Evaluation of two automated metrics for analyzing partner preference tests

Area of Science:

• Ethology
• Animal Behavior
• Neuroscience

Background:

• The partner preference test (PPT) is crucial for studying social and sexual preferences in rodents, particularly pair bonding in prairie voles.
• Current PPT analysis relies on manual scoring, which is laborious, prone to bias, and limits high-throughput applications.

Purpose of the Study:

• To compare manual scoring with two automated metrics, "social proximity" and "immobile social contact", for PPT analysis in prairie voles.
• To determine if "immobile social contact" can accurately replicate manual scoring of "huddling" and serve as a sensitive measure of partner preference.

Main Methods:

• Manual scoring (real-time and 16x) and automated scoring ("social proximity", "immobile social contact") were compared using data from prairie voles.
• The "immobile social contact" metric was hypothesized to best correlate with manual "huddling" scores.
• Automated metrics were applied to four datasets varying in cohabitation length and sex to assess group partner preferences.

Main Results:

• Both automated metrics showed high correlation with manual scoring (R>0.90).
• "Immobile social contact" demonstrated a very strong correlation with manual "huddling" (R=0.99, P<0.001).
• "Immobile social contact" detected significant partner preference in all datasets, while "social proximity" did so in only three out of four.

Conclusions:

• Automated scoring systems, particularly those measuring "immobile social contact", are effective for high-throughput PPT analysis in prairie voles.
• Automated "immobile social contact" provides results comparable to traditional manual scoring, enhancing efficiency and reliability.
• This validates the use of automated PPT analysis for robust investigation of social preferences and pair bonding.