A comparative study of analysis methods in quantitative label free proteomics

The large amounts of data generated by modern proteomics experiments necessitates the use of software pipelines to conduct the bulk of the post-processing. While many software packages (both commercial and open-source) are available to perform some or all of the necessary post-processing steps, it is usual for each research group to use only the instrumentation and software packages with which they are most familiar and/or which are available to analyse their unknown data. The intention of the studies presented within this thesis was to assess the correlation between the experimental results obtained when; - a single result dataset is obtained and post-processed in parallel using four separate software pipelines - a single sample is analysed on two different mass spectrometers and post-processed in parallel and; - when different identification thresholds are applied to a dataset prior to parallel quantitation of the resultant data sets Correlation between different mass spectrometry instruments was assessed and found to yield high r values, especially at the protein level, and was also found to improve following the application of abundance thresholds, however the result of applying score thresholds was unpredictable. The use of manual FDR thresholds prior to importing data into Progenesis LC-MS yielded interesting results, which suggest that a threshold of 1% peptide FDR and 1 or 2% protein FDR is most effective in terms of yielding accurate ratios while maintaining acceptable sensitivity. In addition, a consensus method is suggested to utilise the results from multiple software pipelines in order to increase sensitivity and reduce the FDR, through the use of the QPROT tool and manual post-processing.