Fig. 1
From: Ultrafast data mining of molecular assemblies in multiplexed high-density super-resolution images
Schematic illustration of the direct TC algorithm. a Schematic illustration of dTC computation. Considering three sets of coordinates from blue, red, and green channel, the triple-correlation is calculated as: Each coordinate \({\mathbf{r}}_i^{{\mathrm{CH}}_{\mathrm{R}}}\) in the red channel is visited as origin O (zoom-in, highlighted red). The density of blue coordinates positioning at rRB = (rRB, θ) (highlighted blue) within the differential area (ΔS = rRB dθdr) and that of the green coordinates positioning at rRG = (rRG, θ + Δθ) (highlighted green) within the differential area (ΔS = rRG dθdr) are calculated and multiplied. The product is then integrated along θ through [0, 2π] or appropriate range [φ1, φ2] for edge correction (Supplementary Figure 1). b The same calculation in a is iteratively performed and integrated as each red coordinate is visited as an origin, followed by normalizing with the size of the canvas and the average molecular density of the three molecular species. c The computed triple-correlation profile is then plotted as a function of the triple-wise distances {rRB, rRG, rBG}. d The local maximum(s) significantly higher (≥mean + 2.5 SD) than the fluctuation of the triple-correlation profile (c) is found at certain \(\left\{ {r_{{\mathrm{RB}}}^{{\mathrm{max}}},r_{{\mathrm{RG}}}^{{\mathrm{max}}},r_{{\mathrm{GB}}}^{{\mathrm{max}}}} \right\}\) and drawn as a triangle that represents probable geometric configuration among these three species. The size of the circles at the vertexes represent the correlation amplitude
