Computational analyses for quantitation of eukaryotic proteomes

Our lab is interested in proteins, how they are expressed, translated and regulated and how these processes can adapt during different biological processes, such as during development or under stress. In order to understand how global protein levels are regulated we rely on mass spectrometry driven proteomics techniques, which aim to quantify proteomes based on the signal derived from peptides observed in the mass spectrometer. A variety of techniques are used, including both label-mediated and label-free techniques. This project seeks to benchmark, refine and improve bioinformatics tools need to generate reliable quantitative information from proteomics data, as well as analyse the results to understand different aspects of biological function.

We will seek, initially, to build on one tools designed for selection of “quantotypic” peptides, in order to improve label-free approaches and generate more robust data for exemplar systems such as yeast. By exploiting data sets in-house, as well as external repositories, we can improve our machine learning tools and include peptide detectability (Eyers et al 2011) and peptide missed cleavage (Lawless and Hubbard, 2012) as factors for machine learning classifiers.

By comparison with existing transcriptome data sets we will then aim to understand how yeast cells regulate translation from mRNA to protein, analysing and modelling our data in conjunction with external data sets to rationalise why some genes see a massive increase in the ratio of mRNA:protein (> 250,000) and others are much more modest (< 100).

Finally, we aim to extend these analyses to metazoans, exploiting the data available i from our BBSRC sLoLa project, to observe differences in the quantitative proteomes over several time points in embryonic development of the fruit fly.

This project will hence provide a broad-based training and exposure to many aspects of post-transcriptional control of gene expression using post-genomics data.