Determination of melting temperature for multi-peak differential melting curves of DNA

This study was carried out because of a widely spread mistaken viewpoint that melting temperature $(T_m)$ cannot be used as a reasonable parameter in the case of multi-peak differential melting curves (DMC) of DNA. However, there were no theoretical or experimental evidences or disclaimers of this viewpoint. In this work such study has been carried out. We have tested various definitions of $T_m$. Indeed, some of them give unreasonable dependences of $T_m$ on relative concentration of modifications $(r_b)$. At same time, the average temperature of the helix-coil transition calculated as the integral over the temperature of the product of DMC and temperature gives reasonable smooth dependences $T_m(r_b)$. The same definition is the best for the dependence of $T_m$ on the average $GC$-content.