Fig. 2

Fly pair detection accuracy and quantitative measures obtained from multi-view integration in MCAM fly-fly social behavioral assay. a. Body parts used in fly pair, fly 1 (green) and fly 2 (red), in different view angles for locomotion analysis. Head (square, ■), center (triangle, ▲), and tail (diamond, ♦) in the top view and the side view; and cross (circle, ●), an average location representation when a fly’s head or tail is towards the camera, in the cross view. b. Fly pair detection rate (%) comparison among different DeepLabCut pose estimation implementation in MCAM fly aggression experiment setup. Detection rates of 3 sampling rates for each method are presented, with 60 Hz as original recording frame rate, and 10 Hz and 1 Hz as downsampling results from original recording. The median fly pair detection rate of multi-view integration of top, side, and cross view is 92.4 at 60 Hz, 95.4 at 10 Hz, and 98.4% at 1 Hz, across all the fly pairs, n = 199, in this study. Of all experiments, n = 86 were for studying fly pair with R26E01-GAL4 labeled neurons, n = 83 were for studying fly pair with R72A10-GAL4 labeled neurons, and n = 30 were for an independent group of TRPA1x w¹¹¹⁸ fly pair. c1. Illustration for the definitions of fly whole body length, head + thorax length, abdomen length, and fly pair distance. In c2., c3., and c4., TRPA1x w¹¹¹⁸ fly pairs were used as the baseline to evaluate the tracking performance, including female-female (F—F) pairs, n = 43, and male-male (M—M) pairs, n = 44. As a result, for individual female flies (F), n = 86 and individual male flies (M), n = 88. c2. Length comparison of whole body, head + thorax, and abdomen length between individual female and male flies of TRPA1 > w¹¹¹⁸ line. For non-normal distributions via Shapiro–Wilk test, Mann–Whitney U test was performed for whole body and abdomen length comparison between female and male flies: ***, p < 0.0005 For normal distributions via Shapiro–Wilk test, independent t-test was performed for head + thorax length comparison between female and male flies: ***, p < 0.0005 c3. Left: fly pair distance (mm) comparison between female-female and male-male of TRPA1 > w¹¹¹⁸ line across 30 min. Independent t-test: ns, p ≥ 0.05. Right: scatter plot of fly pair distance vs. the manual count of behavior of interest, head butt for female and lunge for male. Linear regression: female-female, R² = 0.327, ***, p < 0.0005; male-male, R² = 0.244, **, p < 0.005. c4. Left: fly pair traversal speed (mm/s) comparison between female-female and male-male of TRPA1 > w¹¹¹⁸ line across 30 min. Mann–Whitney U test: ***, p < 0.0005. Right: scatter plot of fly pair traversal speed vs. the manual count of behavior of interest, head butt for female and lunge for male. Linear regression: female-female, R² = 0.180, **, p < 0.005; male-male, R² = 0.000, ns, p ≥ 0.05.