Brown, A.E.X. et al. Proc. Natl. Acad. Sci. USA 110, 791–796 (2013).

In the examination of subtle behavioral phenotypes, manual scoring has several drawbacks. Automated methods are increasingly being applied to mitigate these problems: predefined patterns can be rapidly and reproducibly identified by computer algorithms in both static images and video of many model organisms. Brown et al. show with Caenorhabditis elegans that behavior can also be automatically analyzed without any predefinition, solely on the basis of the patterns present in the data. Starting with video data of about 300 mutant worm strains, the researchers identified recurring behavioral motifs that could be strung together into a profile describing the phenotype of a given strain. Phenotypic profiles can be further compared between strains to generate hypotheses about gene function.