Positional Independence Models

Positional-independence models assume that each nucleotide position in the binding site contributes independently and additively to the overall binding-free energy of the DNA motif. In a PWM, the positional independence model is stored as a table of frequencies of nucleotides per position. In a PSAM, the positional independence model is stored as a table of relative affinities for each nucleotide at each position. The overall probability of binding for a PWM and the overall relative affinity for a PSAM are calculated by taking the product of all the marginal frequencies or relative affinities respectively. In the PSAM below for the Mybl2 protein, the relative affinity of the binding site TTAACCGTCG is the product 1.0 *1.0*1.0*1.0*1.0*.29*1.0*1.0*1.0*1.0 = .29. Thus, TTAACCGTCG has roughly a third of the affinity of the highest-affinity site TTAACGGTCG.



Note that PWMs and PSAMs also assume that there is only one affinity maxima in the affinity space of the protein – and therefore cannot capture possible different binding modes. However, with the inclusion of additional parameters, more complex positional independence models like pHMMs can model multiple affinity maxima and also capture specificity differences due to different binding modes.