Poisson noise model for low-abundance labels in iTRAQ

This dataset/code forms part of Andrew Landels' thesis: "Improving proteomic methods and investigating H2 production in Synechocystis sp. PCC6803" http://etheses.whiterose.ac.uk/id/eprint/19034

Proteomic iTRAQ analyses generate background noise, which can generate false positive results in very low abundance analyses. In this analysis, a sample dataset (published in Chiverton et al) is used to generate a model for noise - this is possible because within the experiment, two iTRAQ labels were strategically left blank.

This code, written in R, produces two figures. The first is a histogram of the empty labels in the dataset; and the second is a histogram of a series of values produced using a Poisson distribution of random noise. This comparison shows that due to the discrete nature of mass spectrometer data at low intensities (on the scale of individual counts, as opposed to hundreds or thousands in typical measurements), a Poisson model would need to be used to accurately model the noise.