A novel computational approach applicable to human microbiome studies – urinary tract microbiome exaple

Peptide mass fingerprinting is a term which describes technique which utilizes ESI or MALDI MS followed by tandem mass spectrometry sequencing. This technique has become a cornerstone for protein identification. Today, applications using peptide mass fingerprinting in biomedical analyses are a major driving force behind its rapid development. However, efficient and accurate analyses of generally big protein tandem mass spectrometry data sets require robust software. In terms of final goal, which is data interpretation, the role of software and underlying algortihms is at least equally important as the technique itself, a fact which is often neglected. High-throughput mass spectrometry instruments can readily generate hundreds of thousands of spectra. This fact combined with the ever growing size of genomic databases imposes tremendous demands for potential successful softvare solutions. In fact, it is the process of comparing large-scale mass spectrometry data with large databases that remains the toughest bottleneck in proteomics. Here we present a completely novel approach based on natural language processing which is not just another improvement of existing approaches, but represents a paradigm shift. It doesn't rely on peak intesity for database peptide matching and it uses newly developed concept of microbial proteome fingerptints for strain/species identification. Since this new algorithm doesn't rely on sequence alignment but instead utilizes a concept of singular proteome fingerprints rather than sets of unrelated peptides, it proposes an elegant solution for this most troubling step in proteome analyses. Abandoning BLAST and other alignment based methods. results in far superior processing speed, accuracy and sensitivity. The above mentioned algortithm can be used to analyse not only proteomes but also metaproteomes coming from mixed microbe communities as in the case presented – human urine samples taken from a hospital. The method itself is completely generic, not developed with any specific platform in mind, which makes it highly versatile, able to turn any existing device into highly efficient metaproteome analyzer without siginificant costs related to purchase of new equipment. This work was funded by HRZZ (Croatian Science Foundation) research project “Clinical proteomics of microorganisms”.