[Biojava-dev] Changes to Sequence in BioJava3

[Andy Yates]

Hi everyone,

As a caution to people with implementations already built on the Sequence interface I'm proposing a couple of changes to it. This will cause a binary class incompatibility & will have impacts in the methods you need to implement but I'll sort them out at the BioJava core end.

1). Removal of getSequenceAsString(Integer,Integer,Strand)
** The implementation is patchy & buggy often exposing data from backing stores

2). Addition of SequenceView<C> getReverse()
** Will return the sequence in the reverse strand
** Also complemented if applicable

3). Addition of isComplementable() to CompoundSet
** Used to support the above function

This means substrings of Sequences are retrieved as so:

DNASequence d = new DNASequence("ATGCGC");
d.getSubSequence(2, 5).getSequenceAsString(); //Returns TGCG
d.getSubSequence(2, 5).getReverse().getSequenceAsString(); //Returns CGCT

To support -ve strand indexes you can use the Location objects (the returned Location is expressed in +ve coordinates):

Location l = Location.Tools.location(5, 2, Strand.NEGATIVE, d.getLength());
SequenceView<NucleotideCompound> locationSeq = l.getSubSequence(d);
locationSeq.getSequenceAsString(); //Returns CGCT

Hopefully the implications of these changes will be small & will benefit the code

Andy 

p.s. If you are wondering why I am not proposing a deprecation is because I do not want developers writing quite complex code depending on this functionality. If this was not an alpha release then deprecation would be the only way to go