Figure 8
From: Position dependent mismatch discrimination on DNA microarrays – experiments and model
Comparison of the two-state nearest neighbor (TSNN) model and the zipper model (partition function approach – PFA). To investigate for which sequences the difference between TSNN free energies and PFA free energies is largest, we have created a large set of 5000 random 25 mer sequences with a similar nucleobase composition. Scatter plots of TSNN free energies versus PFA free energies (left) show a very good correlation at a temperature of 310 K (A). At higher temperatures (340 K and 360 K shown in B and C) we find significant deviations between the two models. We have selected the 5% of sequences with the largest residuals (highlighted by red symbols) and determined the position-dependent distribution of NN free energies (shown right) by averaging (→ averaged NN pair free energy versus NN-pair position. The Gibbs free energies in upper, middle and lower plots refer to temperatures T = 310 K, 340 K and 360 K, respectively). At 310 K the sequences with the most stable ΔGPFA have their weak NN pairs at the outermost two base positions (dashed black line) and therefore the more strongly binding NN pairs in the interior. Vice versa sequences with the weakest ΔGPFA (solid green line) have strong NN pairs located at the outermost positions. The mean NN free energy (average over all sequences) is indicated by the dotted red line. At 340 K for the most stable sequences (according to PFA) the weakest NN-pairs are concentrated at the six outermost base positions (at each duplex end). At 360 K (which is above the melting temperature of the duplexes) the NN pair stabilities follow a parabolic position dependence.