Unraveling the Thousand Word Picture: An Introduction to Super-Resolution Data Analysis

A Bayesian Nonparametric Approach to Single Molecule Förster Resonance Energy Transfer

We develop a Bayesian nonparametric framework to analyze single molecule FRET (smFRET) data. This framework, a variation on infinite hidden Markov models, goes beyond traditional hidden Markov analysis, which already treats photon shot noise, in three critical ways: (1) it learns the number of molecular states present in a smFRET time trace (a hallmark of nonparametric approaches), (2) it accounts, simultaneously and self-consistently, for photophysical features of donor and acceptor fluorophores (blinking kinetics, spectral cross-talk, detector quantum efficiency), and (3) it treats background photons. Point 2 is essential in reducing the tendency of nonparametric approaches to overinterpret noisy single molecule time traces and so to estimate states and transition kinetics robust to photophysical artifacts. As a result, with the proposed framework, we obtain accurate estimates of single molecule properties even when the supplied traces are excessively noisy, subject to photoartifacts, and of short duration. We validate our method using synthetic data sets and demonstrate its applicability to real data sets from single molecule experiments on Holliday junctions labeled with conventional fluorescent dyes.

Efficient image reconstruction of high-density molecules with augmented Lagrangian method in super-resolution microscopy

High-density molecules localization algorithm is crucial to obtain sufficient temporal resolution in super-resolution fluorescence microscopy, particularly in view of the challenges associated with live-cell imaging. In this work, an algorithm based on augmented Lagrangian method (ALM) is proposed for reconstructing high-density molecules. The problem is firstly converted to an equivalent optimization problem with constraints using variable splitting, and then the alternating minimization method is applied to implement it straightforwardly. We also take advantage of quasi-Newton method to tackle the sub-problems for acceleration, and total variation regularization to reduce noise. Numerical results on both simulated and real data demonstrate that the algorithm can achieve using fewer frames of raw images to reconstruct high-resolution image with favorable performance in terms of detection rate and image quality.

Single molecule force spectroscopy at high data acquisition: A Bayesian nonparametric analysis

Bayesian nonparametrics (BNPs) are poised to have a deep impact in the analysis of single molecule data as they provide posterior probabilities over entire models consistent with the supplied data, not just model parameters of one preferred model. Thus they provide an elegant and rigorous solution to the difficult problem encountered when selecting an appropriate candidate model. Nevertheless, BNPs' flexibility to learn models and their associated parameters from experimental data is a double-edged sword. Most importantly, BNPs are prone to increasing the complexity of the estimated models due to artifactual features present in time traces. Thus, because of experimental challenges unique to single molecule methods, naive application of available BNP tools is not possible. Here we consider traces with time correlations and, as a specific example, we deal with force spectroscopy traces collected at high acquisition rates. While high acquisition rates are required in order to capture dwells in short-lived molecular states, in this setup, a slow response of the optical trap instrumentation (i.e., trapped beads, ambient fluid, and tethering handles) distorts the molecular signals introducing time correlations into the data that may be misinterpreted as true states by naive BNPs. Our adaptation of BNP tools explicitly takes into consideration these response dynamics, in addition to drift and noise, and makes unsupervised time series analysis of correlated single molecule force spectroscopy measurements possible, even at acquisition rates similar to or below the trap's response times.

Super-resolution Imaging of Individual Human Subchromosomal Regions in Situ Reveals Nanoscopic Building Blocks of Higher-Order Structure

It is widely recognized that the higher-order spatial organization of the genome, beyond the nucleosome, plays an important role in many biological processes. However, to date, direct information on even such fundamental structural details as the typical sizes and DNA content of these higher-order structures in situ is poorly characterized. Here, we examine the nanoscopic DNA organization within human nuclei using super-resolution direct stochastic optical reconstruction microscopy (dSTORM) imaging and 5-ethynyl-2'-deoxyuridine click chemistry, studying single fully labeled chromosomes within an otherwise unlabeled nuclei to improve the attainable resolution. We find that, regardless of nuclear position, individual subchromosomal regions consist of three different levels of DNA compaction: (i) dispersed chromatin; (ii) nanodomains of sizes ranging tens of nanometers containing a few kilobases (kb) of DNA; and (iii) clusters of nanodomains. Interestingly, the sizes and DNA content of the nanodomains are approximately the same at the nuclear periphery, nucleolar proximity, and nuclear interior, suggesting that these nanodomains share a roughly common higher-order architecture. Overall, these results suggest that DNA compaction within the eukaryote nucleus occurs via the condensation of DNA into few-kb nanodomains of approximately similar structure, with further compaction occurring via the clustering of nanodomains.

Consistent prediction of GO protein localization

The GO-Cellular Component (GO-CC) ontology provides a controlled vocabulary for the consistent description of the subcellular compartments or macromolecular complexes where proteins may act. Current machine learning-based methods used for the automated GO-CC annotation of proteins suffer from the inconsistency of individual GO-CC term predictions. Here, we present FGGA-CC⁺, a class of hierarchical graph-based classifiers for the consistent GO-CC annotation of protein coding genes at the subcellular compartment or macromolecular complex levels. Aiming to boost the accuracy of GO-CC predictions, we make use of the protein localization knowledge in the GO-Biological Process (GO-BP) annotations to boost the accuracy of GO-CC prediction. As a result, FGGA-CC⁺ classifiers are built from annotation data in both the GO-CC and GO-BP ontologies. Due to their graph-based design, FGGA-CC⁺ classifiers are fully interpretable and their predictions amenable to expert analysis. Promising results on protein annotation data from five model organisms were obtained. Additionally, successful validation results in the annotation of a challenging subset of tandem duplicated genes in the tomato non-model organism were accomplished. Overall, these results suggest that FGGA-CC⁺ classifiers can indeed be useful for satisfying the huge demand of GO-CC annotation arising from ubiquitous high throughout sequencing and proteomic projects.

Bridging intracellular scales by mechanistic computational models

The impact of intracellular spatial organization beyond classical compartments on processes such as cell signaling is increasingly recognized. A quantitative, mechanistic understanding of cellular systems therefore needs to account for different scales in at least three coordinates: time, molecular abundances, and space. Mechanistic mathematical models may span all these scales, but corresponding multi-scale models need to resolve mechanistic details on small scales while maintaining computational tractability for larger ones. This typically results in models that combine different levels of description: from a microscopic representation of chemical reactions up to continuum dynamics in space and time. We highlight recent progress in bridging these model classes and outline current challenges in multi-scale models such as active transport and dynamic geometries.

Robust model-based analysis of single-particle tracking experiments with Spot-On

Single-particle tracking (SPT) has become an important method to bridge biochemistry and cell biology since it allows direct observation of protein binding and diffusion dynamics in live cells. However, accurately inferring information from SPT studies is challenging due to biases in both data analysis and experimental design. To address analysis bias, we introduce 'Spot-On', an intuitive web-interface. Spot-On implements a kinetic modeling framework that accounts for known biases, including molecules moving out-of-focus, and robustly infers diffusion constants and subpopulations from pooled single-molecule trajectories. To minimize inherent experimental biases, we implement and validate stroboscopic photo-activation SPT (spaSPT), which minimizes motion-blur bias and tracking errors. We validate Spot-On using experimentally realistic simulations and show that Spot-On outperforms other methods. We then apply Spot-On to spaSPT data from live mammalian cells spanning a wide range of nuclear dynamics and demonstrate that Spot-On consistently and robustly infers subpopulation fractions and diffusion constants.

Single-molecule tracking in liveYersinia enterocoliticareveals distinct cytosolic complexes of injectisome subunits

Single-molecule tracking of bound (blue trajectories) and diffusive (red trajectories) injectisome subunits reveals the formation of distinct cytosolic complexes.

Photobleaching of YOYO-1 in super-resolution single DNA fluorescence imaging

Super-resolution imaging of single DNA molecules via point accumulation for imaging in nanoscale topography (PAINT) has great potential to visualize fine DNA structures with nanometer resolution. In a typical PAINT video acquisition, dye molecules (YOYO-1) in solution sparsely bind to the target surfaces (DNA) whose locations can be mathematically determined by fitting their fluorescent point spread function. Many YOYO-1 molecules intercalate into DNA and remain there during imaging, and most of them have to be temporarily or permanently fluorescently bleached, often stochastically, to allow for the visualization of a few fluorescent events per DNA per frame of the video. Thus, controlling the fluorescence on-off rate is important in PAINT. In this paper, we study the photobleaching of YOYO-1 and its correlation with the quality of the PAINT images. At a low excitation laser power density, the photobleaching of YOYO-1 is too slow and a minimum required power density was identified, which can be theoretically predicted with the proposed method in this report.