[Biojava-dev] Contributing chromatogram support to BioJava

[Rhett Sutphin]

Hi Matthew,

I thought about this some, and decided that your alignment suggestion 
was the right way to go.  (I was particularly persuaded by the memory 
footprint argument.)  I've posted a revised version of the API in the 
same place as the previous one:  
http://www.detailedbalance.net/~rhett/dev/biojava/chromat-contrib/doc/ 
.  The main difference is that getBaseCalls now returns an alignment of 
at least two SymbolLists -- one for the calls and one for the offsets 
-- and the BaseCall class has been dropped.  I've extensively modified 
SimpleChromatogram to support this  -- in particular I added factory 
methods for creating the Alignments and SymbolLists so that subclasses 
can optimize those as they see fit.

I also introduced a ChromatogramTools class to permit various forms of 
access to the offsets as primitive ints and some shortcuts for commonly 
used accesses.

I look forward to comments from potential users before I finalize the 
interface.

Thanks,
Rhett

--
Rhett Sutphin
Research Assistant (Software)
Coordinated Laboratory for Computational Genomics
   and the Center for Macular Degeneration
University of Iowa - Iowa City, IA 52242 - USA
4111 MEBRF - email: rhett-sutphin at uiowa.edu

On Monday, March 10, 2003, at 09:27  AM, Matthew Pocock wrote:

> Hi Rhett,
>
> I totaly see your point. Perhaps the simples should be made more 
> visible. Darn, I guess we should have had a .symbols.spi package :) 
> This sort of issue hits up hard against one of the limitations of 
> java-like o o programming languages - it's hard to come up with a 
> single right way to represent things, but it's also hard to 
> dynamically map different interfaces and stoorage paradigms between 
> one another without writing loads of adaptors.
>
>
> The other way to do this named properties milarkie is to treat each 
> property as an entry in an alignment. So, you could publish some 
> well-known labels (say on a *Tools class, or an extention of the 
> Alignment interface) called things like DNA, OFFSET etc., and then 
> when you parse the scf file, spit out one of these alignments with all 
> the bits labeled up. Then the access API would look something like:
>
>   Alignment scf = // get an scf object from some source
>   Symbol dna = scf.getSymbolAt(ScfTools.DNA, 7);
>   Symbol offset = scf.getSymbolAt(ScfTools.OFFSET, 10);
>
> and if someone wanted just one of the quality values (which I don't 
> know the names of, but let's call it Q4)...
>
>   SymbolList q4Vals = scf.symbolListForLabel(ScfTools.Q4);
>
> This isn't perfect, but it may be better on the applications memory 
> footprint (when looking at loads of traces) and will play nicer with 
> the ..gui.sequence package. Any way, see what you think. You're in 
> charge of the package :)
>
> Matthew
>
> -- 
> BioJava Consulting LTD - Support and training for BioJava
> http://www.biojava.co.uk
>
>