Lisser and Margalit (1994) attempted to determine common structural features in Escherichia coli promoters with computer analysis. They studied four intrinsic sequence-dependent characteristics in Escherichia coli (E. coli) promoter regions: "helix stability, helix flexibility, and two conformational parameters represented by the DNA tendencies for B-Z and B-A transition" (p. 823). Quantitative measures for these properties were available for the ten dinucleotides from experimental or theoretical studies; these measures were summarized in Table 1. By summing the values of the dinucleotides along the sequence, a property value for a DNA segment was obtained (p. 824).
Helix stability was evaluated by "calculating the sequence-dependent free-energy values for helix to coil transition." DNA flexibility values, and B-Z and B-A transition tendencies, were determined by experimental and theoretical studies (p. 284).
The authors compiled a data set of E. coli promoter sequences to include 300 mRNA promoters whose transcriptional-start position had been defined experimentally. Redundant sequences were excluded; all promoter sequences were verified in the literature. Next, 300 non-promoter sequences were randomly selected from E. coli coding regions (p. 825).
Comparison between promoter and random sequences was evaluated by a t-test (p. 825). Discriminant analysis was done with the SPSS package and a computing program developed in Fortran 77 and executed on VAX 9000-210 (VMS V5.5). This statistical analysis was used to show the distinction between promoter and random sequences and their relative contribution. Discriminant analysis, discriminates between two or more groups, with identification of attributes whose distributions are different between the groups; variables are weighted, combined linearly, and standardized (p. 824).
The hypothesis for this study was "that the promoter i
...