How does the program locate 5 and 3 untranslated regions?
The primary indicator of 5′ and 3′ untranslated regions is an additional stop codon in the mRNA. In such cases, the program always attempts to find the 5′ and 3′ untranslated region boundaries. GrailEXP examines the mRNA and finds the highest scoring stop-codon-less run (the score being based on the length of the run and the Grail coding score). It then examines that “run” to find the highest scoring start codon (based on proximity to the beginning of the run, which exon it falls in, Grail start site score, and coding/noncoding scores around the boundary). GrailEXP correctly identifies the start site in a complete mRNA about 95% of the time. The only way to make this number close to 100% is to add protein similarity searching to the system, which is planned for future versions.
