Regd 'diffseq'

[Peter Rice]

Anand Rathi wrote:
> 	While using diffseq to report differences between similar sequences, I
> am getting the following warnings:
> 
> Warning: bad /replace value 'G'
> Warning: bad /replace value 'A'
> Warning: bad /replace value 'C'
> 	Can somebody please elucidate the cause and fix for them.

This is a bad value ... according to the official EMBL/genBank feature 
table documentation:

http://www.ebi.ac.uk/embl/Documentation/FT_definitions/feature_table.html

... but you have to go back to version 2.2 to find this statement:

/replace="change_location"
where change_location is a sequence  descriptor or feature name that
replaces the indicated sequence to yield the difference sequence.
Alternatively any valid location is acceptable and legal.

The correct syntax is /replace="g" It appears you have /replace="G" in 
the feature table.

Can you point me to the entries that are causing the problem?

The check is in ajax/ajfeat.c in function featTagSpecialAllReplace

However ... There are already several fixes in the checking of /replace 
values for known errors in EMBL/GenBank entries. This is one of them. It 
was fixed a week ago, so that /replace="G" (and other upper case 
characters, and the ambiguity codes) will be accepted by the next EMBOSS 
release. Some of these values will break something if we try to use them 
to build the "replace" sequence itself... but that is a separate problem.

Diffseq will still work, but will have to report these features with 
note='*replace: G' rather than replace=G

regards,

Peter Rice