A Validation Study of TrAQ: a Versatile and Automated Behavioural Analysis Software

Quantitative metrics of laboratory animals’ locomotion are crucial data in behavioural neuroscience studies. Commercial tracking software are stable and versatile, but quite expensive and require various degrees of user interaction, while open source tracking software are often task specific and/or cumbersome to use. Here we describe the main features of TrAQ, a behavioural analysis software developed for off-line video tracking of laboratory animals. TrAQ is a MATLAB-based tracking software developed to minimize the user interventions providing the position of the animal’s centroid, head and tail, as well as other two-dimensional behavioural parameters among which the number of body-centred turns. We tested TrAQ with smartphone-recorded videos of Open Field Test performing rats bearing 6-hydroxydopamine (6-OHDA) (n=19) induced degeneration of the nigrostriatal pathway or with sham lesion (n=19). In the 6-OHDA model of Parkinson’s Disease the number of contralateral body-centred turns performed after the administration of dopaminergic agonists is the gold standard in physiological assement of lesion extension. We validated TrAQ against the commercial software EthoVision XT comparing the centroid’s X and Y coordinates and total distance travelled for 6-OHDA and sham lesioned rats. Behavioural parameters were collected in sham and 6-OHDA treated rats before lesion (t=0, n=38) and at 1 and 21 days post-surgery (dps) in spontaneous (F, 10 min test) and under pharmacological treatment (A, 60 min test). The number of contralateral turns made by the rats after subcutaneous apomorphine injection were automatically calculated with TrAQ and compared with manual annotation. The mean difference between the X and Y centroid coordinates measured with TrAQ and EthoVision XT was less than 1‰ of the size of the arena (50 cm), while the number of body centred turns measured with TrAQ (the only free software with this feature) was in excellent agreement with the human operator counting, suggesting that TrAQ can reduce intra- operator errors, providing more robust data in a full automatic modality. At 1 dps both sham and 6-OHDA lesioned animals show statistically significant change in the average speed with respect to the pre-surgery (pre-s, t=0) test in F condition (pre-s VS 6-OHDA F 1 dps and sham F 1 dps, p<0.0001 2-way ANOVA TK pot-hoc), suggesting that the mechanical injury may affects the activity of sham lesioned animals. At 21 dps sham lesioned animals in spontaneous condition show a recovery in the average speed with respect to the 1 dps test while 6-OHDA lesioned animals' average speed remains significantly lower with respect to the pre-surgery test (6-OHDA F 21 dps VS pre-s p<0.0001 2-way ANOVA TK pot-hoc). Sham and 6-OHDA lesioned animals present an opposite trend after apomorphine administration: the former shows a decrease in the average speed with respect to the F condition, the latter a slight improvement in locomotor activity (even if not statically relevant). It is not clear to what extension the improvement is occurring and the physiological importance, if any. We observed a significant increase, between 1 and 21 dps, in the contralateral rotations number of the 6-OHDA group only (6-OHDA A 21 dps VS 6-OHDA A 1 dps, p<0.001 2-way ANOVA TK pot-hoc) because of the neurodegeneration progression. TrAQ is a versatile tracking software for bidimensional behavioural assessment of laboratory animals. Moreover, the MATLAB environment allows any kind of data analysis useful in characterising animal’s behaviour. In principle, TrAQ can be used with different animal models and it was successfully applied to track also mice and groundwater copepods