The rate ofcis–transconformation errors is increasing in low-resolution crystal structures

Abstract

Cis-peptide bonds (with the exception ofX-Pro) are exceedingly rare in native protein structures, yet a check for these is not currently included in the standard workflow for some common crystallography packages nor in the automated quality checks that are applied during submission to the Protein Data Bank. This appears to be leading to a growing rate of inclusion of spuriouscis-peptide bonds in low-resolution structures both in absolute terms and as a fraction of solved residues. Most concerningly, it is possible for structures to contain very large numbers (>1%) of spuriouscis-peptide bonds while still achieving excellent quality reports fromMolProbity, leading to concerns that ignoring such errors is allowing software to overfit maps without producing telltale errors in, for example, the Ramachandran plot.