Лекции профессора Хенрика Флювберга 2 и 3 октября 2013 года

H. Flyvbjerg

2 и 3 октября в 446 аудитории химического факультета МГУ состоятся лекции профессора факультета микро- и нанотехнологий Технического Университета Дании (Копенгаген) Хенрика Флювберга «Integrated view of genome structure and sequence of a single DNA molecule in a nanofluidic device» и «Optimized localization-analysis for single-molecule tracking and super-resolution microscopy».

Начало первой лекции 2 октября – 12 часов. Приглашаются студенты, аспиранты, научные сотрудники, все желающие. Лекции – на английском языке. Если у вас нет пропуска в МГУ, обращайтесь к Гудилину Евгению Алексеевичу.

Integrated view of genome structure and sequence of a single DNA molecule in a nanofluidic device

Is it possible to see the genetic contents of a single DNA molecule in a light microscope? Obviously not: A light microscope can only distinguish points more than 100 nm apart while base-pairs in DNA pack three to one nanometer. On the other hand: DNA is inhomogeneous on many length scales, so the real question is: How much can one see, if one can observe DNA with 1 kb resolution? In order to get that resolution, one must stretch a DNA molecule fully, turn off its Brownian motion, and make its large-scale structure light up. We did that with an intercalating fluorescent dye and simple physical methods. In long (~2 Mb) DNA molecules extracted from whole chromosomes in a nanofluidic lab-ona- chip, we detected repetitive elements, known structural variation, as well as unique structural variations. After being mapping in this manner, a molecule of interest was rescued from the chip, amplified, and analyzed with standard methods of genetics, including sequencing to 1-bp resolution. This confirmed what we had observed with light microscopy. I will explain how theoretical physics contributed critically to the success of this project.

Optimized localization-analysis for single-molecule tracking and super-resolution microscopy

In single-molecule tracking and localization microscopy, spatial resolution of a few nanometers is achieved. This is a hundred times below the diffraction limit. Also, far-field super-resolution techniques now exist, such as photo-activated localization microscopy (PALM) and stochastic optical reconstruction microscopy (STORM), which sequentially isolate each fluorescent probe in a densely labeled sample to resolve intracellular protein localization patterns to within a few nanometers. Additional development and application of super-resolution microscopy is likely to provide insight into cellular processes at both systems and mechanistic levels. With such popularity of a technique, it is natural to ask whether it is optimal. That is: How does one optimally localized isolated fluorescent beads and molecules imaged as diffraction-limited spots? How does one optimally determine the orientation of fluorescent molecules? How does one obtain reliable formulas for the precision of various localization methods? Answer: Use maximum-likelihood estimation and physically correct point-spread functions. Doing that, both theory and experimental data show that
(A) the very popular non-weighted least-squares fit of a Gaussian point-spread function to data squanders one-third of the available information,
(B) a popular formula for its precision exaggerates beyond the information limit, and
(C) a weighted least-squares fit may look like an improvement, but may do worse, whereas
(D) maximum-likelihood fitting is practically optimal.
I will explain everything in terms of basic physics and a little statistics.