dan2097 / jna-inchi Goto Github PK

Wrapper to access InChI from Java

License: GNU Lesser General Public License v2.1

Java 100.00%

jna-inchi's Introduction

JNA-InChI

Wrapper to access InChI and RInchI from Java. This wraps the latest version of InChI (1.06) and RInChI (1.00) using JNA. A simple native Java interface can then be used to call InChI. Java 8 or higher is required. Detailed information about the capabilities and limitations when converting from and to RInChI can be found in the io.github.dan2097.jnarinchi package documentation.

Examples

Mol file to StdInChI

InchiOutput output = JnaInchi.molToInchi(molText);
if (output.getStatus() == InchiStatus.SUCCESS || output.getStatus() == InchiStatus.WARNING) {
  String inchi = output.getInchi();
}

SMILES to StdInChI

InchiOutput output = SmilesToInchi.toInchi(smiles);
if (output.getStatus() == InchiStatus.SUCCESS || output.getStatus() == InchiStatus.WARNING) {
  String inchi = output.getInchi();
}

InChI to InChIKey

InchiKeyOutput output = JnaInchi.inchiToInchiKey(inchi);
if (output.getStatus() == InchiKeyStatus.OK) {
  String inchiKey = output.getInchiKey();
}

Custom molecule to StdInChI

InchiInput inchiInput = new InchiInput();
inchiInput.addAtom(atom);
inchiInput.addBond(bond);
inchiInput.addStereo(stereo);
InchiOutput output = JnaInchi.toInchi(inchiInput);

Reaction file to RInChI

RinchiOutput rinchiOutput = JnaRinchi.fileTextToRinchi(reactionFileText);
if (rinchiOutput.getStatus() == Status.SUCCESS || rinchiOutput.getStatus() == Status.WARNING) {
  String rinchi = rinchiOutput.getRinchi();
}

RInChI to RInChIKey-Long

RinchiKeyOutput rinchiKeyOutput = JnaRinchi.fileTextToRinchiKey(reactionFileText, RinchiKeyType.LONG);
if (rinchiKeyOutput.getStatus() == Status.SUCCESS) {
  String longRinchiKey = rinchiKeyOutput.getRinchiKey();
}

Decompose RInChI into its constituent InChIs (and associated AuxInfo if any)

RinchiDecompositionOutput rinchiDecompositionOutput = JnaRinchi.decomposeRinchi(rinchi, rauxInfo);
if (rinchiDecompositionOutput.getStatus() == Status.SUCCESS) {
  String[] inchis = rinchiDecompositionOutput.getInchis();
  String[] auxInfos = rinchiDecompositionOutput.getAuxInfos();
}

Supported platforms

InChI and RInChI are C libraries and hence require platform-specific binaries. The following table lists the availability of the binaries for specific platforms. Pull requests for other platforms are welcome.

Platform	InChI	RInChI
Linux x86	Yes	Yes
Linux x86-64	Yes	Yes
Linux ARM	Yes	Yes
Linux ARM64	Yes	No
Mac x86-64	Yes	No
Mac ARM64	Yes	No
Windows x86	Yes	Yes
Windows x86-64	Yes	Yes

Maven artifacts

The simplest way to use the library is with:

<dependency>
  <groupId>io.github.dan2097</groupId>
  <artifactId>jna-inchi-all</artifactId>
  <version>1.2.1</version>
</dependency>

which includes binaries for most common platforms, support for converting SMILES to InChI/InChIKey and MDL RDfile and RXN to RInChI/RInChIKey.

If you don't need SMILES to InChI support, RInChI support, or do not require support for all platforms, the dependency size can be reduced by only including the required modules.

Artifact	Description
jna-inchi-all	Includes all artifacts
jna-inchi-smiles	JNA-InChI API with SMILES to InChI support
jna-inchi-core	JNA-InChI API with binaries for all platforms
jna-inchi-api	JNA-InChI API
jna-inchi-darwin-aarch64	InChI 64-bit ARM Mac support
jna-inchi-darwin-x86-64	InChI 64-bit Intel Mac support
jna-inchi-linux-aarch64	InChI 64-bit ARM Linux support e.g. Apple M1
jna-inchi-linux-arm	InChI 32-bit ARM Linux support e.g. Raspberry Pi
jna-inchi-linux-x86	InChI 32-bit Linux support
jna-inchi-linux-x86-64	InChI 64-bit Linux support
jna-inchi-win32-x86	InChI 32-bit Windows support
jna-inchi-win32-x86-64	InChI 64-bit Windows support
jna-rinchi-core	JNA-RInChI API with binaries for Windows and Linux
jna-rinchi-linux-arm	RInChI 32-bit ARM Linux support e.g. Raspberry Pi
jna-rinchi-linux-x86	RInChI 32-bit Linux support
jna-rinchi-linux-x86-64	RInChI 64-bit Linux support
jna-rinchi-win32-x86	RInChI 32-bit Windows support
jna-rinchi-win32-x86-64	RInChI 64-bit Windows support

For example, jna-inchi-core omits SMILES support. If you only need 64-bit linux support, depending on if SMILES support was desired, you would choose jna-inchi-smiles + jna-inchi-linux-x86-64, or jna-inchi-api + jna-inchi-linux-x86-64

License

This project is licensed under the GNU Lesser General Public License v2.1 or later

jna-inchi's People

Contributors

Stargazers

Watchers

Forkers

mdcatapult johnmay webbres ideaconsult uli-f shunsunsun mjw99

jna-inchi's Issues

Use static ints for InChI code definitions

cf. current implement of InchiCheckStatus

DataSGroups

I have an issue with DataSGroups. In my environment (accessing inchi.dll via old C interface) I always get valid output. With jna-inchi, I get only valid output without DataSGroups. Is this an issue on my side or can you reproduce this behaviour?
test_IXA_DataSGroup.mol.txt
test_IXA.mol.txt

Incorrect SMILES to InChI allenal stereochemistry handling

In an allenal system e.g. AB(A')=C=D(E)E', the original InChI API uses A,B,D,E as reference atoms. However the IXA API uses A,A',E,E'. This does better capture that the system is effectively an extended tetrahedran.

The SMILES to InChI input incorrectly assumes the former convention. This needs to be fixed and documented as to what the expected behaviour is. Adding static convenience methods to construct appropriate InchiStereo objects for double bond/tetrahedral/allenal stereo may help.

The case where a reference atom is an implicit hydrogen needs to be handled.

Cleaner InChI mass interaction/API quality of life

Since InChI stores mass as a delta you need to know what the base line it is a delta from. The InChI API (and by extension JNA InChI) is asymmetric in that you pass in an exact mass and get a "isotopic shift" out - the caller then needs to translate this to something sane. I think perhaps it accepts both an exact 'mass number' or a 'isotopic shift' on input so it's not quite asymmetric but I believe it always gives an 'isotopic shift' on output.

I've just fixed the issue in CDK and found the table in the INCHI-BASE directory:
cdk/cdk#936. It would also be possible to derive the table by round-tripping in enough values.

I wonder if doing the translation in JNA InChI would be reasonable and a more usable API?

JNAInchi and OSGi

I have a cheminformatics app for Cytoscape called chemViz that uses CDK. I was very excited to see that CDK was moving to use JNAInchi, but I have been unable to make it work with OSGi. The problem is that the native code is not on a classpath since it's part of the OSGi bundle. I can load native code using OSGi's implementation of System.loadLibrary(), which is bundle-aware, but System.load is not bundle aware, so JNA can't find the native modules. Do you have anyone who has successfully used JNAInchi in an OSGi setting?

GPL?

Noel pointed me here, looks cool but any reason for the GPL, wouldn't a better option just be to use the InChI license?

Achieve performance parity with JNI-InChI

Type mapping incurrs a performance penalty with JNA. ~15% performance gains are achievable by using Pointer rather than subclasses of PointerType, and by using byte[] rather than String.

Failure to generate InChI for stereo molecule with JNA 5.13.0 and later

We have a project that uses jna-inchi version 1.2. We recently added a new dependency to our project, which caused the indirect dependency on JNA to be updated from version 5.12.1 to 5.13.0. This change caused jni-inchi to break: some molecules for which an InChI was succesfully generated with JNA 5.12.1 now fail with the error message 'Atom ID is invalid'. The problem seems to affect molecules with stereochemistry. The same problem is exhibited with the latest version of JNA, 5.14.0.

We've found that we can work around the problem by replacing the dummy vertices with the actual indices in the DoubleBond case statement of the JnaInchi.addStereos(...) method, but we can't see how this fix works because we don't have access to the C code of jna-inchi's native methods. If you think it might take a while for you to investigate this problem, we'd appeciate it if you would make the code of the native methods available to us so that we can investigate the problem ourselves. Then, if we're able to find a fix, we'll make a pull request.

The following code fragment illustrates the problem:

InchiInput input = new InchiInput();

InchiAtom atom0 = new InchiAtom("F");
input.addAtom(atom0);

InchiAtom atom1 = new InchiAtom("C");
atom1.setImplicitHydrogen(1);
input.addAtom(atom1);

InchiAtom atom2 = new InchiAtom("C");
atom2.setImplicitHydrogen(1);
input.addAtom(atom2);

InchiAtom atom3 = new InchiAtom("Cl");
input.addAtom(atom3);

input.addBond(new InchiBond(atom0, atom1, InchiBondType.SINGLE));
input.addBond(new InchiBond(atom1, atom2, InchiBondType.DOUBLE));
input.addBond(new InchiBond(atom2, atom3, InchiBondType.SINGLE));

input.addStereo(InchiStereo.createDoubleBondStereo(atom0, atom1, atom2, atom3, InchiStereoParity.ODD));

InchiOutput output = JnaInchi.toInchi(input);
System.out.println(output.getStatus() + "  " + output.getMessage());

For version 5.12.1, the output is
SUCCESS

For version 5.13.0, the output is
ERROR Atom ID is invalid

Runtime exception within Tomcat

This is more of a question than anything else.
I recently modified a Java 8 application to use jna-inchi. It works fine locally but when called from a servlet running under Tomcat, there is an exception:
java.lang.NoClassDefFoundError: Could not initialize class io.github.dan2097.jnainchi.inchi.InchiLibrary

First noticed this error on my Windows laptop and was able to get rid of the error by extracting the appropriate dll from the platform-specific JAR (jna-inchi-win32-x86-64-1.0.1.jar) and copying it to tomcat/bin.
On the production Linux server, the same strategy didn't work. (Extracted and copied the .so file from the corresponding JAR files to tomcat/bin.)

So, what I'm looking to find out: has anyone else come across the java.lang.NoClassDefFoundError: Could not initialize class io.github.dan2097.jnainchi.inchi.InchiLibrary when running an application that uses jna-inchi under Tomcat on Linus? How do you avoid the error?