Replace hard-coded sequence utils with an Alphabet model

[tdanford]

As part of the work on Gene models, we introduced the SequenceUtils object (in adam.utils), which is just a hacky container for some simple "complement" operations on DNA sequences themselves.

This should be replaced by an actual Alphabet model, with support for ambiguity codes, different alphabets, etc.

[ansalaza]

I am interested in picking up this issue. I was thinking of creating an abstract class and extending it with appropriate methods/functions for DNA and RNA sequence and then adding IUPAC codes. Does this approach address this issue? It would also mean changing some code in /adam-core/src/main/scala/org/bdgenomics/adam/models/Gene.scala

[fnothaft]

There had been discussion about using the Alphabet classes from BioJava. However, BioJava is LGPL so we can't package it under Apache 2... @heuermh are you aware of any non-LGPL toolkits that implement Alphabet functionality?

[heuermh]

You're right, BioJava is LGPL and unfortunately is not explicit as to which version is in use. I tried to add "either version XXX of the License, or (at your option) any later version" language at some point but was unsuccessful.

I don't understand though, what the conflict with packaging?

Would LGPL version 3 or later be compatible with packaging under Apache 2?

[Agent007]

Has anyone taken a look at https://github.com/bioscala/bioscala/tree/master/src/main/scala/bio/sequence ?

[fnothaft]

@Agent007 nice! Thanks for the heads up there; I wasn't aware of that, but it's definitely worth a look.

[fnothaft]

Closed by #653.