Added argument for specifying NetMHCIIpan/NetMHCIIpanEL version

[ldhtnp]

I added the argument --netmhc_pan_version to pvacseq, pvacbind, pvacfuse, pvacsplice, and pvacvector. The argument supports four versions: 4.0, 4.1, 4.2, and 4.3. I also added an argument that can be used to list the valid versions valid_netmhcpan_versions. Alongside the additions I had to make to get the arguments implemented, I had to modify the code used to remove the sample name and grab the method name in output_parser.py because the method name can now include a version number. Before this modification, the version number would be trimmed and dropped off of the method name resulting in issues downstream.

This is for issue #457

[susannasiebert]

This looks pretty good overall. I made a few comments for suggested changes.

[susannasiebert]

Also, you'll want to add tests for your new command in the test_pvacseq|bind|vector|splice.py files. Check out how the valid_alleles and valid_algorithms are tested in there and implement similar checks. That should probably get your coverage up as well and make the coveralls check pass.

[susannasiebert]

This block of code concerns me. NetMHCIIpan_el is elution algorithm and as such doesn't return an ic50. Is there a specific version combo that returns this header?

[susannasiebert]

Digging into this some more, it looks like standalone IEDB returns the headers as "ic50" instead of score under NetMHCpan_EL version 4.2 and 4.3. This works but I think it would be better if we could specify a set of rules to pick one over the other instead of using a try and except. This would require us to adapt this module to track whether or not IEDB was used standalone.