I would like to add a remark to my last comment on the "Rox'n'Rab" repeat mentioned in my previous comment to Flybase. This repeat is present at many X-linked sites, starting with the 3' end of the roX1 gene, the 1A region and the region 9C-9E around six Rab GTPases paralogs (RabX2, Rab9D, Rab9Db, Rab9E, Rab9Fa and Rab9Fb). I have noticed that the above repeat contains a very well conserved 25 nt sequence -CATTTTTTGTAAGGGGTAACATCAT- which has the following features:
-at least 153 perfect matches are present on the X of D. melanogaster (in addition to over a hundred 24 nt matches). Several perfect matches are also present on U extra chromosomes.
-at least 1 perfect match (1 to 5 copies) is present on each chromosome arm of D. melanogaster (except on 3L that has apparently only a 24 nt match).
Interestingly, there are over a hundred perfect matches in the current releases of the genomes of all other species of the melanogaster subgroup (D. simulans, D. sechelllia, D. erecta and D. yakuba), whereas it is absent in more distant species, thus appearing very abruptly in D. ananassae. The possibility that this sequence and its surrounding genomic DNA is involved in dosage compensation or X-inactivation during spermatogenesis seems worth considering.