#154 RECODE bug

[Anonymous]

Hi, I was using RECODE to find EarI restriction endonuclease sites and
suggest silent mutations to ablate these sites. Strangely, the output
came back as if the software performed the analysis on the antisense of
the input gene. As EarI is a type IIs restriction endonuclease and is
non palindromic there are two recognition sequences:

5'-ctcttc-3'

Or

5'-gaagag-3'

In my case the EarI recognition sequence was the result of adjacent
glutamate codons: gaa*gag. What RECODE recognized was a leucine
phenylalanine codon pair (ctc*ttc) that was the antisense of the genuine
coding sequence and the recommend changes ctd*ttc or ctc*ttt instead of
gag*gag or gaa*gaa.

When I tried other type IIs sites in RECODE it did not make this same
mistake.