-frame translation searches, we identified 275 novel exons to known genes. In searching our putative transcript database we were able give proteomics-based support for 78 transcripts for which there was previously only weak expression data evidence.