We studied the occurrence of mammalian interspersed repeats (MIRs) in DNA a
nd RNA of vertebrates, invertebrates, and bacteria using the data from GenB
ank. A special algorithm based on a weight position matrix with optimal ali
gnment using dynamic programming was developed to search for the traces of
MIR dissemination. This allowed us to search for highly divergent MIRs carr
ying deletions and insertions. MIRs were detected in genomes of various fis
hes, including Latimeria. This suggests that the origin of MIRs dates back
more than 400 million years. The method to search for similarity between hi
ghly divergent sequences may be used to find the genome fragments from vari
ous ancient repeat families and from various gene families.