Multiple-Emitter Super-resolution Imaging using the Alternating Descent Conditional Gradient Method

Abstract

This thesis examines the state-of-the-art 2D super-resolution technique alternating descent conditional gradient (ADCG) method's ability to accurately localize fluorophores in diffraction-limited single molecule images (SMI) and analyze the impact of pre-processing and post-processing modules on ADCG's fluorophore localization. A synthetic dataset obtained from the 2013 Grand Challenge localization microscopy and a temporally linked dataset obtained from an unpublished set of Optical DNA mapping experiments performed by Jonathan Jeffet at the NanoBioPhotonix Lab at Tel-Aviv University were initially segmented to extract their noise parameters. Subsequently, the fluorophores in both datasets were localized on the order of nanometers and tens of nanometers, respectively, using ADCG. Additionally, leveraging intrinsic fluorophore behavior, as detailed in Jaqaman et al., improved the specificity of the temporally linked localizations. Moreover, by time averaging the fluorophore positions, as detailed in Jeffet et al., the precision of the temporally linked localizations was improved by a factor of 2. Consequently, it is concluded that ADCG can rapidly generate super-resolution images, robustly track single molecules, and generate DNA barcodes for diffraction-limited SMIs in an automatable fashion.

Popular Abstract

Plants, viruses, "the birds and the bees" all function due to sophisticated bio-molecular machinery, whose cumulative and simultaneous action results in what we call life.

Miraculously, an organism's DNA, or RNA, contains the blueprints for all the little bits and pieces of bio-molecular machinery necessary to sustain life. If we could read such blueprints, we could determine an organism's ancestry, create treatment schemes for hereditary diseases, or even identify an organism's mode of infection and consequently deduce ways to stop its spread. These blueprints are written in a language that consists of four letters separated from one another by approximately a third of a nanometer. However, this presents a problem as reading such tightly spaced ``letters" requires a level of precision that is unattainable, per the inescapable laws of physics, which state that in a microscope, the resolution is limited by the wavelength of light observed and the numerical aperture of the lens system; this is known as Abbe's diffraction limit.

Scientists have worked around the aforementioned problem by using next-generation sequencing (NGS) to determine the genes present in a sample. Such a process is quite expensive and time-consuming. Furthermore, the generated DNA sequences are short snippets of the sequenced gene. Hence, a tremendous amount of processing power is needed to piece these snippets together. Generally, this results in imperfect genome reconstructions, which may stall and even prevent the development of life-saving treatments. Thus, complementary methods are crucial in further improving genome reconstructions. One such method is optical gene mapping, where pictures of the gene labeled using fluorescent proteins at commonly appearing DNA regions provide a map for the final genome assembly. However, the Abbe diffraction limit prevents a detailed skeleton, as the light sources in the optical maps blend, resulting in the individual sources being indistinguishable from one another.

Fortunately, novel solutions that aim to defy the Abbe diffraction limit have been proposed. Among these solutions, the alternating descent conjugate gradient method, ADCG for short, has proven to be a state-of-the-art tool for resolving single-molecule images containing fluorescent dots. Using ADCG, highly detailed maps can be created in which the light sources, or fluorophores, are localized with precision in the tens of nanometers. These maps can then be used to rapidly identify and characterize bacterial strains in an economical and automatable manner, allowing scientists to react to mutations, consequently saving lives.

A typical single-molecule image consists of a DNA segment that has been chemically labeled and imaged using a microscope. Unfortunately for such images, Professor Abbes' law seemingly holds, as finding the individual segments--—which are shaped like dots--—seems impossible. Alas, ADCG proposes a relatively simple solution. First, it assumes that the image comprises the minimum number of dots possible; in the context of information theory, this makes its predictions "sharp." Consequently, ADCG sequentially constructs synthetic fluorescent dots and moves them within the image until it effectively reconstructs the original image. This is done by minimizing the difference between the real and ADCG-generated images.

Using ADCG, the molecular motion of life can be captured and viewed at "super-resolution" that surpasses the laws of physics, which lends itself to assisting and, at times, replacing NGS and other sequencing methods as a fast and cheap alternative. Hence, developing an updated software library featuring additional functionalities in a modern computing language will expedite scientific discovery in molecular biology and nano-biophysics.

• author: Illouz, Dolev
• supervisor: Tobias Ambjörnsson
• organization: Computational Biology and Biological Physics - Undergoing reorganization; Department of Astronomy and Theoretical Physics - Undergoing reorganization
• course: FYTK02 20201
• year: 2023
• type: M2 - Bachelor Degree
• subject: Physics and Astronomy
• keywords: super-resolution, alternating descent conditional gradient method, ADCG, multiple-emitter localization, nanoscale imaging, single molecule tracking
• language: English
• id: 9136489
• date added to LUP: 2023-09-06 16:47:48
• date last changed: 2023-09-06 16:47:48