[Bioperl-l] Genbank/EMBL output

[Alex Hudek]

I'm working by the site

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

As Keith mentioned in his post, we have the definitions

complement(location)
Find the complement of the presented sequence in the span specified by
"location" (i.e., read the complement of the presented strand in its
5'-to-3' direction)

join(location,location, ... location)
The indicated elements should be joined (placed end-to-end) to form one
contiguous sequence

The complment operator produces the complement of the sequence in the
5'-to-3' direction of the presented strand. So taking Keith's example I'm
interpreting it this way.

     e1  e2
  5' ACT GGT
     TGA CCA 5'

complement(join(e1,e2))

The join produces ACTGGT, and the complement operator produces TGACCA.

join(complement(e2),complement(e1))

The complement operators produce e1': TGA e2': CCA and the join(e2', e1')
produces CCATGA.

Additionally, the bioperl CVS on Dec 28, 2002 seems to rewrite

complement(join(e1,e2)) 

as 

join(complement(e1),complement(e2))

which I've come to think is correct. I'm by no means an expert on Genbank
format though and originally I interpreted the specefication the same way
you guys did. I only thought deeper about it when I noticed how bioperl
was writing/rewriting my genbank files. 

My original question was actually whether it was possible to move the
complement to the outside of the join as a colleague would rather have it
in that form. So it was an unreleated question really and I don't actually
have a problem with how bio-perl is currently operating.

However, I'm now curious about the correct interpretation of the
genbank complement operator as well. 

Alex


On Sat, 18 Jan 2003, Ewan Birney wrote:

> 
> 
> On Fri, 17 Jan 2003, Alex Hudek wrote:
> 
> >
> >
> > At first glance I thought this as well. Apparently the complement operator
> > is complement only rather than reverse complement as one might think. So
> > you do not have to reverse the ranges when moving the complment tag inside
> > the join.
> 
> Where did you get this statement? Everyone I know think that they are the
> same?
> 
> 


Alexander K. Hudek
The Hospital for Sick Children
Room 9107, The Centre for Applied Genomics
555 University Avenue
Toronto, Ontario, M5G 1X8, Canada
Lab phone: 416-813-7638
Fax: 416-813-8319
E-mail: alex@genet.sickkids.on.ca