Image processing and calculating ΔF/F

We performed all calcium image processing and analyses using scripts written in MATLAB (MathWorks). After acquiring the images for a trial, we first applied a Gaussian filter (size 5×5 pixel, Ʃ = 3) and intensity threshold to minimize background noise. For calculating the GCaMP6 fluorescence change relative to the baseline (ΔF/F), we used the lowest average fluorescence level in a 10-frame window as the baseline fluorescence during that trial. For cases in which calcium signals were reduced relative to baseline (e.g., 19Aa neurons), we used the average fluorescence level in a 10-frame window at the beginning of each trial as the baseline. Because not all flies co-expressed tdTomato, we did perform image registration to correct for sample movement. From those flies that did co-express tdTomato, we observed that movement of the VNC was negligible.

We defined three parameters to analyze the temporal dynamics of calcium signals, as shown in Figure 1G: peak ΔF/F during the stimulation window, the time after stimulation at which the ΔF/F reaches 50% of the peak value (Figure 1I), and the half-decay time after the peak ΔF/F is reached (Figure 1J). For quantification of adaptation in Figure 5D, we calculated an adaptation index as 1 − F_offset/F_peak, where F_peak indicates the peak ΔF/F, and F_offset is ΔF/F 19 s after the stimulus onset (where the stimulation offset typically occurs in club/10Ba neurons). An adaptation index of 1 would indicate 100% decay to baseline, while an index of 0 would indicate no adaptation. Negative index values indicate an increase in the calcium signal over time.