Transposable element subfamily annotation has a reproducibility problem

Table 4 Expected conversion between Alu subfamiles based on a simple model of substitution mutations

	Non-match	Non-match	Non-match	Combined
	AluJ (%)	AluS (%)	AluY (%)	mismatch
AluJ	0.72	2e-5	8e-10	0.73%
AluS	2e-3	7.76	4e-3	7.77%
AluY	3e-5	0.02	6.69	6.71%

For each pair of subfamilies, we computed the probability of switching from one subfamily to another, based on the probability of changing the necessary number of diagnostic sites. Subfamily pairs were clustered by type, capturing the probability of converting from one of the types either within type (diagonal) or between types (off-diagonal). The final column is the sum of all probabilities of converting from the row header to any other subfamily

ISSN: 1759-8753