A Robust Algorithm for Automated Analysis of DNA Molecules in AFM Images

E. Ficarra, E. Macii, L. Benini, and G. Zuccheri (Italy)

Keywords

AFM, DNA, Length, Image processing

Abstract

DNA molecules with a high curvature profile present, in AFM images, more indented and curved shapes. The higher the complexity of DNA molecular profiles, the higher is the difficulty in the analysis of molecular shape. We present an automated algorithm for DNA sizing with high precision and good robustness for a wide range of DNA curvature profiles and background noise contamina tions. The algorithm has been tested on AFM images con taining two sets of molecules with very different curvature profiles. In the first set the molecules were characterized by symmetric peaks of curvature of 0.04 rads. In the sec ond one, the molecules were characterized by a high curved region with a peak of 0.196 rads. In the first set, the algo rithm provided a length distributions with a mean value of 629.82nm very close to the theoretical one of 633.4nm (i.e. 0.56% of deviation), and a standard deviation of 12.5nm on mean value (about 1.98%). For the second one, we ob tained a length distributions with a standard deviation of 27.5nm on mean value of about 2.5% and a mean value of 1085nm once again very close to the theoretical one of 1098nm, that is 1.18% of deviation.

Important Links:

Go Back