An analysis of structural instances of low complexity sequence segments

Amino acid sequence databases contain many low complexity, compositionallybiased sequence segments. However, only a limited number ofrelatively short instances of these segments occur in proteinsof known structure. An analysis is presented of structural instancesof these low complexity sequence segments in the BrookhavenProtein Data Bank with regard to preferences for sequence composition,secondary structural conformation and the local atomic environment.The complexity varies almost linearly with segment length, reflectingthe absence of very long, low complexity segments in the structuraldatabase. The low complexity segments identified are not disorderedand have temperature factors which are generally the same asthe rest of the protein. It is observed that these segmentsare predominantly exposed and either helical or coiled, in excessof what would be expected by chance. Secondary structure predictionmethods perform well in correctly predicting those low complexitysegments which are helical but poorly in correctly predictingsegments that are strands.