How are gene message alignments calculated?
The program reads in a sequence file and a list of Grail Exon Candidates. It replaces all non-exonic portions of the sequence with “n”‘s and blasts that sequence against the search database. The program examines the resulting alignments briefly to determine which ESTs/cDNAs are worth examining further. It throws away quite a few ESTs at this stage. The program then does a second blast, in which it blasts the entire sequence against the EST/cDNAs of interest. It takes these raw BLAST alignments and applies a dynamic programming algorithm to arrange the alignments into gene structures. It does a merge check, in which it locates the pieces that BLAST failed to join and merges those alignments into one. It then optimizes all the edges, left, right, and internal, doing an exhaustive search of all splice sites in the vicinity of each edge. At this stage, the donor splice site scoring system (developed by Victor Olman) is used to help get the correct splice site. The resulting alignments, wit
