I wrote some code to read a chunk of DNA sequence from a file in Jim Kent's
blat ".nib" file format.  This is a simple format using four bits/base.

  I didn't attach the code, to avoid spamming the whole list; but it, and a
(very crude!) JUnit test, are at

http://keyfitz.org/jburdick/read_nib_file_java/NibFile.java
http://keyfitz.org/jburdick/read_nib_file_java/NibFileTest.java

  You could use 2 bits/base, but then you can't have ambiguous bases.  4 bits/base
seems like a reasonable compromise; plus sites that have "blat" installed will need to
have the .nib files on a server somewhere anyway, and this way repeat-masking can be
included, which may be convenient.

  It would need some tweaking to be included in BioJava.  The package name should
be changed.  There are methods to return a String of ACGT, and a SymbolList
object.  (Returning the SymbolList will just 
  Also, the coordinate information should probably come from a Range object.
  And the JUnit test could do more than just print stuff.

  If similar code is already somewhere in BioJava, please ignore this; but I couldn't
find it with thirty seconds of Googling, so I figured it hadn't been written...

Josh Burdick
programmer, Vivian Cheung's lab, Children's Hospital of Philadelphia
jburdick@keyfitz.org