Use of artificial neural networks for the evaluation UV-Vis spectra of adenine and cytosine in mixtures

• Authors: TOPINKOVÁ Jana; TRNKOVÁ Libuše; HAVEL Josef
• Year of publication: 2002
• Type: Article in Proceedings
• Conference: Chemické listy
• MU Faculty or unit: Faculty of Science
• Field: Analytic chemistry
• Keywords: adenine;cytosine;UV-Vis spectra;artificial neural networks;TRAJAN 6;chemometrics
• Description: Recently we have demonstrated advantages of artificial neural networks (ANNs) for quantification of nucleic acid bases mixtures from overlapping voltammetric signals. In this work ANNs were applied to analyze adenine-cytosine mixtures from UV-Vis spectra. Spectra of these compounds are highly similar (lambda max 261 and 270nm, respectively) and common spectral approaches of analysis are failing. The optimal ANN architecture, able to model the system, was established. Software of ANNs method was from TRAJAN 6 program. Using the suitable experimental design with soft ANN modeling adenine - cytosine mixtures can be quantified with an excellent precision (less than 1%). Generally, ANNs open many possibilities for the analysis of mixtures of biologically important substances with very overlapped signals, even without the explicit knowledge of their behavior.

Related projects

• Electrodes modified with nucleic acids and proteins. New tools in biochemical and biomedical research
• New trends in electrochemistry of nucleic acids and their applications in enviromental chemistry
• Application of adsorptive transfer and elemination techniques in oligonucleotides and nucleic acids analysis.
• Structure and character of bonding, properties and analysis of synthetic and natural molecular ensembles