[Biojava-dev] Problems with RichLocations

[Mark Schreiber]

This is one area where we don't have many unit tests:

Volunteers???

On 11/3/06, Richard Holland <holland at ebi.ac.uk> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Thanks for pointing this out.
>
> > In RichLocation.Tools.construct(Collection members) the merging is apparently missing. Biojavax always calls merging before construct but the javadoc indicates logically that merging should be done inside the body of the method.
>
> The JavaDocs were wrong. There are instances (GenbankFormat etc.) where
> we need to construct without merging. Therefore, merging is always left
> to the caller. I have amended the JavaDocs to remove this ambiguity.
>
> > In CompoundRichLocation, blocIterator should return an iterator over contiguous blocs sorted in ascending order according to Location javadoc. It does not seem that the list of members is sorted.
>
> I've altered this now and it will return sorted blocks.
>
> > In RichLocation.Tools.flatten(Collection members), members are checked for being instance of SimpleRichLocation. This does not sound correct. First, SimpleRichLocation is not an interface and therefore there is no contract with it and it should not be here; secondly compound members are probably instance of CompoundRichLocation which extends SimpleRichLocation and therefore these compound members are not flattened.
>
> Oops! Fixed now.
>
> > In RichLocation.Tools.modulateCircularLocation(int start, int end, int seqLength), end could be far up and we are probably missing something like
> > while (locationLength>=seqLength) locationLength-=seqLength;
>
> This is deliberate, as circular locations are allowed to wrap round and
> cover the entire sequence as many times as they like. Therefore
> modulateCircularLocation is supposed to return a location of identical
> length, just translated so that start is before end and that start is as
> close to the actual start of the sequence as possible, in order to make
> it easier for the subsequence methods to work out how to extract
> circular sequence. (in the resulting coordinates, start will lie within
> the sequence but end could be far far beyond the actual end of the
> sequence, which will be interpreted as wrapping round and starting again
> from the beginning).
>
> > Hope this helps.
>
> It did! :) All above fixes are in CVS.
>
> cheers,
> Richard
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.2.2 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFFSh494C5LeMEKA/QRAixVAJ0a6rF+/sdG/8P+FfVBrF3i78OuagCePThK
> nVT3GCpLl4AeHBnobZD9zDE=
> =/CAn
> -----END PGP SIGNATURE-----
> _______________________________________________
> biojava-dev mailing list
> biojava-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/biojava-dev
>