fix read count report

[tedsharpe]

No description provided.

[codecov-io]

Codecov Report

Merging #5883 into master will decrease coverage by 0.005%.
The diff coverage is 63.158%.

@@               Coverage Diff               @@
##              master     #5883       +/-   ##
===============================================
- Coverage     86.843%   86.838%   -0.005%     
- Complexity     32321     32323        +2     
===============================================
  Files           1991      1991               
  Lines         149308    149327       +19     
  Branches       16486     16483        -3     
===============================================
+ Hits          129663    129672        +9     
- Misses         13630     13647       +17     
+ Partials        6015      6008        -7

Impacted Files	Coverage Δ	Complexity Δ
...er/tools/AnalyzeSaturationMutagenesisUnitTest.java	97.849% <100%> (-0.029%)	29 <8> (ø)
...hellbender/tools/AnalyzeSaturationMutagenesis.java	46.968% <30.769%> (+0.077%)	11 <2> (+2)	⬆️
...lotypecaller/readthreading/ReadThreadingGraph.java	88.971% <0%> (+0.245%)	159% <0%> (ø)	⬇️
...nder/utils/runtime/StreamingProcessController.java	67.773% <0%> (+0.474%)	33% <0%> (ø)	⬇️

[SHuang-Broad]

Better structure!
Some suggestions on possible further improvements.

I'd recommend also providing a more descriptive commit message.

[SHuang-Broad]

First, I understand that this absorbs this the functionality of the old PairedReadReport.updateCounts(...).

I find this new architecture a little inconsistent: before this PR, there's the class ReadReport, and PairedReadReport, which now are merged into a single final class ReadReport. Now in this new ReadReport, there's a ctor that takes in two ReadReport's. A reader would be confused by these two: why are there two functions essentially having the same signature (sure, return types are different)?

So I'd recommend looking for a solution to absorb this into the class ReadReport too.

[tedsharpe]

This function figures out which set of moleculeCounts to manipulate and passes that to the updateCounts function of the correct ReadReport (which isn't always a combined report). The reports are combined only in the case where they're not empty and they overlap. It's the parent class that has things chopped up into different buckets of molecule counts, and so it has a function that looks at the reports and chooses the right bucket. It's not really a concern of the ReadReport, which just updates whichever moleculeCount its given. Again, it's a separation of concerns thing. Sorry, but I like it this way better than pushing it all into the ReadReport constructor, which would then have to know about all the buckets and choose the right one.

[SHuang-Broad]

Maybe combineReports should be renamed to updateCountsForPair? Would that be better?

Yep, I think this name reflects the intention better.

[tedsharpe]

I've made a couple of small changes, but pushed back on two of your suggestions. I'll improve the commit description when I rebase and squash. (I'll retain only the comment on this 3rd commit.)
Let me know if you can live with it.

[tedsharpe]

This function figures out which set of moleculeCounts to manipulate and passes that to the updateCounts function of the correct ReadReport (which isn't always a combined report). The reports are combined only in the case where they're not empty and they overlap. It's the parent class that has things chopped up into different buckets of molecule counts, and so it has a function that looks at the reports and chooses the right bucket. It's not really a concern of the ReadReport, which just updates whichever moleculeCount its given. Again, it's a separation of concerns thing. Sorry, but I like it this way better than pushing it all into the ReadReport constructor, which would then have to know about all the buckets and choose the right one.

[tedsharpe]

Maybe combineReports should be renamed to updateCountsForPair? Would that be better?

[SHuang-Broad]

Basically only one meaningful comment about using a different name for combineReports(...).
👍

[tedsharpe]

Thanks very much for the review, Steve.