Dataset Information

Detection of Steps in Single Molecule Data.

ABSTRACT: Over the past few decades, single molecule investigations employing optical tweezers, AFM and TIRF microscopy have revealed that molecular behaviors are typically characterized by discrete steps or events that follow changes in protein conformation. These events, that manifest as steps or jumps, are short-lived transitions between otherwise more stable molecular states. A major limiting factor in determining the size and timing of the steps is the noise introduced by the measurement system. To address this impediment to the analysis of single molecule behaviors, step detection algorithms incorporate large records of data and provide objective analysis. However, existing algorithms are mostly based on heuristics that are not reliable and lack objectivity. Most of these step detection methods require the user to supply parameters that inform the search for steps. They work well, only when the signal to noise ratio (SNR) is high and stepping speed is low. In this report, we have developed a novel step detection method that performs an objective analysis on the data without input parameters, and based only on the noise statistics. The noise levels and characteristics can be estimated from the data providing reliable results for much smaller SNR and higher stepping speeds. An iterative learning process drives the optimization of step-size distributions for data that has unimodal step-size distribution, and produces extremely low false positive outcomes and high accuracy in finding true steps. Our novel methodology, also uniquely incorporates compensation for the smoothing affects of probe dynamics. A mechanical measurement probe typically takes a finite time to respond to step changes, and when steps occur faster than the probe response time, the sharp step transitions are smoothed out and can obscure the step events. To address probe dynamics we accept a model for the dynamic behavior of the probe and invert it to reveal the steps. No other existing method addresses the impact of probe dynamics on step detection. Importantly, we have also developed a comprehensive set of tools to evaluate various existing step detection techniques. We quantify the performance and limitations of various step detection methods using novel evaluation scales. We show that under these scales, our method provides much better overall performance. The method is validated on different simulated test cases, as well as experimental data.

SUBMITTER: Aggarwal T

PROVIDER: S-EPMC3743561 | biostudies-literature | 2012

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Detection of Steps in Single Molecule Data.

Aggarwal Tanuj T Materassi Donatello D Davison Robert R Hays Thomas T Salapaka Murti M

Cellular and molecular bioengineering 20120101 1

Over the past few decades, single molecule investigations employing optical tweezers, AFM and TIRF microscopy have revealed that molecular behaviors are typically characterized by discrete steps or events that follow changes in protein conformation. These events, that manifest as steps or jumps, are short-lived transitions between otherwise more stable molecular states. A major limiting factor in determining the size and timing of the steps is the noise introduced by the measurement system. To a ...[more]

PMID: 23956798

Similar Datasets

Project description:Caveolae are plasma membrane invaginations whose formation requires caveolin-1 (Cav1), the adaptor protein polymerase I, and the transcript release factor (PTRF or CAVIN1). Caveolae have an important role in cell functioning, signaling, and disease. In the absence of CAVIN1/PTRF, Cav1 forms non-caveolar membrane domains called scaffolds. In this work, we train machine learning models to automatically distinguish between caveolae and scaffolds from single molecule localization microscopy (SMLM) data. We apply machine learning algorithms to discriminate biological structures from SMLM data. Our work is the first that is leveraging machine learning approaches (including deep learning models) to automatically identifying biological structures from SMLM data. In particular, we develop and compare three binary classification methods to identify whether or not a given 3D cluster of Cav1 proteins is a caveolae. The first uses a random forest classifier applied to 28 hand-crafted/designed features, the second uses a convolutional neural net (CNN) applied to a projection of the point clouds onto three planes, and the third uses a PointNet model, a recent development that can directly take point clouds as its input. We validate our methods on a dataset of super-resolution microscopy images of PC3 prostate cancer cells labeled for Cav1. Specifically, we have images from two cell populations: 10 PC3 and 10 CAVIN1/PTRF-transfected PC3 cells (PC3-PTRF cells) that form caveolae. We obtained a balanced set of 1714 different cellular structures. Our results show that both the random forest on hand-designed features and the deep learning approach achieve high accuracy in distinguishing the intrinsic features of the caveolae and non-caveolae biological structures. More specifically, both random forest and deep CNN classifiers achieve classification accuracy reaching 94% on our test set, while the PointNet model only reached 83% accuracy. We also discuss the pros and cons of the different approaches.

Dataset Information

Detection of Steps in Single Molecule Data.

Publications

Detection of Steps in Single Molecule Data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets