Supplementary Figure 3: Principal component analysis of reliability of evoked motor responses.

(a) Principal component significance plot from principal component analysis on the combined data set of MSE and non-specific ventral location summary metrics (see Methods). The first component (PC1) is the only component identified as significant relative to the significance threshold (dashed red line). (b) PC1 loading pattern represents a reliability signature of low latency to the first motorneuron spike, low standard deviation in the latency to the first motorneuron spike, a high response fraction among a set of ten trials, high similarity in the pattern of motorneuron spikes, and a high fraction of dual root responses with negligible contribution of rostral-caudal location. (c) PC1 association scores (mean and standard error) for MSE evoked L5 (red), MSE evoked L2 (purple), non-specific ventral evoked L5 (orange), and non-specific ventral evoked L2 (blue) responses. A positive association score to PC1 corresponds to greater reliability. (d) Examples of hemisected transverse sections for a transynaptic MSE-focused optical stimulation (left) and a non-specific ventral interneuron stimulation (right). The blue bar represents the dorsal-ventral position of light exposure. MSE and intracord experiments had comparable numbers of ChR2-expressing neurons. Transynaptic (MSE) experiments were only considered if the spinal cords met a minimum criteria for efficiency (at least 15 interneurons in a single peak 50 μm section), and the range was 15-30 interneurons in a peak section in the transynaptic ChR2 experiments. In comparison, intracord injections had an average of 35 ± 10 interneuron per section.