Picking up someone else's project is rarely top of a researcher's wish list, especially if it has been around for more than four decades. But if the project happens to be that of Jerry Hirsch, then perhaps the idea is worth reconsidering. In the 1950s and 1960s, this 'drosophilist' sought to analyse the genetic basis of behaviour by artificially selecting lines of Drosophila that had an extreme preference for moving towards or against gravity in a vertical maze. Although he was able to establish, for the first time, that this so-called geotaxic behaviour — indeed any behaviour — has a genetic basis, getting to the underlying genes just wasn't possible at that time. However, by applying cDNA microarray experiments and mutant analysis to the original lines generated by Hirsch, Daniel Toma, along with Ralph Greenspan, Kevin White and Jerry Hirsch himself, have now partly realized the original researcher's aim by identifying three genes involved in fly geotaxic behaviour. The genetic basis of any selected phenotype is rather impenetrable, even today, making this work all the more remarkable.

If Hi (flies that like to go 'up') and Lo (flies that like to go 'down') lines behave differently, then the genes involved in this divergent phenotype are probably differently expressed in the two lines. In a microarray experiment to assess this, the authors identified 250 genes whose expression levels differed at least twofold between the two lines.

Toma et al. decided to pursue only those candidates from their microarray analysis for which mutants with neurological defects already exist. This left them with four mutant lines — cryptochrome (cry), Pendulin (Pen), Pigment-dispersing factor (Pdf) and prospero (pros) — which were tested for their preference to go up or down. The geotaxic score of three of the mutants, the exception being pros, was significantly different from that of wild-type flies and correlated with the difference in mRNA levels seen in the selected Hi and Lo lines. The dosage effect of gene expression on behaviour was also tested by generating transgenic flies that expressed wild-type Pdf and pros in backgrounds with varying copies of the endogenous transcript. Although altering the level of pros had no significant effect on the geotaxic score (as predicted from the mutant data), altering the dosage of Pdf produced a graded effect, which differed between the sexes.

This file might have been an old one but, despite the qualitative advance reported here, it still isn't closed. How do Pen, Pdf and cry influence behaviour, as their functions give us little clue? How do we identify the remaining genes, which are probably pleiotropic and of small effect? Regardless of the outstanding questions, this work shows that behaviour can be genetically dissected by combining classical quantitative analysis, genomic approaches and mutant characterization — a new 'modern synthesis' for understanding the genetic architecture of complex traits.