Figure 1

Scaling and weighting. The left panel shows how the scaling parameter affects the values of the pan-matrix when computing distances between genomes. With no scaling (default) copy numbers beyond 1 are ignored, i.e. for a gene cluster occurring in 1 or more copies the pan-matrix will only have the value 1. If the scale is 1 (maximum) the pan-matrix value is identical to the copy number. Any scale value between 0 and 1 will produce intermediate pan-matrix values. The right panel shows two gene cluster weight strategies available in the geneWeights function. The blue curve is the shell weighting, giving large weight to gene clusters occurring frequently (large detection probability). The red curve is the cloud weighting, emphasizing rarely occurring gene clusters.