Automatic DNA Geometric Analysis from Atomic Force Microscopy Images

摘要

This paper presents a fully automated DNA sample analysis technology, derived for analyzing images acquired from Atomic Force Microscope (AFM) systems, sampling on long-chained biopolymers such as the string-like DNA strands. Since AFM systems are capable of acquiring nanometer resolution images in a digital form, the biopolymer's Worm-Like-Chain (WLC) models geometric characteristic are embedded in this digital format. However, there is inconsistency in developing an estimation method Where there are . In order to accurately retrieve lc and lp, this paper details two crucial problem that was never dealt with before: the DNA self-crossing tracing problem and the DNA contour tangent analysis. On the first issue, bio-polymer self-cross tracing from digital image are dealt with from scanning small grids of the whole image window, and by utilizing the geometric relationship between image pixels, crossing is automatically identified from database matching, along with its trace analyzed. Where its digital Chain-Code (CC) is updated and reflects its head to tail trace. For the second issue regarding contour tangent, a novel directional sorting database is also created for the digital pixel components, where elements of the contour are individually verified for reconstructing its original tangential variation. Both of the two technologies are fully tested against simulation data for ground truth, and showed more than two times improvement with the fully automatic development.