chromosomes should return String array or make GenomicRanges accept Symbol array

[Marlin-Na]

I think GenomicRanges perhaps should accept Array{Symbol,1} as chromosomes as well since chromosomes function is returning Array{Symbol,1}. Alternatively, chromosomes should return a String Array (which I prefer, since strings might be easier to work with).

Currently, I have to use map to replicate a GenomicRanges:

GenomicRanges(map(String, chromosomes(gr)), starts(gr), ends(gr), chr_info(gr))

Ideally, this should work:

GenomicRanges(chromosomes(gr), starts(gr), ends(gr), chr_info(gr))

Anyway, this should be a minor issue on UX.

[phaverty]

I thought about this a bunch when I did my last big batch of work on this project. Symbols seem like the right thing to use internally because I want them to be immutable labels. I've even played with making them part of the type definition by making the chromosomes slot a NamedTuple. I probably want both Symbol and String to work in the user API. I should experiment with this more now that Strings are more efficient (in 1.5+).

[Marlin-Na]

Thanks @phaverty . I don't have much to comment about the Symbol vs String for internal use. But your idea of making them as part of type definition is interesting. Even chromosome lengths could be part of the type definition, so that it could be used to parameterize scalar integer as "genopos" (mentioned in #25 ). But such complex type probably would slow down JIT compiling. Also it won't work with megagenome, which might contain thousands of contigs.

[phaverty]

I did a bit of work to get GenomicVectors updated for some recent updates in Julia and dependency packages. I'll be able to have a look at this now.

[phaverty]

Fixed by 47078ec