Penalized Spline Regression for Comparing Spectroscopic Analyses of Protein Unfolding: Methods in a Bayesian Framework

• Activity Number: 295
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Miranda Lynch*+
• Companies: UConn Health Center
• Keywords: Bayesian smoothing ; curve comparison ; Bayesian inference ; nonparametric regression ; spectroscopy ; protein stability

Abstract

Spectroscopic methods such as circular dichroism are used to provide important information about protein stability and secondary structure under different experimental conditions, but formal quantitative methods for comparing spectroscopic profiles for complicated biomolecules are lacking. Absorbance measurements taken across wavelengths or taken across temperature readings in thermal denaturation experiments provide profiles that can be compared across experimental conditions. Flexible nonparametric smooth curves are fit using spline basis functions to this repeated measures data in a Bayesian framework. Bayesian inferential methods are presented to compare the curves globally over the entire range of measurements, as well as over local ranges of particular interest where characteristic behavior is present. Secondarily, the smoothed thermal denaturation profiles are compared to a parametric alternative, where the parametric model may be subject to model misspecification. The methods are exemplified using simulated and experimental data from transcriptional regulatory metalloproteins from pathogenic bacteria.