Choosing Thresholds for Cromwell

Cromwell is an implementation of our algorithms for low-level processing of mass spectrometry proteomics data. Cromwell uses the undecimated discrete wavelet transform to denoise spectra; a critical parameter is the threshold to use for denoising. We have developed graphical tools to aid users in making this decision.

Sizer Plots

We have provided a readme file and two MATLAB scripts in a zip file.

The output of the first script is passed as input to the second, along with display ranges. Atypical output is displayed below. The horizontal axis is the time index in the spectrum; the vertical axis for the heatmap is te threshold. Colors indicate the deviation of the spectrum (at that threshold level) from an "infinitely" smoothed spectrum, represented by the overlaid white curve. The raw spectrum is overlaid in black, along with the smooth curves corresponding to the minimum (in red) and maximum (in green) displayed thresholds

In this example, the peak near time 8850 remains in the spectrum throughg threshold level 20, but is markedly dampened by that point. The peak near 8200 disappears from the smooth spectrum when the threshold is between 6 and 8; the peak near 9100 disappears when the threshold is around 12. By varying the time index and the displayed thresholds, the user can make an informed decision about the threshold level that best preserves the peaks that he or she finds believable.

©2003-2010 The University of Texas MD Anderson Cancer Center
1515 Holcombe Blvd, Houston, TX 77030
1-800-392-1611 (USA) / 1-713-792-6161      Legal Statements