Key points are not available for this paper at this time.
A reliable and widely used transmembrane protein structure prediction algorithm was applied to five representative genomic sequence data sets in order to re-examine the hypothesis that in contrast to globular proteins there are no favored transmembrane protein fold families. When the number of predicted membrane spanning segments and the topology of these segments is taken into account then definite biases are observed which suggest that certain transmembrane topologies are significantly more common than others.
David T. Jones (Fri,) studied this question.