Adding utilities for read trimming.

[fnothaft]

This pull request adds a few utilities for read trimming, as a first start towards read error correction. This pull request contains:

• An update to the ADAMRecord schema to add fields for bases trimmed from the read start/end. After alignment, these fields are redundant (with hard clipping in the CIGAR), but the fields are important for trimming performed before alignment. These fields are added in the SAMRecordConverter.
• Two new RDD functions. One performs a fixed length trim to all reads. A second function trims the prefix/suffix of all reads in a read group based on the geometric mean of the error probability at that position.
• An update to the Transform CLI module, to add these two functions as possible transforms.
• A correction to the CycleCovariate that correctly calculates the sequencing cycle for hard clipped reads.

[AmplabJenkins]

All automated tests passed.
Refer to this link for build results: https://amplab.cs.berkeley.edu/jenkins/job/ADAM-prb/327/

[massie]

Shouldn't this be an optional field?

[fnothaft]

Hmmm, something went screwy with the rebase here. Let me fix this...

[AmplabJenkins]

All automated tests passed.
Refer to this link for build results: https://amplab.cs.berkeley.edu/jenkins/job/ADAM-prb/328/

[massie]

Do you mean getBasesTrimmedFromEnd here?

[fnothaft]

Yes, yes, I do... I'll fix this.

[tdanford]

Are we still maintaining updates to the CHANGES file(s) as part of a pull request?

[fnothaft]

@tdanford I believe we've dropped the CHANGES.txt, and are just keeping the CHANGES.md. However, CHANGES.md should probably be updated via the script whenever we do a release.

[AmplabJenkins]

All automated tests passed.
Refer to this link for build results: https://amplab.cs.berkeley.edu/jenkins/job/ADAM-prb/332/

[tdanford]

Frank, I think you've got some rebasing to do -- it looks like there are some commits in this branch that are duplicated from master, right?

[fnothaft]

@tdanford correct; not sure how that happened...

[fnothaft]

@tdanford this is rebased now, thanks for the heads up!

[tdanford]

Again! I just merged Matt's edit to CONTRIBUTING.md, you're still out of date! :-)

[fnothaft]

@tdanford u r rebase troll :'(

Re-rebased...

[tdanford]

I like my commit histories like I like my spaces: linear.

[tdanford]

Okay, a more substantive comment: Frank, do you have somewhere (comments? a change file?) where you can write a little bit about the motivation behind these trimming additions? Are you trying to recreate specific functionality from Picard (I'm thinking the MergeBamAlignments command, but maybe something else)? From somewhere else? Does this satisfy a requirement or need for avocado or variant calling in particular? I understand there are a lot of reasons why one might "trim" reads, I'm just hoping for a little context here.

[fnothaft]

@tdanford Good point; I don't have this documented yet. When I started on this, I was planning to implement read error correction first (à la Quake), but I then decided that read trimming was a better first start. The need is indeed for variant calling/assembly.

[AmplabJenkins]

All automated tests passed.
Refer to this link for build results: https://amplab.cs.berkeley.edu/jenkins/job/ADAM-prb/336/

[AmplabJenkins]

All automated tests passed.
Refer to this link for build results: https://amplab.cs.berkeley.edu/jenkins/job/ADAM-prb/337/

[massie]

Thanks, Frank!