Figure 6
The multiple data set alignment feature was tested by comparing a) the simultaneous alignment of the three simulated data sets in Table 3, to b) the result from merging data set b ( D b ) and c ( D c ) and then merging this result with the largest data set a ( D a ). The theoretical allele frequencies "Freq." and overlap probabilities "Overlap" are provided for each alignment, where bin i from lab 1 and a bin j from lab 2 overlap if they both align with one or more of the same theoretical alleles. Their overlap probability o ij can be estimated by the fraction of the sampled alignments where overlap occurs. This figure illustrates the importance of merging all data sets simultaneously rather than conducting a series of pair-wise merges. (a) Simultaneous alignment of all three data sets gave the correct alignment with posterior probability 0.55. This posterior probability is lower than the posterior probability for the alignment presented in part b) shown below because the posterior probability of the alignment of D b with D a was low (0.509). (b) The alignment of data set b with data set c was incorrect, but MicroMerge finds a high posterior probability for their alignment (0.997) because the bin frequencies match well. Since this alignment D bc was not accurate, the alignment of D bc with D a was also inaccurate (posterior probability 0.697).