No abstract

No abstract

Proton transfer in malonaldehyde was studied by molecular dynamics simulations with the projector augmented wave (PAW) method, which combines classical dynamics with ab initio quantum mechanical forces. The PAW trajectories were calculated for several temperatures between 1 and 600 K, for evolution time periods up to 20 ps, and with a constant time interval of 0.12 fs. At elevated temperatures proton transfer is not associated with a well-defined C 2v -symmetric transition state, but takes place in widely differing geometric situations. Although a short O ± O distance favors proton transfer, it is neither a sufficient nor a necessary condition. Analysis of the data by a discriminant method and with a neural network yielded several relevant molecular parameters, and the resulting discrimination functions predicted the occurrence of proton transfer with an accuracy greater than 95 %. The energetics of the proton motion was modeled by calculating time evolutions of the potential energy along a properly chosen reaction coordinate within a heavy ± light ± heavy atom approximation. At any instant the proton motion is governed by this potential, but while the proton moves, the potential also changes due to the dynamics of the molecule. Three extremes can be distinguished: i) Normal periods, in which the proton is trapped at one oxygen atom. The proton is stationary within an approximately constant, strongly asymmetric potential; the frequency of about 2850 cm À1 is close to the experimentally observed n(OH) frequency. ii) Statistical isolated proton-transfer transitions, in which the proton rapidly moves from one oxygen atom to the other. The process starts and ends with strongly asymmetric potentials, but passes through (nearly) symmetric double-or single-minimum potentials. iii) Proton-shuttling periods, which include several consecutive nonstatistical transitions. These are not true proton transfers. The proton is (quasi)-stationary within a (nearly) symmetric single-minimum potential, which remains approximately constant for a longer time period; the motion corresponds to a n(OH) vibration with a frequency of about 2000 cm À1

3lp N]V[R spectroscopy has been used to study the intercalation of the anthracyclines doxorubicin 1, daunorubicin 2, 4-demethoxydaunorubicin 3, morpholinodoxorubicin 4, methoxymorpholinodoxorubicin 5 and 9-deoxydaunorubicin 6 with the DNA fragment d(CGTACG)2. The individual phosphate resonances of the oligonucleotide were assigned in the free as well as in the intercalated species. The 3~p chemical shitt vadations allowed us to identify the intercalation sites, which resulted to be the same for all compounds i e. between the terminal CG base-pairs of the helix (two molecules of drug per duplex). The binding coustants, the dissociation rete constants and AG ~ values have been determined in different conditions of ionic strength and temperature. The kinetic constant (ko~) of the slow step of the anthracycline/duplex intercalation process has been directly measured by NOE exchange techniques. Binding constants depend on the ionic strength and on the self-association process so greatly, that their use to study by NMR anthracycline/DNA interactions is questionable. On the contrary, the kof r are not affected by these phenomena and presentan interesting trend for 1~, thus showing that the average lifetime of the drug in the intercalation site appears to be important for determining the cytotoxicity and the antimitotic activity.

Cardiotocography signals were sampled during labour in 53 patients. A recurrent artificial neural network with hidden layer feedback was trained and performance was compared with that of several conventional systems. Correct and false positive rates of all systems tested were calculated. To ensure that the performance of neural networks was not just caused by using different cut-off levels, the threshold used for conventional methods were also adapted and optimised. The correct positives rate of neural networks was between 0.72 and 0.9, and the false positive rate between 0.2 and 0.4. Before optimising, conventional algorithms produced a very low correct positive (0.02-0.5) and a low false positive rate (0.0-0.08). After adjusting the parameters, the tested neural networks still performed better than optimised conventional systems.