Trans-splicing is an unusual process
in which two separate RNA strands are spliced together to yield a
mature mRNA. We present a novel computational approach which has an
overall accuracy of 82% and can predict 92% of known trans-splicing
sites. We have applied our method to Chromosomes 1 and 3 of
Leishmania major, with high confidence predictions for 85% and 88% of
annotated genes respectively. We suggest some extensions of our
method to other systems.