Detection of nonrandom clustering of colonies in agar culture

We report a quantitative method for the detection of clustering in agar cultures. A statistical model was constructed on the assumption that colony formation by a cell in agar culture was independent of its spatial relationship to other cells in the culture. Deviation from this model was then tested by a chi-squared test of the actual observations compared with values obtained from the model. This method provides a technique that will identify interactions between cells of like or unlike type. Using a cell line with an extremely high cloning frequency, we show that the assumptions of the method are valid. Two situations are examined in which there was a suggestion of colony clustering on observational grounds. In both cases the test confirmed this impression and allowed us to analyze the possible causes of this behavior. We believe that this method will be of value in the identification of as yet unknown growth factors or cell-cell interactions.