Cram code base #9
Conversation
|
I notice all the new work is licensed under the Apache License 2.0 whilst the rest of the project is licensed under MIT. I assume this is intended? |
|
Historical reasons -- cramtools is under AL2. Does anyone have an opinion on AL2 vs MIT in hts-jdk? The default is MIT I guess. |
|
I'm fine with both licenses. In theory if we changed to ASL2 I should go back through the technology licensing bureaucracy at Broad and get that approved, which is the only downside. |
|
Hold on, I'll check with our bureaucracy :) |
These changes were from before the Picard-public/htsjdk split. This branch is synchronous with the branch of the same name in Picard-public
Still lots of work to do. Just a commit before going home on 5/29
Alterations to implement Queue interface complete pending unit testing and usage (likely will find bugs)
to reflect previous change in DiskBackedQueue to implement Queue, store file records a little more simply
DiskBackedQueue now appropriately implements a queue. Added improvements and additional cases to the unit test. Unit tests pass. Added a TemporaryDuplicateMarkedFlag to SAMRecord. This is set temporarily during Picard-public MarkDuplicatesWithMateCigar to indicate that a record is available for emission from the algorithm. For use with in-memory records that are never spilled to disk.
Messing with SAMRecord was a bad design choice.
| <natures> | ||
| <nature>org.eclipse.jdt.core.javanature</nature> | ||
| </natures> | ||
| <name>cramone</name> |
There was a problem hiding this comment.
removed .classpath and .project as these are Eclipse files
updated javadocs.
|
|
||
| private Sam2CramRecordFactory sam2CramRecordFactory; | ||
| private OutputStream os; | ||
|
|
There was a problem hiding this comment.
fileName and os are never initialized. When a new CRAMFileWriter is instantiated all subsequent calls with return NPEs as a result. There should probably be a constructor that can take a File or an OutputStream similar to BAMFileWriter.
multiple fixes in CRAM writer to make it actually work.
CRAMFileWriter: proper initialization of output stream and record factory; Slice: write empty MD5 if not set; CRAMIterator: nextContainer clears records so that hasNext returns false.
fixed missing pairing in CRAMFileWriter; fixed header written twice; added CRAM reader to SamFileReader make method;
added tests to .classpath added testng output to .gitingore;
tested with CramFileWriterTest.
|
@vadimzalunin have we fixed the secondary read mate info issue? |
|
I think so, although I have not tested it as I'm on ASHG with somewhat On Thu, Oct 16, 2014 at 3:23 AM, Nils Homer [email protected]
|
|
I'm going to close this pull request. The cram code base was merged in to master after a rebase and the SamReader changes needed to take advantage of the cram code base. There is one change that I need to cherry pick from this branch to bring master up to date. After that I'll delete this branch and open a few issues for @vadimzalunin. |
* Adding MAINTAINERS.md file with information about the current maintainers. Please keep this up to date.
# This is the 1st commit message: move Container.setByteOffset() call inside ContainerIO.readContainer() - also ContainerHeaderIO.readContainerHeader() restrict access to setByteOffset() and encapsulate Container.slices # This is the commit message #2: oops # This is the commit message #3: a little unrelated cleanup # This is the commit message #4: oops # This is the commit message #5: better CRAIEntryTest # This is the commit message #6: test improvements and a fix # This is the commit message #7: comment # This is the commit message #8: javadoc # This is the commit message #9: review comments # This is the commit message #10: comments and clarification for CRAMBAIIndexer
Fixes samtools#9. Also add footnote referencing the GZIP RFC.
Just rebased the branch from master. Briefly: