Improve error message when we can't find a ReferenceRegion for a contig

[fnothaft]

Seen on the mailing list. If calling ReferenceRegion(contig: NucleotideContigFragment) returns None, we throw a nondescript error.

[InvisibleTech]

Is this issue still open? I was looking for a starting point to help out and noticed this issue has an assignee but was also labeled "pick me up".

[heuermh]

@InvisibleTech go for it! We're not too strict about the Assignee label.

[InvisibleTech]

@fnothaft @heuermh

Is there something that explains what defines a valid fragment? Also do you have perferences regarding how to fail at this point? Default quietly to an empty value? Raise a RuntimeException? Meaning what is the style of error processing you follow here in this project?

Requirements question. I am looking at fixing this issue with some defensive coding in:

org.bdgenomics.adam.converters.FragmentCollector#apply
Normally in a Java world I would do something like throw a RuntimeException like IllegalArgumentException but I am not sure in FP Scala land if this is legit.

Looking at the code in:

org.bdgenomics.adam.models.ReferenceRegion#apply

I see that if any of the following referenced fields aren't valued we would get None because of the generators used in the for-comprehension:

  contig <- Option(fragment.getContig)
  contigName <- Option(contig.getContigName)
  startPosition <- Option(fragment.getFragmentStartPosition)

So which of these fields is required to properly create a ReferenceRegion? Is this an error or is it more legit for me to silently create some empty Seq[(ReferenceRegion, String)] if ReferenceRegion(fragment) produces a None?