Bayesian inference assessment of protein secondary structure analysis using circular dichroism data – how much structural information is contained in protein circular dichroism spectra?

Abstract

Bayesian modelling capturing uncertainty and correlations in circular dichroism (CD) spectra suggests it is not possible to identify more than 3 distinct secondary structure classes from CD spectra above 175 nm.

Circular dichroism for secondary structure determination of proteins with unfolded domains using a self-organising map algorithm SOMSpec

“…We recently showed that, although one can annotate a protein with many independent structures, there is only enough information in a far UV circular dichroism (CD) spectrum to identify 3 types of structure, broadly α-helix or β-sheet or ‘other’. 27 So we present results in terms only of α-helix, β-sheet, and other structures.…”

Spectroscopy of model-membrane liposome-protein systems: complementarity of linear dichroism, circular dichroism, fluorescence and SERS

“…It is widely accepted that CD spectra down to 190 nm contain three independent pieces of information whereas data down into the vacuum UV region (say 175 nm) can provide more information [50]. However, we have recently shown, using Bayesian inference, that it is not possible to identify more than 3 distinct secondary structure classes from CD spectra above 175 nm, though more data do slightly improve the quality of the helix and sheet estimates [51]. The quality of the reference set has more impact on the reliability of a secondary structure estimate from CD.…”

The PCDDB (Protein Circular Dichroism Data Bank): A Bioinformatics Resource for Protein Characterisations and Methods Development

“…For example, a method of atomistically refining the structural ensemble of intrinsicallydisordered peptides was facilitated by experimental measurements using circular dichroism spectroscopy, with the PCDDB being used as a source of 411 proteins with known structures and associated CD spectra for validating the results of this investigation [30]. In other studies, a Bayesian approach to secondary structure prediction was investigated using the 71 SRCD spectra present in the SP175 dataset available in the PCDDB [31]). A study using accelerated molecular dynamics methods was validated by reverse calculation of CD data, where 107 protein spectra from the PCDDB were used to validate the results [32], and spectra prediction from classical electromagnetic theory was facilitated specifically using the specialist SRCD spectra that are available in the PCDDB [33].…”