Automated and Robust Quantification of Colocalization in Dual-Color Fluorescence Microscopy: A Nonparametric Statistical Approach

Abstract

Colocalization is a powerful tool to study the interactions between fluorescently labeled molecules in biological fluorescence microscopy. However, existing techniques for colocalization analysis have not undergone continued development especially in regards to robust statistical support. In this paper, we examine two of the most popular quantification techniques for colocalization and argue that they could be improved upon using ideas from nonparametric statistics and scan statistics. In particular, we propose a new colocalization metric that is robust, easily implementable, and optimal in a rigorous statistical testing framework. Application to several benchmark data sets, as well as biological examples, further demonstrates the usefulness of the proposed technique.

Spatially Adaptive Colocalization Analysis in Dual-Color Fluorescence Microscopy

“…Three different regions of interest (see Figure 16) in the left most images of Figures 7, 8 and 11 were chosen to evaluate the performance of previous ROIbased methods. Following the 3-step procedure, we applied Pearson correlation coefficient [2], Manders' split coefficients (M 1 , M 2 ) [3], Intensity correlation quotient (ICQ) [4], and maximum truncated Kendall tau correlation coefficient τ * (MTKT) [21]. For M 1 , M 2 , and τ * , the thresholds or lower bound of thresholds were determined by Otsu's method.…”

“…Our general framework can be applied with a generic colocalization index, such as Pearson's correlation or Mander's split coefficient. To be specific, we shall characterize colocalization using a general and robust criterion recently introduced by [21]:…”

“…All other parameters were chosen according to guidance in Section III-B. Specifically, we considered five different conditions of HeLa cells with corresponding images (Figure 7, 8, and 11 are from [21]). In the first four conditions (Figure 7-10), HIV-1-Gag (green channel) was fused to cyan fluorescence protein (CFP) and MS2 protein (red channel), which tracks the viral RNA genome, was fused to yellow fluorescent protein (YFP).…”

“…Here (X, Y ) and ( X, Ỹ ) are independent, identical copies following distribution F . Assuming that for all pixels i in a ROI, (X i , Y i ) ∼ F , the level of colocalization, as measured by (3), in the ROI with index set I can be conveniently measured by Kendall's tau coefficient over all pixels in I [21] for further details.…”

Transport Dependency: Optimal Transport Based Dependency Measures

“…Regarding functional data analysis, we refer to (Petersen and Müller 2019;Dubey and Müller 2019). For colocalization problems in cell microscopy, we mention Wang et al (2017), who propose a statistically optimal colocalization metric based on Kendall's 𝜏, and Tameling et al (2021), who suggest certain surrogates of the optimal transport plan for quantifying colocalization.…”