Fix for issue #513: NPE in ValidateSamFile with CRAM

[ekazachkova]

Fix for issue #513:

Current Htsjdk implementation throws NullPointerException during NM tag validation in SamFileValidator because getSequenceDictionary() method in ReferenceFileWalker returns null without .dict file. We propose check result of getSequenceDictionary() for null. If this method returns null don't throw any exceptions and continue processing as if reference isn't indexed. Also, we added new unit test to check code correctness.

Checklist

• Code compiles correctly
• New tests covering changes and new functionality
• All tests passing

[coveralls]

Coverage increased (+0.01%) to 69.821% when pulling 6d3f109 on EpamLifeSciencesTeam:epam-ls_npe_in_ValidateSamFile_with_cram into 45ee380 on samtools:master.

[ZLyanov]

@yfarjoun could you take a look, please?

[droazen]

Can you review?

[vadimzalunin]

minor change to ensure there is no index file used

[vadimzalunin]

just to be sure there is no index/dictionary involved:

final ReferenceSequenceFile reference = new FastaSequenceFile(new File(TEST_DATA_DIR, "nm_tag_validation.fa"), true);

final SamReader samReader = SamReaderFactory.
makeDefault().
validationStringency(ValidationStringency.SILENT).
referenceSource(new ReferenceSource(reference)).
...

[ekazachkova]

@vadimzalunin I can not understand why do we need this check: this test will always fail since cram reader can't work with reference without index. Could you please clarify your comment?

[ekazachkova]

Also, we have discovered that our test data aren't valid: cram file is not paired with reference. So we have regenerated them.

[vadimzalunin]

Could you please check the CRAM file? Reference MD5 checksums don't seem to match.

cat nm_tag_validation.fa | tail -n +2 | tr -d '\n' | md5sum
f8c08a4411f07717451464d546b3706d -

samtools view -H nm_tag_validation.cram | grep '^@sq' | grep -P 'M5:\w+' -o
M5:351f64d4f4f9ddd45b35336ad97aa6de

[ekazachkova]

ooops, @vadimzalunin thanks for remark

[yfarjoun]

verbal thumbs from vadim. 👍