Generate NonStandard InChI

[johnmay]

Via Nina:

Using InChIGenerator with non-standard options generates standard InChI (starting with 1S/ ) with 1.5.14. The same code generated non-standard InChI in previous versions (1.5.13 and before)

List<INCHI_OPTION> options = new ArrayList<INCHI_OPTION>();
options.add(INCHI_OPTION.FixedH);
options.add(INCHI_OPTION.SAbs);
options.add(INCHI_OPTION.SAsXYZ);
options.add(INCHI_OPTION.SPXYZ);
options.add(INCHI_OPTION.FixSp3Bug);
options.add(INCHI_OPTION.AuxNone);
InChIGeneratorFactory factory = InChIGeneratorFactory.getInstance();
InChIGenerator gen1 = factory.getInChIGenerator(mol1, options);

[johnmay]

@vedina have you been able to confirm if this is a dependency problem?

[vedina]

it certainly affects cdk 1.5.14 and not cdk 1.5.13.

had not checked if the underlying inchi library is different - is it?

[johnmay]

No it's not - but there are lots of test cases that generate non-standard InChIs thats didn't fail. Can you check on 2.0-SNAPSHOT.

John

[johnmay]

Oh and I added you exact test case to check - the output is correct:

https://jenkins.bigcat.unimaas.nl/job/cdk/lastCompletedBuild/testReport/org.openscience.cdk.inchi/InChIGeneratorTest/nonStandardInChIWithEnumOptions/

https://github.com/cdk/cdk/blob/master/storage/inchi/src/test/java/org/openscience/cdk/inchi/InChIGeneratorTest.java#L230

[johnmay]

@vedina will close 1 day please let me know if you want to keep open

[vedina]

will check during the weekend, there was a problem when upgrading to 1.5.14

[vedina]

@johnmay here is the test

https://github.com/ideaconsult/examples-cdk/blob/master/maven-single-module/src/test/java/net/idea/examples/cdk/maven_single_module/InchiTest.java

Works with cdk <=1.5.13, fails with 1.5.14 . There is maven profile for CDK versions .

[vedina]

@johnmay it's weird, the tests (including the one in CDK) fail only on Windows machines (both JDK 7 and 8). In fact other tests in the cdk/storage/inchi folder are also failing on Windows.

Any hints what might have changed since 1.5.13 will be very useful!

-------------------------------------------------------
 T E S T S
-------------------------------------------------------
Running org.openscience.cdk.graph.invariant.InChINumbersToolsTest
Tests run: 14, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 0.963 sec <<< FAILURE! - in org.openscience.cdk.graph.invariant.InChINumbersToolsTest
fixedH(org.openscience.cdk.graph.invariant.InChINumbersToolsTest)  Time elapsed: 0.105 sec  <<< FAILURE!
java.lang.AssertionError: 
Expected: is "AuxInfo=1/1/N:6,7,5,8,2,4,9,3,1/E:(1,2)(3,4)(6,7)(8,9)/F:7,6,8,5,2,9,4,1,3/rA:9NCNCCCCCC/rB:s1;d2;s3;d4;s5;d6;s7;s1s4d8;/rC:;;;;;;;;;"
     but: was "AuxInfo=1/1/N:6,7,5,8,2,4,9,3,1/E:(1,2)(3,4)(6,7)(8,9)/rA:9NCNCCCCCC/rB:s1;d2;s3;d4;s5;d6;s7;s1s4d8;/rC:;;;;;;;;;"
	at org.hamcrest.MatcherAssert.assertThat(MatcherAssert.java:20)
	at org.hamcrest.MatcherAssert.assertThat(MatcherAssert.java:8)
	at org.openscience.cdk.graph.invariant.InChINumbersToolsTest.fixedH(InChINumbersToolsTest.java)

Running org.openscience.cdk.inchi.InChIGeneratorFactoryTest
Tests run: 12, Failures: 2, Errors: 0, Skipped: 0, Time elapsed: 0.256 sec <<< FAILURE! - in org.openscience.cdk.inchi.InChIGeneratorFactoryTest
testGetInChIGenerator_IAtomContainer_List(org.openscience.cdk.inchi.InChIGeneratorFactoryTest)  Time elapsed: 0 sec  <<< FAILURE!
org.junit.ComparisonFailure: expected:<InChI=1[]/ClH/h1H> but was:<InChI=1[S]/ClH/h1H>
	at org.junit.Assert.assertEquals(Assert.java:115)
	at org.junit.Assert.assertEquals(Assert.java:144)
	at org.openscience.cdk.inchi.InChIGeneratorFactoryTest.testGetInChIGenerator_IAtomContainer_List(InChIGeneratorFactoryTest.java)

testGetInChIGenerator_IAtomContainer_String(org.openscience.cdk.inchi.InChIGeneratorFactoryTest)  Time elapsed: 0 sec  <<< FAILURE!
org.junit.ComparisonFailure: expected:<InChI=1[]/ClH/h1H> but was:<InChI=1[S]/ClH/h1H>
	at org.junit.Assert.assertEquals(Assert.java:115)
	at org.junit.Assert.assertEquals(Assert.java:144)
	at org.openscience.cdk.inchi.InChIGeneratorFactoryTest.testGetInChIGenerator_IAtomContainer_String(InChIGeneratorFactoryTest.java)

[johnmay]

Ahhhh that makes more sense. Putting this together with something I saw on the InChI mailing list. InChI in it's infinite wisdom has different syntax depending on the OS. Windows uses '/' and everything else uses '-'.

Windows:
inchi-1 /STDIO
Everything Else:
inchi-1 -STDIO

When we modified the handling to allow the 15T/KET etc I added output as '-KET -15T' etc. I really thought the '/' was not needed at the API level... but apparently this is the case.

I'll get this fixed ASAP, hopefully by tomorrow.

Here (https://github.com/cdk/cdk/blob/master/storage/inchi/src/main/java/org/openscience/cdk/inchi/JniInChIInputAdapter.java#L79) the '-' needs to be replaced with a constant based on OS (like line separation).

[vedina]

I'see ....