Summary of PRED-A-TERM program. (A) The putative trans-splicing site is first predicted by fragmenting 700 nucleotides at each occurrence of AG. The longest, pyrimidine rich inter-AG fragment is selected as the sequence upstream of the splicing site. (B) A large scanning matrix subsequently skims through 1000 nucleotides upstream of this position, identifying pyrimidine rich areas with adenosines upstream of them. For each large matrix position, a smaller matrix scans 75 positions in both directions from the larger matrix's hypothetical poly(A) site in order to pinpoint adenosine residues contrasted to pyrimidines. The position of the smaller matrix displaying the highest sum of bit-scores is retained as the putative poly(A) site.
Smith et al. BMC Bioinformatics 2008 9:158 doi:10.1186/1471-2105-9-158